Psychological Statistics

Download as pdf or txt
Download as pdf or txt
You are on page 1of 170

ALAGAPPA UNIVERSITY

(Accredited with ‘A+’ Grade by NAAC (with CGPA: 3.64) in the Third Cycle and Graded
As category - I University by MHRD-UGC)
(A State University Established by the Government of Tamilnadu)

KARAIKUDI – 630 003

DIRECTORATE OF DISTANCE EDUCATION

B.Sc. (PSYCHOLOGY)

Second Year – Third Semester

11933 –PSYCHOLOGICAL STATISTICS

Copy Right Reserved For Private Use only


Author:
Dr. M. Ramya Maheswari
Professor & Head
Department of Psychology,
Ethiraj College for women,
Chennai-8.

“The Copyright shall be vested with Alagappa University”

All rights reserved. No part of this publication which is material protected by this copyright notice may be
reproduced or transmitted or utilized or stored in any form or by any means now known or hereinafter invented,
electronic, digital or mechanical, including photocopying, scanning, recording or by any information storage or
retrieval system, without prior written permission from the Alagappa University, Karaikudi, Tamil Nadu.
SYLLABI BOOK MAPPING TABLE
PSYCHOLOGICAL STATISTICS
Syllabi Mapping in Book

BLOCK I: STATISTICAL MEASURES 1-13

UNIT 1: Statistics
Definitions of statistics Importance of Statics in
Psychological and Research Collections of Data Primary
and Secondary-classification of Data Formation of
Frequency distribution
UNIT 2: Diagrammatic and Graphical 14-24
Diagrammatic and Graphical representation of Statistical
Data Simple Multiple sub divided, percentage Bar
Diagram and Pie diagram – Histogram Frequency
polygon Frequency Curve, Ogives

UNIT 3: Measures of Central Tendency 25-34


Measures of Central Tendency Mean Median Mode Geometric Mean
and Harmonic Mean

BLOCK II: CONCEPTS OF DISPERSION, 35-46


CORRELATION
AND PROBABILITY
UNIT 4 : MEASURE OF DISPERSION
Measure of Dispersion range Quartile Deviation, Mean Deviation (about
mean) Standard Deviation and Co efficient of variance –concept of
Skeweness –Karl Pearson and Bowley’s coefficient of Skeweness –
Simple Problems- Kurtosis 9concpet only

UNIT 5: CORRELATION 47-60


Correlation-Scatter diagram – Karl Pearson co efficient correlation
Spearmen’s Rank correlation – regression equation –Properties of
regressive coefficient simple problems

UNIT 6: CONCEPT OF PROBABILITY 61-69


Concept of probability basic Definition- mathematical and statistical
approach
– Addition and multiplication Theorems(without proof) simple
Problems

BLOCK III : LEARNING, INTELLIGENCE AND 70-82


PERSONALITY
UNIT 7 : NORMAL DISTRIBUTION

Concept of Probability, Meaning, Importance and Characteristics of


Normal Probability Curve, Deviations from Normal Probability Curve
83-93
UNIT 8: BINOMIAL DISTRIBUTIONS
Skewness and Kurtosis, Applications of Normal Probability Curve,
Binomial Distributions.

UNIT 9: ANALYSIS OF VARIANCE 94-110

Purpose and Assumptions of Analysis of Variance, One way


and Two way Analysis of Variance

UNIT 10: NON-PARAMETRIC TESTS 111-121


Nature and Assumptions, Distribution Free Statistics, Chi-square,
Contingency Co- efficient, Median and Sign test, Friedman test

BLOCK IV: EDUCATION FOR ALL 122-126


UNIT 11: Preparation of Data for Computer Analysis

Familiarization with Software Packages for Statistical Analysis and their


Applications.

UNIT 12: INFERENTIAL STATISTICS 127-135


Introduction to Inferential Statistics and Hypothesis Testing about the
Difference between Two Independent Means: The meaning of Statistical
Inference and Hypothesis Testing; Hypothesis Testing about the
difference between Two Independent means; Null and the Alternative
Hypotheses;

UNIT 13: THE RANDOM SAMPLING DISTRIBUTION 136-151

The Random Sampling Distribution of the Difference between Two


Sample Means; Properties of the Sampling Distribution of the Difference
between Means; Choice of HA: One-Tailed and Two- Tailed Tests;

UNIT 14: INTERPRETING THE RESULTS OF HYPOTHESIS 152-158


TESTING
A Statistically Significant Difference versus a Practically Important
Difference; Errors in Hypothesis Testing; Power of a Test; Levels of
Significance versus p-Values.

Model Question Paper 159-160


CONTENTS
UNIT 1: STATISTICS 1-13
1.1. Definitions of statistics:
1.2. Importance of statistics in psychology and research.
1.2.1: Applications of Statistics
1.3. Branches of Statistical methods:
1.4. Variables and Measurement
1.4.1: Types of Variables
1.4.2: Scales of measurement.
1.4.2.1: Nominal
1.4.2.2: Ordinal
1.4.2.3: Interval
1.4.2.4: Ratio
1.5. Collection of data:
1.5.1 Primary
1.5.1.1 Observation method
1.5.1.2 Interview method
1.5.1.3 Questionnaires
1.5.1.4 Schedules,
1.5.2 Secondary
1.6. Classification of Data: Formation of Frequency distribution.
1.7. Let us Sum up
1.8. Unit – End exercises
1.9. Answer to check your progress
1.10. Suggested readings

UNIT 2: DIAGRAMMATIC AND GRAPHICAL


REPRESENTATION OF STATISTICAL DATA 14-24
2.1 Introduction
2.1.1 General principles of graphical representation of data
2.1.2 Advantages of graphical representation of data
2.1.3 Disadvantages of graphical representation of data
2.2 Graphs For Interval and Ratio Data
1.2.1: Histogram
1.2.2: Frequency Polygons
1.2.3: Cumulative Percentage Curve or Ogive.
2.3 Graphs for Nominal and Ordinal Data
2.3.1: Bar Diagram
2.3.1:1: Simple bar diagram
2.3.1:2: Multiple bar diagram
2.3.1:3: Sub-divided bar diagram
2.3.1: 4: Percentage bar diagram
2.3.2: Pie Charts
2.4 Let us Sum up
2.5 Unit – End exercises
2.6 Answer to check your progress
2.7 Suggested readings
UNIT 3: MEASURES OF CENTRAL TENDENCY 25-34
3.1 Introduction
3.2 Arithmetic Mean
3.2.1: Computation of mean from ungrouped data
3.2.2: Computation of mean from grouped data
3.2.3 : Properties of the mean
3.3Median
3.3.1: Computation of median from ungrouped data
3.3.2: Computation of median from grouped data
3.3.3 : Properties of the median
3.4 Mode
3.4.1: Computation of mode from ungrouped data
3.4.2: Computation of mode from grouped data
3.4.3Properties of the mode

3.5 Geometric Mean


3.6 Harmonic Mean.
3.7 Let us Sum up
3.8 Unit – End exercises
3.9 Answer to check your progress
3.10 Suggested readings

BLOCK II: CONCEPTS OF DISPERSION, CORRELATION 35-46


AND PROBABILITY
UNIT 4: MEASURES OF DISPERSION

1.1 Introduction
1.2 Range
1.2.1 Properties of the range
1.3 Quartile deviation
4.3.1 Properties of the quartile deviation
1.4 Standard deviation and Variance for populations
1.4.1 Computation of standard deviation from
ungrouped and grouped data
1.4.2 Properties of the Standard deviation
1.4.3 Computation of variance
4.5 Standard deviation and Variance for samples
4.5.4: Sampling variability and degrees of freedom
4.6 Concept of skewness and kurtosis
4.6.1 Karl Pearson co-efficient of skewness
4.6.2 Bowley’s Co-efficient of skewness
4.6.3 Kurtosis
4.7 Let us Sum up
4.8 Unit – End exercises
4.9 Answer to check your progress
4.10 Suggested readings
UNIT V: CORRELATION 47-60

5.1: Introduction to Correlation


5.1.1: Characteristics of relationship
5.2 Scatter Diagram
5.3 Methods of computing co-efficient of correlation
5.3.1Karl Pearson co-efficient of correlation
1.3.2 Spearman’s rank correlation
5.4 Regression
5.4.1: Regression Line
5.4.2: Regression Equation
1.4.3 Properties of regression co-efficient.
1.4 Let us Sum up
1.5 Unit – End exercises
1.6 Answer to check your progress
5.8 Suggested Readings

UNIT 6: CONCEPT OF PROBABILITY 61-69

6.1 Introduction
6.2 Approaches to Probability
6.3 Random Sampling
6.4 Basic Rules of Probability
6.4.1Addition theorem
6.4.2Multiplication theorem for dependent and independent events
6.5 Let us Sum up
6.6 Unit – End exercises
6.7 Answer to check your progress
6.8 Suggested Readings

BLOCK III: DISTRIBUTIONS AND VARIANCE 70-82


UNIT 7: NORMAL DISTRIBUTION

7.1 Meaning and Importance


7.2 Characteristics of Normal Probability Curve
1.2.1 Probability and Normal Distribution
7.3 Deviations of Normal Probability Curve
7.3.1 Skewness
1.3.2 Kurtosis
7.4 Applications of the Normal Probability Curve.
7.5 Let us Sum up
7.6 Unit – End exercises
7.7 Answer to check your progress
7.8 Suggested Readings

UNIT 8: BINOMIAL DISTRIBUTIONS 83-93

8.1 Introduction
8.2 The Binomial Distribution
8.3 The Binomial Test
8.3.1Hypotheses for the binomial test
8.3.2The data for the binomial test
8.3.3The test statistic for the binomial test
8.4 Let us sum up
8.5 Unit End Exercises
8.6 Answers to check your progress
8.7 Suggested Readings

UNIT 9: ANALYSIS OF VARIANCE 94-110

9.1 Purpose and Logic of Analysis of Variance


9.2 Assumptions of Analysis of Variance.
9.3 One Way Analysis of Variance.
9.3.1 Illustration
9.4 Two Way Analysis of Variance
9.4.1 Illustration
9.5 Let us Sum up
9.6 Unit – End exercises
9.7 Answer to check your progress
9.8 Suggested Readings

10 UNIT 10: NON- PARAMETRIC TESTS 111-121

10.1.Nature and Assumptions


10.1.1 Advantages of Parametric test
10.1.1 Disadvantages of Parametric test
10.2 Chi-square
10.2.1 Assumptions of the Chi-square test
10.2.2 Null hypothesis for chisquare
10.2.3 Steps to compute Chi-square.
10.2.4 Uses of chi-square test
10.3 Contingency Co-efficient
10.4 Median test
10.4.1 Steps in the computation of the median test
10.5 Sign test
10.6 Friedman test.
10.7 Let’s sum up
10.8 Unit End Exercises
10.9 Answers to check your progress
10.10 Suggested Readings

BLOCK IV: ANALYSIS AND INTERPRETATION OF DATA


UNIT 11: PREPARATION OF DATA FOR COMPUTER 122-126

11.1Introduction:
11.2 Software packages that are used for statistical analysis
11.2.1. Programs for Quantitative Research.
11.2.2 Programs for Qualitative Research
11.3Let’s sum up
11.4 Unit End Exercises
11.5 Answers to check your progress
11.6 Suggested Readings

UNIT 12: INFERENTIAL STATISTICS 127-135

12.1 Introduction to Inferential Statistics


12.2 Types of Hypothesis
12.3 The Meaning of Statistical Inference
12.3.1 Relationship between probability and Inferential Statistics
12.4Methods of statistical Inference
12.4.1 Statistical Estimation
12.4.2 Statistical hypothesis testing
12.5 Let us sum up
12.6 Unit end Exercises
12.7 Answers to check your progress
12.8 Suggested Readings

UNIT 13: THE RANDOM SAMPLING DISTRIBUTION 136-151

13.1Population and Samples


13.2The Distribution of sample means
13.3Properties of the distribution of sample means
13.3.1 The shape of the distribution of sample means
13.3.2 The mean of the distribution of sample means: The
expected value of M
13.3.3 The standard error of M
13.4 Sampling distribution of differences between means
13.5 Introduction to t statistics
13.5.1: Illustration of hypothesis testing with t statistics
for Independent measures research design
13.5.2 : Illustration of hypothesis testing with t statistics
Repeated measures research design
13.6 Choice of alterative hypothesis: One tailed and two
tailed tests of significance
13.7 Let us sum up
13.8Unit end exercises
13.9 Answers to check your progress
13.10 Suggested Readings
UNIT 14: INTERPRETING THE RESULTS OF HYPOTHESIS 152-158
TESTING

14.1A statistically significant difference versus practically important difference.


14.2Errors in Hypothesis testing.
14.3Power of the test.
14.4Statistical decision making: Levels of Significance versus p –values
14.5Let us sum up
14.6Unit End exercises
14.7Answers to check your progress
14.8Suggested Readings

Model Question Paper 159-160


BLOCK- I Statistics

NOTES
UNIT-I – STATISTICS
1.1. Definitions of statistics:
1.2. Importance of statistics in psychology and research.
1.2.1: Applications of Statistics
1.3. Branches of Statistical methods:
1.4. Variables and Measurement
1.4.1: Types of Variables
1.4.2: Scales of measurement.
1.4.2.1: Nominal
1.4.2.2: Ordinal
1.4.2.3: Interval
1.4.2.4: Ratio
1.5. Collection of data:
1.5.1 Primary
1.5.1.1 Observation method
1.5.1.2 Interview method
1.5.1.3 Questionnaires
1.5.1.4 Schedules,
1.5.2 Secondary
1.6. Classification of Data: Formation of Frequency distribution.
1.7. Let us Sum up
1.8. Unit – End exercises
1.9. Answer to check your progress
1.10. Suggested readings

1.1 Definitions of statistics:

Statistics according to Horace Secrist is ‗Aggregates of facts, affected to


marked extent by multiplicity of causes, estimated or enumerated
according to reasonable standards of accuracy, collected in a systematic
manner, for a pre-determined purpose and placed in relation to each
other‖. From the above data we understand that individual figures are not
statistics.
Statistics refer to numerical facts. It also signifies the method or methods
of dealing with numerical facts. Further statistics refers to the summarized
figures of numerical facts such as percentages, averages, means, medians,
modes etc. This indicates that qualitative data do not qualify as statistics.

1 Self-instructional Material
Statistics
The term Statistics refers to a set of mathematical procedures for
organising, summarizing and interpreting information.
NOTES
1.2 Importance of statistics in psychology and research:

Statistics renders valuable services in collection of evidences or facts.


Statistics are used to organise and summarize the information so that the
researcher can see what happened in the research study and can
communicate the results to others. Statistics help the researcher to answer
the questions that initiated the research by determining exactly what
general conclusions are justified based on the specific results that were
obtained.
Statistical procedures help to ensure that the information and
observation are presented and presented and interpreted in an accurate and
informative way. In addition statistics provide researchers with the set of
standardised techniques that are recognised and understood throughout the
scientific community. Thus, the statistical methods used by one researcher
are familiar to other researchers, who can accurately interpret the statistical
analyses with the full understanding of how analysis was done and what
the results signify.

1.2.1: Applications of statistics

(1) Business
Statistics plays an important role in business. Statistics helps businessmen
to plan production according to the taste of the customers, and the quality
of the products can also be checked more efficiently by using statistical
methods. Thus, it can be seen that all business activities are based on
statistical information. Businessmen can make correct decisions about the
location of business, marketing of the products, financial resources, etc.

(2) Economics
Economics largely depends upon statistics. National income accounts are
multipurpose indicators for economists and administrators, and statistical
methods are used to prepare these accounts. In economics research,
statistical methods are used to collect and analyze the data and test
hypotheses. The relationship between supply and demand is studied by
statistical methods; imports and exports, inflation rates, and per capita
income are problems which require a good knowledge of statistics.
(3) Mathematics
Statistics plays a central role in almost all natural and social sciences. The
methods used in natural sciences are the most reliable but conclusions
drawn from them are only probable because they are based on incomplete
2
Self-instructional Material
evidence. Statistics helps in describing these measurements more precisely. Statistics
Statistics is a branch of applied mathematics. A large number of statistical
methods like probability averages, dispersions, estimation, etc., is used in
mathematics, and different techniques of pure mathematics like integration, NOTES
differentiation and algebra are used in statistics.
(4) Banking
Statistics plays an important role in banking. Banks make use of statistics
for a number of purposes. They work on the principle that everyone who
deposits their money with the banks does not withdraw it at the same time.
The bank earns profits out of these deposits by lending it to others on
interest. Bankers use statistical approaches based on probability to estimate
the number of deposits and their claims for a certain day.
(5) State Management (Administration)
Statistics is essential to a country. Different governmental policies are
based on statistics. Statistical data are now widely used in making all
administrative decisions. Suppose if the government wants to revise the
pay scales of employees in view of an increase in the cost of living, and
statistical methods will be used to determine the rise in the cost of living.
The preparation of federal and provincial government budgets mainly
depends upon statistics because it helps in estimating the expected
expenditures and revenue from different sources. So statistics are the eyes
of the administration of the state.
(6) Natural and Social Sciences
Statistics plays a vital role in almost all the natural and social sciences.
Statistical methods are commonly used for analyzing experiments results,
and testing their significance in biology, physics, chemistry, mathematics,
meteorology, research, chambers of commerce, sociology, business, public
administration, communications and information technology, etc.

(7) Astronomy
Astronomy is one of the oldest branches of statistical study; it deals with
the measurement of distance, and sizes, masses and densities of heavenly
bodies by means of observations. During these measurements errors are
unavoidable, so the most probable measurements are found by using
statistical methods. For example, the distance of the moon from the earth is
measured. Since history, astronomers have been using statistical methods
like method of least squares to find the movements of stars.
1.3 Branches of Statistical methods:

Although researchers have developed a variety of different


statistical procedures to organize and interpret data, these different
procedures can be classified into two general categories.
a) Descriptive Statistics: These are statistical procedures used to
summarise, organize and simplify data.

3 Self-instructional Material
Statistics
b) Inferential Statistics: Consists of techniques that allow us to study
samples and then make generalizations about the populations from
NOTES which they are selected.

Check your Progress -1

1. Qualitative data are not ________.


2. What are the two categories of statistical techniques?

1.4 Variables and Measurement:


The scores that make up the data from a research study are the result of
observing and measuring variables. The variables in a study can be
characterized by the type of values that can be assigned to them
1.4.1: Types of variables: Continuous and discrete variables

a) Continuous Variable: For a Continuous Variable, there are an


infinite number of possible values that fall between any two observed
values . A continuous variable is divisible into an infinite number of
fractional parts Examples are height , weight etc all of which when
measured can be broken down to an infinite number of possible points
between any two points on a scale . Two other factors apply for a
continuous variable.
1) When measuring a continuous variable, it should be very rare to obtain
identical measurements for two different individuals, as it has an infinite
number of possible values. If the data show a substantial number of tied
scores, then you would suspect the measurement procedure is very crude or
that the variable is not really continuous.
2) When measuring a continuous variable, each measurement category is
actually an interval that must be defined by boundaries. For example, two
people who both claim to weigh 150 pounds are probably not exactly the
same weight. One person may weigh 149.6 and the other 150.3. Thus, a
score of 150 is not a specific point on the scale but instead is an interval.
To differentiate a score of 150 from a score of 149, we must set up
boundaries on a scale of measurement. These boundaries are called real
limits and are positioned exactly half way between adjacent scores. Thus ,
a score of X = 150 pounds is actually an interval bounded by a lower real
limit of 149.5 at the bottom and a real upper limit of 150.5 at the top. Any
individual whose weight falls between these real limits will be assigned a
score of X =150.
b) Discrete Variable: Discrete variables consist of separate indivisible
categories. No values can exist between two neighbouring categories.
Discrete variables are commonly restricted to whole countable numbers-
for example number of children in the family , number of students

4
Self-instructional Material
attending class. If you observe class attendance from day to day, you may Statistics
count 18 students one day and 19 students the next day. However it is
impossible ever to observe a value between 18 and 19. A discrete variable
may also consist of observations that differ qualitatively. For example NOTES
people classified by gender (male/female), by occupation (nurse,
teacher,etc)
1.4.2 Scales of Measurement:
Data collection requires that we make measurements of our
observation. Measurement involves assigning individuals or events to
categories. The categories can simply be names such as male/female,
employed/unemployed, or they can be numerical values such as 68 inches
or 175 pounds. The categories used to measure a variable make up scales
of measurement, and the relationships between categories determine
different types of scales. The distinction among the scales are important
because they identify the limitations of certain types of measurements and
because certain statistical procedures are appropriate for scores that have
been measured on some scales but not others.. Stevens (1956) has
recognized four type of scales - nominal, ordinal, interval and ratio.
1.4.2.1Nominal Scale:
In nominal measurement numbers are used to name, identify a
classify persons, objects, groups etc. For example a sample of persons
being studied may be classified as (a) Hindu (b) Muslim (c) Sikh or the
same sample may be classified on the basis of sex, rural - urban variable
etc.
In nominal measurement, members of any two groups are never
equivalent but all members of the same group are equivalent. In case of
nominal measurement admissable statistical operations are counting a
frequency, percentage proportion, mode and co-efficient of contingency.
The drawback of nominal measurement is that it is most elementary and
simple.
1.4.2.2 Ordinal Scale:
In ordinal measurement numbers denote the rank order of the
objects or the individual. Ordinal measures reflect which person on object
is larger or smaller, heavier or lighter, brighter or duller, harder or softer
etc.
Persons may be grouped according to physical or psychological
traits to convey a relationship like ―greater than‖ or ―less than‖. Socio-
economic status is a good example of ordinal measurement.
In ordinal measurement besides the relationship of equivalence a
relationship of greater than or less than exists because all members of any
particular subclass are equivalent to each other and at the same time greater
or lesser than the members of other subclasses. The permissible statistical
operations in ordinal measurement are median, percentiles and rank
correlation co-efficient plus all those permissible for nominal
measurement. The drawback of ordinal measurement is that ordinal
5 Self-instructional Material
Statistics
measures are not absolute quantities, and do they convey that the distance
between the different rank values is equal. This is because ordinal
NOTES measurement are not equal-interval measurement nor do they incorporate
absolute zero point.
1.4.2.3 Interval or Equal-Interval Scale

This measurement includes all the characteristics of the nominal


and ordinal scales of measurement. The salient feature of interval
measurement is that numerically equal distances on the scale indicate equal
distance in the properties of the objects being measured. In other words,
here the unit of measurement is constant & equal. Since the numbers are
after equal intervals, they can legitimately be added and subtracted from
each other ie on an interval measurement the intervals or distances (not the
quantities or amounts) can be added. The difference or interval between the
numbers on the scale reflects differences in magnitude.
Although it is true that the intervals can be added, it does not mean
that the process of additivity can be carried out in the absolute sense. The
process of additivity of intervals or distances on an interval measurement
has only a limited value because in such measurement zero point is not true
but rather arbitrary. Zero point, here, does not tell the real absence of the
property being measured. The common statistics used in such measurement
are arithmetic mean, standard deviation, pearson‘s r. and other statistics
based on them. Statistic like the t test and F test which are widely used
tests of significance, can also be legitimately applied. In psychology,
sociology and education we frequently encounter interval measurement.

1.4.2.4 Ratio Scale of Measurement


It is the highest level of measurement and has all the properties of
nominal. Ordinal and interval scales plus an absolute or true zero point.
The salient feature of the ratio scale is that the ratio of two numbers is
independent of the unit of measurement and therefore, it can be
meaningfully equated.
Common examples of ratio scale are the measures of weight, width,
length, loudness and so on. It is common among physical sciences than
social sciences.

Check your progress -2

3. The number of eggs in a basket is an example of ____________


variable.
4. The different television shows make up a _____________ scale of
measurement.

6
Self-instructional Material
5. When measuring height to the nearest half inch , what are the real Statistics
limits for the score of 68 inches?
1.5 : Collection of Data:
While deciding about the method of data collection to be used for the NOTES
study, the researcher should keep in mind two types of data viz., primary
and secondary. The primary data are those which are collected afresh and
for the first time, and thus happen to be original in character. The
secondary data, on the other hand, are those which have already been
collected by someone else and which have already been passed through
the statistical process. The researcher would have to decide which sort of
data he would be using (thus collecting) for his study and accordingly he
will have to select one or the other method of data collection.

1.5.1: Primary Data: There are several ways in which primary data are
collected especially in surveys and descriptive researches. We shall review
the four most commonly employed ways of data collection.

1.5.1.1 Observation method: The observation method is the most


commonly used method specially in studies relating to behavioural sciences.
Observation becomes a scientific tool and the method of data collection for
the researcher, when it serves a formulated research purpose, is
systematically planned and recorded and is subjected to checks and controls
on validity and reliability. Under the observation method, the information is
sought by way of investigator‘s own direct observation without asking from
the respondent. For instance, in a study relating to consumer behaviour, the
investigator instead of asking the brand of wrist watch used by the
respondent, may himself look at the watch. The main advantage of this
method is that subjective bias is eliminated, if observation is done accurately.
Secondly, the information obtained under this method relates to what is
currently happening; it is not complicated by either the past behaviour or
future intentions or attitudes. Thirdly, this method is independent of
respondents‘ willingness to respond and as such is relatively less demanding
of active cooperation on the part of respondents as happens to be the case in
the interview or the questionnaire method. This method is particularly
suitable in studies which deal with subjects (i.e., respondents) who are not
capable of giving verbal reports of their feelings for one reason or the other
However, observation method has various limitations. Firstly, it is an
expensive method. Secondly, the information provided by this method is
very limited. Thirdly, sometimes unforeseen factors may interfere with the
observational task. At times, the fact that some people are rarely accessible
to direct observation creates obstacle for this method to collect data
effectively.
1.5.1.2 Interview method: The interview method of collecting data
involves presentation of oral-verbal stimuli and reply in terms of oral-verbal
responses. This method can be used through personal interviews and, if
7 Self-instructional Material
Statistics
possible, through telephone interviews. Personal interview method requires a
person known as the interviewer asking questions generally in a face-to-face
NOTES contact to the other person or persons. This method is particularly suitable
for intensive investigations.
The method of collecting information through personal interviews is
usually carried out in a structured way. Structured interviews involve the use
of a set of predetermined questions and of highly standardised techniques of
recording. As against it, the unstructured interviews are characterised by a
flexibility of approach to questioning. In a non-structured interview, the
interviewer is allowed much greater freedom to ask, in case of need,
supplementary questions or at times he may omit certain questions if the
situation so requires. But this sort of flexibility results in lack of
comparability of one interview with another and the analysis of unstructured
responses becomes much more difficult and time-consuming than that of the
structured responses obtained in case of structured interviews. Unstructured
interviews also demand deep knowledge and greater skill on the part of the
interviewer. Unstructured interview, however, happens to be the central
technique of collecting information in case of exploratory or formulative
research studies. But in case of descriptive studies, we quite often use the
technique of structured interview because of its being more economical,
providing a safe basis for generalisation and requiring relatively lesser skill
on the part of the interviewer.
1.5.1.3 Questionnaires: This method of data collection is quite popular,
particularly in case of big enquiries. It is being adopted by private individuals,
research workers, private and public organisations and even by governments.
In this method a questionnaire is sent to the persons concerned with a request
to answer the questions and return the questionnaire. A questionnaire consists
of a number of questions printed or typed in a definite order on a form or set of
forms. The questionnaire is mailed/or distributed individually to respondents
who are expected to read and understand the questions and write down the
reply in the space meant for the purpose in the questionnaire itself. The
respondents have to answer the questions on their own. The method of
collecting data by mailing the questionnaires to respondents is most
extensively employed in various economic and business surveys. The
following are the merits of the questionnaire method
1. There is low cost even when the universe is large and is widely
spread geographically
2. It is free from the bias of the interviewer; answers are in
respondents‘ own words.
3. Respondents have adequate time to give well thought out
answers.

8
Self-instructional Material
4. Respondents, who are not easily approachable, can also be Statistics
reached conveniently.
5. Large samples can be made use of and thus the results can be
made more dependable and reliable. NOTES
The main demerits of this system are listed below
1. Low rate of return of the duly filled in questionnaires; bias due
to no-response is often indeterminate.
2. It can be used only when respondents are educated and
cooperating.
3. The control over questionnaire may be lost once it is sent.
4. There is inbuilt inflexibility because of the difficulty of
amending the approach once questionnaires have been
despatched.
5. There is also the possibility of ambiguous replies or omission
of replies altogether to certain questions; interpretation of
omissions is difficult.
6. It is difficult to know whether willing respondents are truly
representative.
7. This method is likely to be the slowest of all.
Before using this method, it is always advisable to conduct
‗pilot study‘ for testing the questionnaires. In a big enquiry the
significance of pilot survey is felt very much. Pilot survey is the
rehearsal of the main survey. Such a survey, being conducted
by experts, brings to the light the weaknesses (if any) of the
questionnaires and also of the survey techniques. From the
experience gained in this way, improvement can be effected.
1.5.1.4 Schedules: This method of data collection is very much like the
collection of data through questionnaire, with little difference which lies in
the fact that schedules are being filled in by the enumerators who are
specially appointed for the purpose. These enumerators along with
schedules, go to respondents, put to them the questions from the proforma
in the order the questions are listed and record the replies in the space
meant for the same in the proforma. Enumerators explain the aims and
objects of the investigation and also remove the difficulties which any
respondent may feel in understanding the implications of a particular
question or the definition or concept of difficult terms. This method
requires the selection of enumerators for filling up schedules or assisting
respondents to fill up schedules and as such enumerators should be very
carefully selected. The enumerators should be trained to perform their job
welland the nature and scope of the investigation should be explained to
them thoroughly so that they may well understand the implications of
different questions put in the schedule. This method of data collection is
very useful in extensive enquiries and can lead to fairly reliable results. It
is, however, very expensive and is usually adopted in investigations

9 Self-instructional Material
Statistics
conducted by governmental agencies or by some big organisations.
Population census all over the world is conducted through this method.
NOTES 1.5.2: Secondary data: Secondary data may either be published data or
unpublished data. Usually published data are available in: (a) various
publications of the central, state are local governments; (b) various
publications of foreign governments or of international bodies and their
subsidiary organisations; (c) technical and trade journals; (d) books,
magazines and newspapers; (e) reports and publications of various
associations connected with business and industry, banks, stock
exchanges, etc.; (f) reports prepared by research scholars, universities,
economists, etc. in different fields; and (g) public records and statistics,
historical documents, and other sources of published information. The
sources of unpublished data are many; they may be found in diaries,
letters, unpublished biographies and autobiographies and also may be
available with scholars and research workers, trade associations, labour
bureaus and other public/ private individuals and organisations.
Researcher must be very careful in using secondary data. He must make a
minute scrutiny because it is just possible that the secondary data may be
unsuitable or may be inadequate in the context of the problem which the
researcher wants to study

Check your Progress -3

6. Population Census all over the world is conducted through


___________.
7. Why is it essential to do a pilot survey?
1.6 Classification of Data : Formation of frequency
distribution:
Tests, experiments and survey studies in education and psychology
provide us valuable data, mostly in the shape of numerical stores. For
understanding the meaning and deriving useful conclusion the data
have to the organized or arranged in systematic manner. One such
means of organizing the original data or computed statistics are called
frequency distribution.
A frequency table is a systematic testing of the number of scores of
each value in the group studied. It makes it easy for the investigator to see
a pattern in a large group of scores.

For example a frequency table for a nominal variable with the


following score will look like the following.

5, 7, 4, 5, 6, 5, 4

10
Self-instructional Material
Frequency Table Statistics

Value Frequency Percentage


7 1 14
6 1 14 NOTES
5 3 43
4 2 29

Sometimes there are so many possible values that an ordinary


frequency table is too awkward to give a simple picture of the scores. The
solutions is to make groupings of value that include all values within a
certain range. This combined category is a range of values that includes
many values and is called an interval. A frequency table that uses intervals
is called a frequency distribution. This shows all the possible scores and
their frequency of occurrence.

Guidelines for constructing a frequency distribution

A set of raw scores does not result in a unique set of grouped


scores. A given set of score can be grouped in more than one way. Some
widely accepted conventions that help us to group scores easily are as
follows.

a. Be sure that the class intervals are mutually exclusive

b. Make all intervals of the same width

c. Make the intervals continuous throughout the distribution

d. Place the interval containing the highest score value at the top

e. For most work, use 10 to 20 class intervals.

f. Choose a convenient interval width

g. When possible make the lower score limits multiples of the interval
width

Steps to construct a grouped frequency distribution

a. Finding the range - It is done by subtracting the lowest score from


the highest.

b. Determining the class interval or grouping interval: There are two


different means to get an idea of the size of the class interval, the
range is divided by the number of classes desired. Class interval is
usually denoted by the symbol i.

11 Self-instructional Material
Statistics
Range
i
No. of classes desired
NOTES
To decide the No. of classes desired as a general rule, Tate (1955)
writes.

If the series contains fewer than about 50 item, more than about 10
classes are not justified. It the series contain from about 50 to 100 items, 10
to 15 classes tend to be appropriate. If more than 100 items, 15 or more
class tend to be appropriate. Ordinarily not fewer than 10 classes or more
than 20 are used.

Class interval ‗i‘ can be decided first and then the number of
classes desired. A combined procedure that takes into account the range,
the number of classes and the class interval while planning for a frequency
distribution is must so that we are able to arrive at a frequency distributions
that minimizes grouping error. The wider the class interval width the
greater the potential for grouping error.

c. Writing the contents of the frequency distribution

The classes of the distribution written and the scores given in the
data are taken one by one and tallied in their proper classes. The tallies all
totaled and frequencies of each class interval is noted down.

Classes of Scores Tallies


Frequencies
65 - 69 1 1
60 - 64 111 3
55 - 59 1111 4
50 - 54 1111 11 7
45 - 49 1111 1111 9
40 - 44 1111 11111 11
35 - 39 1111 111 8
30 - 34 1111 4
25 - 29 11 2
20 - 24 1 1

----
Total frequencies No. 50

-----

12
Self-instructional Material
1.7 Lets Sum up: Statistics
In this unit, you have been introduced to basic statistical concepts. The
need and importance of statistics and its applications in the diverse
field has been presented. This knowledge will equip you to formulate NOTES
research questions in various fields of study. The nature of variables
and the scales of measurement will help you to operationally
conceptualise the variable that you choose for your research study, The
section on methods of data collection will provide insight on how to go
about collecting data for the chosen research study.

1.8 Unit – End exercises:


1. List the difference between schedules and questionnaires
2. Give examples of variables that can be measured using the various
scales of measurement.
3. List applications of statistics in day to day life.
1.9 Answer to check your progress
1. Qualitative data are not Statistics.
2. The two categories of statistical techniques are Descriptive and
inferential.
3. The number of eggs in a basket is an example of discrete variable.
4. The different television shows make up a nominal scale of
measurement.
5. 67.5 - 68.5 inches.
6. Population Census all over the world are conducted through schedules
7. Pilot study, is a small scale preliminary study conducted in order to
evaluate feasibility, time, cost, adverse events, and improve upon
the study design prior to performance of a full-scale research project.
1.10 Suggested readings
1. S.P. Gupta, Statistical Methods, Sultan Chand and Sons, New
Delhi, 2002.
2. Henry E., Garret‘s Statistics in Psychology and Education, Feffer
and Simans Private Limited, 1969, p.329.

13 Self-instructional Material
Diagrammatic and
Graphical
UNIT 2: DIAGRAMMATIC AND
NOTES GRAPHICAL
2.1 Introduction
2.1.1 General principles of graphical representation of data
2.1.2 Advantages of graphical representation of data
2.1.3 Disadvantages of graphical representation of data
2.2 Graphs For Interval and Ratio Data
1.2.1: Histogram
1.2.2: Frequency Polygons
1.2.3: Cumulative Percentage Curve or Ogive.
2.3 Graphs for Nominal and Ordinal Data
2.3.1: Bar Diagram
2.3.1:1: Simple bar diagram
2.3.1:2: Multiple bar diagram
2.3.1:3: Sub-divided bar diagram
2.3.1: 4: Percentage bar diagram
2.3.2: Pie Charts
2.4 Let us Sum up
2.5 Unit – End exercises
2.6 Answer to check your progress
2.7 Suggested readings

2.1 Introduction
A graph is another good way to make a large group of scores easy to
understand ―A picture is worth a thousand words‖ - and sometimes a
thousand numbers. A graph is a sort of chart through which statistical
data are represented in the form of lines or curves drawn across the
coordinated points plotted. In other words , it is a geometrical image of a
set of data.

General Principles of Graphic Representation:

There are some algebraic principles which apply to all types of graphic
representation of data. In a graph there are two lines called coordinate axes.
One is vertical known as Y axis and the other is horizontal called X axis.
These two lines are perpendicular to each other. Where these two lines
intersect each other is called ‗0‘ or the Origin. On the X axis the distances
14
Self-instructional Material
right to the origin have positive value (see fig. 7.1) and distances left to the Diagrammatic and
origin have negative value. On the Y axis distances above the origin have a Graphical
positive value and below the origin have a negative value.
NOTES

2.1.1 Advantages of graphical representation of data

1. Acceptability: Such report is acceptable to the busy people because


it easily highlights the theme of the report. This helps to avoid
wastage of time.
2. Comparative Analysis: Information can be compared easily. Such
comparative analysis helps for quick understanding and attention.
3. Decision Making: Business executives can view the graphs at a
glance and can make decision very quickly which is hardly possible
through descriptive report.
4. Helpful for less literate Audience: Less literate or illiterate people
can understand graphical representation easily because it does not
involve going through line by line of any descriptive report.
5. A complete Idea: Such representation creates clear and complete
idea in the mind of audience. Reading hundred pages may not give
any scope to make decision. But an instant view or looking at a
glance obviously makes an impression.
6. Use in the Notice Board: Such representation can be hanged in the
notice board to quickly raise the attention of employees in any
organization.

2.1.2 Disadvantages of Graphical Representation of Data

1. Lack of Secrecy: Graphical representation makes full presentation


of information which may hamper the objective to keep something
secret.

15 Self-instructional Material
Diagrammatic and
Graphical 2. More time: Normal report involves less time to represent but
graphical representation involves more time as it requires graphs
NOTES and figures .
3. Errors and Mistakes: Since graphical representations are
complex, there are chance of errors and mistake. This causes
problems for better understanding to lay people.

Check your progress- 1

1. What is the chief need for representing data graphically?


2. The point of intersection between two co-ordinate axis in a graph is
known as ________

2.2 Graphs For Interval and Ratio Data:


When the data consist of numerical scores that have been measured
on an interval or ratio scale, histograms , polygons and frequency
curves can be constructed

2.2.1: Histogram :To construct a histogram


i. Make a frequency distribution

ii. Put the values of real lower limits of the class interval along the
bottom of the page.

iii. Make a scale of frequencies along the Y axis that goes from 0 at the
bottom to the highest frequency of any value.

iv. Make a bar above each value with a height for the frequency of that
class interval. Each class or interval with its specific frequency is
represented by a separate triangle. The base of each rectangle is the
width of the class interval (i) and the height is the respective
frequency of that class or interval.

16
Self-instructional Material
Diagrammatic and
Graphical

NOTES

2.2.2 Frequency Polygons


Another way to graph a frequency table is to make a special kind of
limit graph called a frequency polygon. A frequency polygon is essentially
a line graph for graphical representation of the frequency distribution. We
can get a frequency polygon from a histogram, if the midpoints of the
upper bases of the rectangles are connected by straight lines. But it is not
essential to plot a histogram first to draw a frequency polygon.

There are five steps for making a frequency polygon.


a) Make the frequency table
b) Put the midpoints of the class intervals along the bottom of the
page, from left to right, starting one value below the lowest value
and ending one value above the highest value. The extra values
helps to close the figure.
c) Make a scale of frequencies along the left edge of the page that
goes from zero at the bottom to the highest frequency for any value.
d) Make a point above each value with the height for the frequency of
that value.
e) Connect the point with lines.

17 Self-instructional Material
Diagrammatic and
Graphical

NOTES

2.2.3 Cumulative percentage curve/ Ogive: When a population consists


of numerical scores from an interval or a ratio scale, it is customary to
draw the distribution with a smooth curve instead of the jagged, step-wise
shapes that occur with histograms and polygons. The smooth curve
indicates that you are not connecting a series of dots (real frequencies) but
instead are showing the relative changes that occur from one score to the
next. The first step in the construction of such graph is organising the data
in the form of cumulative frequency distribution, which is then converted
into cumulative percentages. Technically, a cumulative frequency
distribution is the sum of the class and all classes below it in a frequency
distribution. All that means is you're adding up a value and all of the
values that came before it.

Cumulative percentage indicates the percentage of scores that lies


below the upper real limit of the associated class interval. Therefore, in
constructing cumulative percentage curves, the cumulative percentage is
plotted at the upper real limit of the class interval. We then connect the
points with straight lines and bring the curve down to zero at the lower end
at the upper real limit of the next adjacent class interval. A cumulative
percentage curve never has a negative slope. The result in a cumulative
percentage curve with an S-shaped figure is called an Ogive. We may
determine percentiles or are percentile ranks from the cumulative
percentage curve.

18
Self-instructional Material
Diagrammatic and
Graphical

NOTES

Check your progress -2

3. What is plotted on the x –axis when you construct a frequency


polygon?
4. What is the advantage ogive over histogram and frequency
polygon?
5. What do you understand by the term cumulative frequency?

2.3 Graphs for Nominal and Ordinal Data:


Bar graphs and Pie-charts are widely used for this purpose
2.3.1: Bar Diagram: The bar diagram is very similar to the histogram and
is constructed in the same manner except that space appears between the
rectangles, thus suggesting the essential discontinuity of several categories.
However within categories, sub categories may be displayed as adjacent
bars. Because qualitative categories on a nominal scale of measurement
have no necessary order, we may arrange them in any order. However, for
ordinal scale of measurement, the categories should be arranged in order of
rank.

2.3.1:1: Simple bar diagram:A simple bar chart is used to represent data
involving only one variable classified on a spatial, quantitative or temporal
basis. In a simple bar chart, we make bars of equal width but variable
length, i.e. the magnitude of a quantity is represented by the height or
length of the bars.
The following steps are used to draw a simple bar diagram:

 Draw two perpendicular lines, one horizontally and the other vertically,
at an appropriate place on the paper.

19 Self-instructional Material
Diagrammatic and
Graphical  Take the basis of classification along the horizontal line (X−X− axis)
and the observed variable along the vertical line (Y−Y− axis), or vice
NOTES versa.
 Mark signs of equal breadth for each class and leave equal or not less
than half a breadth between two classes.
 Finally mark the values of the given variable to prepare required bars.

The above is an illustration of a simple bar chart showing the profit of a


particular bank over a five year period
2.3.1:2: Multiple bar diagram: In a multiple bars diagram two or more
sets of inter-related data are represented (multiple bar diagram facilitates
comparison between more than one phenomena). The technique of making
a simple bar chart is used to draw this diagram but the difference is that we
use different shades, colours, or dots to distinguish between different
phenomena.

The above graph is an illustration of the multiple bar diagram showing


import and export trade in a country over a 5 year time period.

20
Self-instructional Material
2.3.1:3: Sub-divided bar diagram: In this diagram, first we make Diagrammatic and
simple bars for each class taking the total magnitude in that class and then Graphical
divide these simple bars into parts in the ratio of various components.
NOTES
1. This chart consists of bars which are sub-divided into two or more
parts.
2. The length of the bars is proportional to the totals.
3. The component bars are shaded or coloured differently.
Below is a numerical table and graphical representation of the numerical
facts in a sub-divided bar diagram

Current and Development Expenditure – Pakistan (All figures in Rs.


Billion)
Current Development Total
Years
Expenditure Expenditure Expenditure
1988-89 153 48 201
1989-90 166 56 222
1990-91 196 65 261
1991-92 230 91 321
1992-93 272 76 348
1993-94 294 71 365
1994-95 346 82 428

2.3.1: 4: Percentage bar diagram:


A sub-divided bar chart may be drawn on a percentage basis. To
draw a sub-divided bar chart on a percentage basis, we express each
component as the percentage of its respective total. In drawing a

21 Self-instructional Material
Diagrammatic and
Graphical percentage bar chart, bars of length equal to 100 for each class are drawn in
the first step and sub-divided into the proportion of the percentage of their
NOTES component in the second step. The diagram so obtained is called a
percentage component bar chart or percentage stacked bar chart. This type
of chart is useful to make comparisons in components holding the
difference of total constants.

2.3.2: Pie Charts: This data is presented in the form of a circle. There are
segments and sectors into which a pie chart is being divided and each of these
segments and sectors forms a certain portion of the total (in terms of
percentage). In the pie-chart, the total of all the data is equal to 360 degrees.
The degree of angles that are used to represent different items are calculated in
the form of proportionality. In other words, the total frequencies or value is
equated to360 degrees , and then the angles corresponding to component parts
are calculated (or the component parts are are expressed as percentages of the
total and then multiplied by 3.6). After determining these angles, the required
sectors in the circle are drawn. In pie charts, the entire diagram looks like a pie
and the components in it resembles the various slices cut from the pie. The
pie-chart is thus used to show the break-up of one continuous variable into its
components parts.

Check your Progress -3

3 When is simple bar diagram used?


4 When do we use a sub-divided bar diagram?

Shows the distribution of sales of the laptop industry between five companies

22
Self-instructional Material
Diagrammatic and
Graphical

NOTES

4.3 Let us Sum up:


The statistical data may be presented in a more attractive form
appealing to the eye with the help of graphic aids. Such presentation
helps in communicating information to both well read and lay people.
It helps viewer‘s to have an immediate and meaningful grasp of the
large amount of data. For graphical representation of data measured on
an interval and ratio scale, histogram, polygons and Ogives are used.
For graphical representation of data measured on nominal and ordinal
scale, bar charts and pie charts are used. The kind of bar diagram you
choose depend on the nature of data.
2.5 Unit – End exercises:
1. What is the process for constructing histogram?
2. Draw a suitable bar diagram for the following data:The data shows
the number of students (Both male and female) graduating from the
year 2012-2017 of a city college.
Year of 2012 2013 2014 2015 2016 2017
graduation
Gender 105 110 95 100 85 115
Male
Female 115 125 105 85 150 200

3. State the advantages and disadvantages of graphical representation


of data.

2.6 Answer to check your progress:


1. The value of graphs lies in the fact that it is an economic device for
presentation, understanding and interpretation of collected
statistical data.
2. Origin
3. Midpoints of the class –interval
4. From the ogive we can infer relative changes in the data.
Percentiles and percentile ranks can be graphically computed from
ogives.
23 Self-instructional Material
Diagrammatic and
Graphical 5. Cumulative frequency is the total of a frequency and all frequencies
so far in a frequency distribution. It is the 'running total' of
NOTES frequencies.
6. We use simple bar diagram when we have data in which one
variable is classified on a spatial, qualitative or temporal basis.
7. A sub-divided or component bar chart is used when we have to
represent data in which the total magnitude has to be divided into
different components.
2.7 Suggested readings:

1. Garrett.H.E.(1981). Statistics in Psychology and Education. Vakils,


Feffer and Simons Ltd.
2. Gupta, S.P. (2014). Statistical Methods. (44the.d.) New Delhi
:Roopak Printers.
3. Mangal S.K. (1987). Statistics in Psychology and Education. New
Delhi: PHI Learning Private Ltd.
4. https://byjus.com/maths/graphical-representation/
5. http://www.psychologydiscussion.net/educational-
psychology/statistics/measures-of-central-tendency-and-how-it-
helps-in-educational-psychology/2741

24
Self-instructional Material
UNIT 3: MEASURES OF CENTRAL Measures Of Central
Tendency
TENDENCY
3.1 Introduction
3.2 Arithmetic Mean
3.2.1: Computation of mean from ungrouped data NOTES
3.2.2: Computation of mean from grouped data
3.2.3 : Properties of the mean
3.3Median
3.3.1: Computation of median from ungrouped data
3.3.2: Computation of median from grouped data
3.3.3 : Properties of the median
3.4 Mode
3.4.1: Computation of mode from ungrouped data
3.4.2: Computation of mode from grouped data
3.4.3Properties of the mode
3.5 Geometric Mean
3.6 Harmonic Mean.
3.7 Let us Sum up
3.8 Unit – End exercises
3.9 Answer to check your progress
3.10 Suggested readings

3.1 Introduction
The purpose of any measure of central tendency is to provide a single
summary figure that best describes the central location of our entire
distribution of observations. Such measures are useful in comparing the
performance of a group with that of a standard reference group.

A measure of central tendency also helps simplify comparisons of


two of more groups listed under different conditions. There are many
measures of central tendency. The three most commonly used in education
and behavioural sciences are:

a) Arithmetic Mean b) Median c) Mode

3.2 Arithmetic Mean:


The arithmetic mean in the sum of all the score in a distribution divided by
the total number of scores.
3.2.1: Computation of mean from ungrouped data

Let X1, X2, X3, X4, X5, X6, X7, X8, X9, X10be the scores obtained by 10
students on an achievement test. The arithmetic mean of the group of 10
students can be calculated as,

M = X1 + X2 + X3 + X4 --------- +X10
_______________________________________
10
25 Self-instructional Material
Measures Of Central The formula for calculating the mean of an ungrouped data is
Tendency
∑X
M= where X stands for the sum of score or values of the
NOTES N
items and N for the total number of items in a series or group.

3.2.2: i)Computation of mean from grouped data(Long Method)

Scores f Mid points (X) fX


65 - 69 1 67 67
60 - 64 3 62 186
55 - 59 4 57 228
50 - 54 7 52 364
45 - 49 9 47 423
40 - 44 11 42 462
35 - 39 8 37 296
30 - 34 4 32 128
25 - 29 2 27 54
20 - 24 1 22 22
---------
-----
fX =
2230
---------
------

In a frequency distribution where all the frequencies are greater


than one, the mean is computed by the formula,

∑ fX
M=
N

where X represents the midpoints of the interval ,f its frequency and N the
total of all frequency

∑ fX 2230
M=   44.6
N 50

ii)Short cut method of computing the mean from grouped


data(Assumed mean method)

Mean is computed with the help of the formula

Self-instructional Material 26
∑ fx1 Measures Of Central
M =A+ 1 Tendency
N

where A = Assumed mean (The midpoint of the class interval with the
highest frequency is usually taken as the assumed mean).
NOTES
i = Class interval

f = Respective frequency of the mid-values of the class


intervals

N = Total frequency

X A
x' =
i

The quotient obtained after division of the difference


between the mid-value of the class interval and assumed
mean by i, the width class intervals

Note: The population mean is denoted by the symbol µ

Scores f X x' (X-A) f x'


65 - 69 1 67 5 5
60 - 64 3 62 4 12
55 - 59 4 57 3 12
50 - 54 7 52 2 12
45 - 49 9 47 1 9
40 - 44 11 42 0 0
35 - 39 8 37 -1 -8
30 - 34 4 32 -2 -8
25 - 29 2 27 -3 -6
20 - 24 1 22 -4 -4
--------- ---------
--
N = 50
fx  26
'

--------- ---------
---

 fx1 26
  A  i  42   5
N 50
27 Self-instructional Material
Measures Of Central = 44.6
Tendency
3.2.3 : Properties of the mean
NOTES
a) Mean is responsive to the exact position of each score on the
distribution.

b) The mean may be thought of as balance point of a distribution. This


says that if we express the scores in terms of negative and positive
deviation their sum is zero.

c) The mean is more sensitive to the presence or absence of score at


the extremes of the distribution. When a measure of central
tendency should reflect the total of the scores the mean is the best
choice.

d) Changing the value of any score changes the mean. For example, a
sample of quiz scores for a psychology lab section consists of 9, 8,
7, 5, and 1. Note that the sample consists of n = 5 scores with ΣX -=
30 .The mean for this sample is 6. Now suppose that the score of X
= 1 is changed to X =8. Note that we have added 7 points to this
individual‘s score, which also adds 7 points to the total (∑X). After
changing the score, the new distribution consists of 9, 8, 7, 5, 8, the
mean becomes 7.4
.
e) Adding a new score to a distribution, or removing an existing score,
usually changes the mean. The exception is when the new score (or
the removed score) is exactly equal to the mean.

f) If a constant value is added to every score in a distribution, the


same constant is added to the mean. Similarly, if you subtract a
constant from every score, the same constant is subtracted from the
mean.

g) If every score in a distribution is multiplied by (or divided by) a


constant value, the mean changes in the same way.
Check your progress – 1

1. Adding a new score to a distribution always changes the mean


(True/ False).
2. A population has a mean of µ = 60. If 5 points were added to every
score, what would be the value for the new mean?
3. What do you understand by the term central tendency?

Self-instructional Material 28
3.3 Median Measures Of Central
The Median of a distribution is the point along the scale of possible Tendency
score below which 50% of the scores fall. The median thus is the value
that divides the distribution into halves. It symbol is Mdn. For raw
scores, we may think of the median as a middle score of a distribution
based on score frequency. To determine the median therefore we need NOTES
to know where all the scores are located.

3.3.1: Computation of median from ungrouped data

To find the median we must first put the scores in rank order from lowest
to highest. If N is odd, the median will be score that has an equal number
of scores below and above it. For example, for the following scores.

1, 7, 8, 11, 15, 16, 20

The median is 11

When there is an even number of scores, there is no middle score,


so the median is taken as the point halfway between the two score that
bracket the middle position. The example, for group of scores:

12, 14, 15, 18, 19, 20

the median is 15 + (18 - 15) / 2 = 16.5

This again shows that the position of each score in the series needs to be
ascertained first and then median should be determined.

3.3.2: Computation of median from grouped data

The formula for computing median from grouped data are as


follows:

 ( N / 2)  F 
Mdn  L   i
 f 

L = Exact lower limit of the median classes

F = Total of all frequencies before the median class

f = frequency of the median class

i = Class interval width

N = Total of all the frequencies

29 Self-instructional Material
Measures Of Central Illustration: Compute the median for the frequency distribution given
Tendency below:

NOTES Scores f
65 - 69 1
60 - 64 3
55 - 59 4
50 - 54 7
45 - 49 9
40 - 44 11
35 - 39 8
30 - 34 4
25 - 29 2
20 - 24 1
-------
N = 50
---------

Since median is the central item, for this distribution it is likely to


fall somewhere between the scores of 25th & 26th items. In a given
frequency distribution table if we add frequencies either from above or
below we may see that the class interval designated as 40-44 is to be
labeled as a class where the score representing median will fall.

By applying the above formula we get

 50 / 2  15 
Mdn  395   5
 11 
10
 395   5  44.05
11

3.3.3 : Properties of the median

The median responds to how many score lie below (or above) it but
not to how far away the scores may be. A little below the median or way
below, both count the same in determining its value. Thus, the median is
less sensitive than the mean to the presence of a few extreme scores.
Therefore in distribution that are strongly asymmetrical or have few very
deviant scores the median may be the better choice for measuring the
central tendency, if we wish to represent the bulk of the scores and not give
undue weight to the relatively few deviant once.

Self-instructional Material 30
In behavioural studies, there are occasion when the researcher Measures Of Central
cannot record the exact value of scores at the upper end of the distribution. Tendency
Distributions like this are called open-ended. In open-ended distribution,
we cannot calculate the mean without making assumption. But we can find
the median. It is also used when there are undetermined (Infinite) scores
that make it impossible to compute a mean NOTES

Of the three measures of central tendency, the median stands


second to the mean in ability to resist the influence of sampling
fluctuations in ordinary circumstances. For large sample taken from a
normal distribution, the median varies about one quarter more from sample
to sample than does the mean. For small samples, the median is relatively
better.

3.4 Mode:
The final measure of central tendency that we consider is called the mode.
In its common usage, the word mode means ―the customary fashion‖ or ―a
popular style.‖ The statistical definition is similar in that the mode is the
most common observation among a group of scores. In a frequency
distribution, the mode is the score or category that has the greatest
frequency.

3.4.1 : Computation of mode from ungrouped data


In ungrouped distribution, the mode is the score that occurs with the
greatest frequency. In grouped data, it is taken as the midpoint of the class
interval that contains the greatest number of scores. The symbol for mode
to Mo.

The mode is a useful measure of central tendency because it can be used to


determine the typical or average value for any scale of measurement,
inducing a nominal scale . Consider, for example, the data given below

Restaurant Frequency
College Grill l5
George & Harry‘s 16
Luigi‘s 42
Oasis Diner 18
Roxbury Inn 7
Sutter‘s Mill 12

These data were obtained by asking a sample of 100 students to name their
favourite restaurants in the city. For these data, the mode is Luigi‘s, the
restaurant (score) that was named most frequently as a favourite place.
Although we can identify a modal response for these data, you should
notice that it would be impossible to compute a mean or a median. For
31 Self-instructional Material
Measures Of Central example, you cannot add the scores to determine a mean (How much is 5
Tendency College Grills plus 42 Luigi‘s?). Also, it is impossible to list the scores in
order because the restaurants do not form any natural order. For example,
NOTES the College Grill is not ―more than‖ or ―less than‖ the Oasis Diner, they are
simply two different restaurants. Thus, it is impossible to obtain the median
by finding the midpoint of the list. In general, the mode is the only measure
of central tendency that can be used with data from a nominal scale of
measurement.

3.4.3 Properties of the Mode:

The mode is easy to obtain, but it is not very stable from sample to
sample further, when quantitative data are grouped, the mode may be
strongly affected by the width and location of class interval. In addition,
there may be more than one mode for a particular set of scores. The mode
in the only measure that can be used for data that have the character of the
nominal scale.

Check your progress – 2

4. If you have a score of 82 on an 80- exam point , then you scored


definitely above the median(True/False).
5. During the month of October, an instructor recorded the number of
absences for each student in a class of n= 20 and obtained the
following distribution. Using the mode, What is the average
number of absences for a class
No:of absences 6 4 3 1 2 0

F 1 2 7 5 3 2

3.5 Geometric Mean:

The Geometric Mean is a special type of average where we multiply the


numbers together and then take a square root (for two numbers), cube
root (for three numbers) .For n numbers: multiply them all together and
then take the nth root (written n√ )

More formally, the geometric mean of n numbers a1 to an is:

n
√(a1 × a2 × ... × an)

Since it takes into account the compounding that occurs from period to
period, in certain contexts the geometric mean is more accurate measure
Self-instructional Material 32
than the arithmetic mean, such as in business where it is employed for Measures Of Central
describing proportional growth, both exponential growth (constant Tendency
proportional growth) and varying growth

3.6 Harmonic Mean


Harmonic mean is a type of average generally used for numbers NOTES
that represent a rate or ratio such as the precision and the recall in
information retrieval. The harmonic mean can be described as the
reciprocal of the arithmetic mean of the reciprocals of the data. It is
calculated by dividing the number of observations by the sum of
reciprocal of the observation. The formula is:

Harmonic means are often used in averaging things like rates (e.g.,
the average travel speed given aduration of several trips)

3.7 Let us Sum up:


The purposeof the central tendency is to determine a single value that
identifies the centre of the distribution. The three standard measures of
central tendency are the mean, median and mode. The mean is the
arithmetic average It is computed by adding all of the scores and then
dividing by the number of scores.Changing any score in the distribution
causes the mean to be changed. When a constant value is added to (or
subtracted from) every score in a distribution, the same constant value
is added to (or subtracted from) the mean. If every score is multiplied
by a constant, the mean is multiplied by the same constant. The median
is the midpoint of a distribution of scores. The median is the preferred
measure of central tendency when a distribution has a few extreme
scores . The median also is used for open-ended distributions and when
there are undetermined (infinite) scores that make it impossible to
compute a mean. Finally, the median is the preferred measure of central
tendency for data from an ordinal scale. . The mode is the most
frequently occurring score in a distribution .For data measured on a
nominal scale, the mode is the appropriate measure of central tendency.
It is possible for a distribution to have more than one mode. Geometric
and Harmonic mean has unique properties but are rarely used in social
sciences.

33 Self-instructional Material
Measures Of Central 3.8 Unit – End exercises:
Tendency
1. For the following sample, find the mean, median and mode. The
NOTES scores are:
5, 6, 8, 9, 11, 5, 12, 8, 9, 6, 9
2. One question on a student survey asks: In a typical week, how
many times do you eat at fast food restaurant? The following
frequency distribution table summarizes the results for a sample of
n=20 students.

Number of times 5 or more 4 3 2 1 0


per week
F 2 2 3 6 4 3
a) Find the mode for this distribution
b) Find the median for this distribution
c) Explain why you cannot compute the mean using the data in
the table.
3. A sample of n-= 5 scores has a mean =12. If one person with the
score of X=8 is removed from the sample , what is the value for
the new mean?
4. Identify the circumstances in which the median rather than the
mean is a preferred measure of central tendency.
5. Compute mean , median and mode from the following data:
Class 20- 25- 30- 35- 40- 45- 50- 55-
intervals 24 29 34- 39 45 49 54 59
F 2 7 10 11 8 8 2 2

3.9 Answer to check your progress


1. False. If the score is equal to the mean, it does not change the
mean..
2. The new mean score would be 45.
3. Central tendency is a statistical measure to determine a single score
that defines the centre of a distribution. The goal of central
tendency is to find a single score that is most typical or most
representative of the entire group.
4. False. The value of the median depends on where all of the scores
are located.The mode is 3.

3.10 Suggested Readings :

1. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural


sciences (9th Ed.). Cengage Learning India Private Ltd.
2. Gupta, S.P. (2014). Statistical Methods. (44the.d.) New Delhi
:RoopakPrinters.

Self-instructional Material 34
UNIT 4: MEASURES OF DISPERSION Measures Of Dispersion

4.1 Introduction NOTES


4.2 Range
4.2.1 Properties of the range
4.3 Quartile deviation
4.3.1 Properties of the quartile deviation
4.4 Standard deviation and Variance for populations
4.4.1 Computation of standard deviation from
ungrouped and grouped data
4.4.2 Properties of the Standard deviation
4.4.3 Computation of variance
4.5 Standard deviation and Variance for samples
4.5.4: Sampling variability and degrees of freedom
4.6 Concept of skewness and kurtosis
4.6.1 Karl Pearson co-efficient of skewness
4.6.2 Bowley’s Co-efficient of skewness
4.6.3 Kurtosis
4.7 Let us Sum up
4.8 Unit – End exercises
4.9 Answer to check your progress
4.10 Suggested readings

4.5 Introduction

Variability expresses quantitatively the extent to which the scores


in the distribution scatter about or cluster together. In general, a good
measure of variability serves two purposes:

1. Variability describes the distribution. Specifically, it tells whether the


scores are clustered close together or are spread out over a large distance.
Usually, variability is defined in terms of distance. It tells how much
distance to expect between one score and another, or how much distance to
expect between an individual score and the mean. For example, we know
that the heights for most adult males are clustered close together, within 5
or 6 inches of the average. Although more extreme heights exist, they are
relatively rare.

2. Variability measures how well an individual score (or group of scores)


represents the entire distribution. This aspect of variability is very
important for inferential statistics, in which relatively small samples are
used to answer questions about populations. For example, suppose that you
selected a sample of one person to represent the entire population. Because
most adult males have heights that are within a few inches of the
population average (the distances are small), there is a very good chance
that you would select someone whose height is within 6 inches of the
35 Self-instructional Material
Measures Of Dispersion population mean. On the other hand, the scores are much more spread out
(greater distances) in the distribution of weights. In this case, you probably
would not obtain someone whose weight was within 6 pounds of the
population mean. Thus, variability provides information about how much
error to expect if you are using a sample to represent a population. The
NOTES various measures of variability are:

4.2 The Range:

The simplest measure of variability is the range. The range is the difference
between the highest and lowest score in the distribution. The range is the
distance and not a location.

For example range for the following set of scores will be


3, 5, 5, 8, 15, 14, 18, 23
Range = 23 – 3 = 20

In a grouped frequency distribution, calculate the range as the


difference between the value of the lowest raw score that could be included
in the bottom class interval and that of the highest score that could be
included in the upper most class interval.

4.2.1Properties of the Range:

The range is easier to compute and is ideal for preliminary work or


in other circumstances where precision is not an important requirement.

The range has some major shortcoming. The range is not sensitive
to the total condition of the distribution. The range is also of little use
beyond the descriptive level.

4.3 Quartile Deviation: (QD)

The quartile deviation, also termed as the semi - inter quartile


range, symbolized by the letter Q, is a more sophisticated measure that
depends only on the relatively stable central portion of a distribution
specifically, on the middle 50% of the score. It is defined as one-half the
distance between the first and third quartile points

Q3  Q1
Q
2
or
P75  P25
Q
2

The quartile points are the three score points that divide the
distribution into four parts, each containing an equal number of cases.

Self-instructional Material 36
4.3.1: Properties of the Quartile deviation: Measures Of Dispersion

QD or the semi-inter quartile range is closely, related to the median


because both are defined in terms of percentage points of the distribution. NOTES
The median is responsive to the number of scores lying below it rather than
to their exact position and P25 and P75 are points defined in a similar way.
We may therefore expect median and the semi inter quartile range to have
properties in common.

The semi inter quartile range is less sensitive to the presence of few
extreme scores than is the standard deviation. If a distribution is badly
skewed or if it contains a few very extreme scores the semi inter qualities
range will respond to the presence of such scores, but it will not give them
undue weight. With open-ended distribution the QD may be reasonable
measure to compute.

Check your progress

1. What is the purpose of measuring variability? The purpose for


measuring variability is to obtain an objective measure of how the
scores are spread out in a distribution.
2. What is range? It is the distanced covered by the scores in the
distribution from the smallest to the largest score
3. Semi interquartile deviation have similar properties to that of

4.4: Standard deviation and Variance for populations:


The standard deviation is the most commonly used and the most
important measure of variability. Standard deviation uses the mean of the
distribution as a reference point and measures variability by considering
the distance between each score and the mean The first step in finding the
standard distance from the mean is to determine the deviation, or distance
from the mean, for each individual score. By definition, the deviation for
each score is the difference between the score and the mean. Deviation is
distance from the mean:
Deviation score = X – µ
For a distribution of scores with µ (Population mean) = 50, if your score is
X = 53, then your deviation score is X-µ= 53 – 50 = 3
If your score is X = 45, then your deviation score is X-µ = 45 – 50 = –5
Notice that there are two parts to a deviation score: the sign (+ or –) and
the number. The sign tells the direction from the mean—that is, whether
the score is located above (+) or below (–) the mean. The number gives the
actual distance from the mean. For example, a deviation score of –6
corresponds to a score that is below the mean by a distance of 6 points.

Because our goal is to compute a measure of the standard distance


from the mean, the

37 Self-instructional Material
Measures Of Dispersion obvious next step is to calculate the mean of the deviation scores. To
compute this mean, you first add up the deviation scores and then divide by
N. When you do this you will find that sum of deviations around the mean
will be zero, because of the positive and negative signs. The solution is to
get rid of the signs (+ and –). The standard procedure for accomplishing
NOTES this is to square each deviation score. Using the squared values, you then
compute the mean squared deviation, which is called variance. Population
variance equals the mean squared deviation. Variance is the average
squared distance from the mean.

Our goal is to compute a measure of the standard distance from the


mean. Variance, which measures the average squared distance from the
mean, is not exactly what we want. The final step simply takes the square
root of the variance to obtain the standard deviation, which measures the
standard distance from the mean. Standard deviation is the square root of
the variance and provides a measure of the standard, or average, distance
from the mean. In other words, Standard deviation of a set of scores also
defined as the square root of the average of the squares of the deviation of
each score from the mean. This simply means standard deviation is the
square root of the variance

( X  M ) 2  x2
SD  
N N
where
X = is the individual score
M = Mean of the given set of scores
N = Total No. of scores
x = Deviation of each score from the mean

Standard deviation is regarded as the most stable and reliable


measure of variability as it employs the mean for its computation. It is
often called root mean square deviation and is denoted by the greek letter
sigma ().

4.4.1 Computation of standard deviation (SD) for ungrouped data.

Standard deviation can be computed from the ungrouped scores by


the formula.

 x2

N

Calculate SD for the following set of scores

Self-instructional Material 38
52, 50, 56, 68, 65, 62, 57, 70 Measures Of Dispersion

Mean of the given scores


NOTES
M = 480 / 8 = 60

Scores X Deviation from the mean x2


52 -8 64
60 -10 100
56 -4 16
68 8 64
65 5 25
62 2 4
57 -3 9
70 10 100
x2=382
Applying the obtained values in the formula we get

382

8
 47.75  6.91

Computation of SD from grouped data

Scores F X M X x2 fx2
127 – 129 1 128 115 13 169 169
124 – 126 2 125 115 10 100 200
121 – 123 3 122 115 7 49 147
118 – 120 1 119 115 4 16 16
115 – 117 6 116 115 1 1 6
112 – 114 4 113 115 -2 4 16
109 – 111 3 110 115 -5 25 75
106 – 108 2 107 115 -8 64 128
103 – 105 1 104 115 -11 121 121
100 – 102 1 101 115 -14 196 196
N = 24 fx2 = 1074
The formula for computing SD from grouped data is

 fx 2 1074
SD    6.69
N 24

4.4.2: Properties of Standard Deviation


The standard deviation, like the mean, is response to be exact
position of every score in the distribution. Because it is calculated by
39 Self-instructional Material
Measures Of Dispersion taking deviations from the mean, if a score is shifted to a position more
deviant from the mean, the standard deviation will increase. If the score is
shifted to a position closer to the mean, the standard deviation decreases.
Thus, the standard deviation is more sensitive to the condition of the
distribution than is either the range or semi-inter quartile range.
NOTES
The standard deviation is more sensitive than the semi - inter
quartile range to the presence or absence of scores that lie at the extremes
of the distribution. This should be obvious because the semi-inter quartile
range is based only on the score points that make the border of the middle
50% of the scores. Because of this characteristic sensitivity, the standard
deviation may not be the choice among measures of variability when the
distribution contains a few very extreme score or when the distribution is
badly skewed. For example, if we compare the variability of two
distributions where only one of them contains several extreme score, the
extreme scores will exert an influence on the standard deviation that is
disproportionate to their relative number. Of course, if N is quite large and
the extreme score are very few they will make little difference.

When we calculate deviation from the mean the sum of squares of


their values is smaller than if they had been taken about any other point. In
other words, it is the point about which the sum of squares of deviation
scores is minimum.

One of the most important points favouring our use of standard


deviation is its resistance to sampling variation. In repeated random
samples drawn from population, of the type most frequently encountered in
statistical works, the numerical value of the standard deviation tends to
jump about less than would that of other means computed on the same
samples. If our concern is in any way associated with inferring variation in
a population from knowledge of variation in the sample, the property is
clearly of worth.

Standard deviation appears explicitly or lies embedded in many


procedures of both descriptive and inferential statistics.

4.4.3 Computation of variance

We will calculate the variance and standard deviation for the following
population of N = 5 scores: 1, 9, 5, 8, 7

Score X Deviation X – µ Squared Deviation


(X –µ ) 2
1 -5 25
9 3 9
5 -1 1
8 2 4
7 1 1
Self-instructional Material 40
40 (Sum of squared Measures Of Dispersion
deviation
For this set of N = 5 scores, the squared deviations add up to 40. The mean
of the squared deviations, the variance, is 40/5 = 8 NOTES

Standard Deviation which is the square root of variance is a


measure of dispersion like variance. But it is used more often than
variance because the unit in which it is measured is the same as that of
mean, a measure of central tendency. The advantage of variance is that it
treats all deviations from the mean the same regardless of direction; as a
result, the squared deviations cannot sum to zero.

4.5: Standard deviation and variance for samples:


The goal of inferential statistics is to use the limited information from
samples to draw general conclusions about populations. The basic
assumption of this process is that samples should be representative of the
populations from which they come. This assumption poses a special
problem for variability because samples consistently tend to be less
variable than their populations. The fact that a sample tends to be less
variable than its population means that sample variability gives a biased
estimate of population variability. This bias is in the direction of
underestimating the population value rather than being right on the mark.
Fortunately, the bias in sample variability is consistent and predictable,
which means it can be corrected.

The calculations of variance and standard deviation for a sample follow the
same steps that were used to find population variance and standard
deviation.

1. Find the deviation from the mean for each score: deviation = X – M (M
represents sample mean while µ represents population mean)

2. Square each deviation: squared deviation = (X – M) 2

3. Add the squared deviations: Sum of squared deviation = Σ (X – M) 2 or


SS (Sum of squares)

Sample variance s2 = Σ (X – M) 2/ n-1

Note that the denominator is n-1 , instead of n. This is the adjustment that
is necessary to correct for the bias in sample variability. The effect of the
adjustment is to increase the value that you obtain. Dividing by a smaller
number (n – 1 instead of n) produces a larger result and makes sample
variance an accurate and unbiased estimator of population variance.

4.5.1: Sample Variability and Degrees of Freedom: Although the


concept of a deviation score and the calculation of Sum of squares (SS)
are almost exactly the same for samples and populations, the minor
41 Self-instructional Material
Measures Of Dispersion differences in notation are really very important. Specifically, with a
population, you find the deviation for each score by measuring its distance
from the population mean, µ. With a sample, on the other hand, the value
of µ is unknown and you must measure distances from the sample mean.
Because the value of the sample mean varies from one sample to another,
NOTES you must first compute the sample mean before you can begin to compute
deviations. However, calculating the value of M places a restriction on the
variability of the scores in the sample. This restriction is demonstrated in
the following example.

Suppose we select a sample of n 3 scores and compute a mean of M= 5.


The first two scores in the sample have no restrictions; they are
independent of each other and they can have any values. For this
demonstration, we assume that we obtained X = 2 for the first score and X
= 9 for the second. At this point, however, the third score in the sample is
restricted

X Sample of n = 3 scores with a mean of M = 5.


2
9
? What is the third score?
. For this example, the third score must be X= 4. The reason that the third
score is restricted to X =4 is that the entire sample of n =3 scores has a
mean of M =5. For 3 scores to have a mean of 5, the scores must have a
total of X =15. Because the first two scores add up to 11 (9 +2), the third
score must be X =4.

In this the first two out of three scores were free to have any values, but the
final score was dependent on the values chosen for the first two. In general,
with a sample of n scores, the first n – 1 scores are free to vary, but the
final score is restricted. As a result, the sample is said to have n – 1
degrees of freedom. For a sample of n scores, the degrees of freedom, or
df, for the sample variance are defined as df= n – 1. The degrees of
freedom determine the number of scores in the sample that are independent
and free to vary.

Check your progress 2

4. What does standard deviation and variance measure?


5. The deviation scores are calculated for each individual in a
population of N = 4. The first three individuals have deviations of
2, 4, and –1. What is the deviation for the fourth individual?
6. What is the standard deviation for the following set of N = 5
scores: 10, 10, 10, 10, and 10?
7. Explain why the formula for sample variance divides SS by n – 1
instead of dividing by n.

Self-instructional Material 42
4.6 Concept of Skewness and Kurtosis: Measures Of Dispersion

Measures of Skewness and Kurtosis, like measures of central tendency and


dispersion, study the characteristics of a frequency distribution. Averages NOTES
tell us about the central value of the distribution and measures of
dispersion tell us about the concentration of the items around a central
value. These measures do not reveal whether the dispersal of value on
either side of an average is symmetrical or not. If observations are arranged
in a symmetrical manner around a measure of central tendency, we get a
symmetrical distribution, otherwise, it may be arranged in an asymmetrical
order which gives asymmetrical distribution. Thus, skewness is a measure
that studies the degree and direction of departure from symmetry.

4.6.1 Karl Pearson Co-efficient of Skewness

Karl Pearson developed two methods to find skewness in a sample.


1. Pearson‘s Coefficient of Skewness method 1 uses the mode. The
formula is:

Where = the mean, Mo = the mode and s = the standard


deviation for the sample.

2. Pearson‘s Coefficient of Skewness method 2 uses the median. The


formula is:

Where = the mean, Mo = the mode and s = the standard


deviation for the sample.

Example : Use Pearson‘s Coefficient #1 and #2 to find the skewness for


data with the following characteristics:
 Mean = 70.5.
 Median = 80.
 Mode = 85.
 Standard deviation = 19.33.
Pearson’s Coefficient of Skewness #1 (Mode):
Step 1: Subtract the mode from the mean: 70.5 – 85 = -14.5.
Step 2: Divide by the standard deviation: -14.5 / 19.33 = -0.75.
Pearson’s Coefficient of Skewness #2 (Median):
Step 1: Subtract the median from the mean: 70.5 – 80 = -9.5.
Step 2: Multiply Step 1 by 3: -9.5(3) = -28.5
Step 2: Divide by the standard deviation: -28.5 / 19.33 = -1.47.
43 Self-instructional Material
Measures Of Dispersion Caution: Pearson‘s first coefficient of skewness uses the mode. Therefore,
if the mode is made up of too few pieces of data it won‘t be a stable
measure of central tendency.
Interpretation
 The direction of skewness is given by the sign.
NOTES  The coefficient compares the sample distribution with a normal
distribution. The larger the value, the larger the distribution differs
from a normal distribution.
 A value of zero means no skewness at all.
 A large negative value means the distribution is negatively skewed
(happens when there are more individuals at the right side (positive
end) of the distribution.
 A large positive value means the distribution is positively skewed
(happens when there are more individuals at the right side (negative
end) of the distribution.
4.6.2 Bowley’s Co-efficient of skewness :

Bowley‘s method of skewness is based on the values of median,


lower and upper quartiles. This method suffers from the same limitations
which are in the case of median and quartiles. Wherever positional
measures are given, skewness should be measured by Bowley‘s method.
This method is also used in case of ‗open-end series‘, where the
importance of extreme values is ignored. Coefficient of skewness lies
within the limit ± 1. This method is quite convenient for
determining skewness where one has already calculated quartiles.

4.6.3 Kurtosis

When there are very few individuals whose scores are near to the
average score for their group (too few cases in the central area of the
curve) the curve representing such a distribution becomes ‗flattened in the
middle. On the other hand, when there are too many cases in central area,
the distribution curve becomes too ‗peaked in comparison to normal‘. Both
these characteristics of being flat (Platykurtic) on peaked (Leptokurtic), are
used to describe the term Kurtosis.

Check your progress 3

8. When the frequency distribution graph is flattened at the top it is


termed as _________
9. In an easy test, many individuals score higher than normal. When
the scores are represented graphically, it is likely to result in
_____________.

4.7 Let us Sum up:


The purpose of variability is to measure and describe the degree to
which the scores in a distribution are spread out or clustered together.
Self-instructional Material 44
In this unit, we covered four measures of variability: the range, the Measures Of Dispersion
quartile deviation. the variance, and the standard deviation. The range
is the distance covered by the set of scores, from the smallest score to
the largest score. The range is completely determined by the two NOTES
extreme scores and is considered to be a relatively crude measure of
variability.
Standard deviation and variance are the most commonly used
measures of variability. Both of these measures are based on the idea
that each score can be described in terms of its deviation, or distance,
from the mean. The variance is the mean of the squared deviations.
The standard deviation is the square root of the variance and provides
a measure of the standard distance from the mean. We also studied
how to describe distributions in which the scores are erratically
scattered through the concept of skewness and kurtosis.
4.8 Unit – End exercises
1. Calculate the variance for the following population of N = 5
scores: 4, 0, 7, 1, 3.
2. For the following set of scores: 1, 5, 7, 3, 4 .Assume that this is a
sample of n = 5 scores and compute SS and variance for the
sample.
3. Write the properties of quartile deviation.
4. Explain the concept of degrees of freedom.

4.9 Answer to check your progress


1. The purpose for measuring variability is to obtain an objective
measure of how the scores are spread out in a distribution.
2. It is the distanced covered by the scores in the distribution from the
smallest to the largest score
3. Median
4. Standard deviation measures the standard distance from the mean
and variance measures the average squared distance from the mean.
5. The deviation scores for the entire set must add up to zero. The first
four deviations add to 5 so the fifth deviation must be –5.
6. Because there is no variability (the scores are all the same), the
standard deviation is zero.
7. Without some correction, sample variability consistently
underestimates the population variability. Dividing by a smaller
number (n – 1 instead of n) increases the value of the sample
variance and makes it an unbiased estimate of the population
variance.
8. Platykurtic
9. Negatively skewed.

45 Self-instructional Material
Measures Of Dispersion 4.10 Suggested readings:
6. Gupta, S.P. (2014). Statistical Methods. (44the.d.) New Delhi
:Roopak Printers.

NOTES 7. Garrett.H.E.(1981). Statistics in Psychology and Education. Vakils,


Feffer and Simons Ltd.
8. https://us.sagepub.com/sites/default/files/upm-
binaries/66956_Tokunaga_Chapter_4.pdf

Self-instructional Material 46
Correlation

UNIT 5: CORRELATION
5.1: Introduction to Correlation NOTES
5.1.1: Characteristics of relationship

5.2 Scatter Diagram

5.3 Methods of computing co-efficient of correlation

5.3.1Karl Pearson co-efficient of correlation

i. Spearman‘s rank correlation


5.4 Regression

5.4.1: Regression Line

5.4.2: Regression Equation

5.6.1 Properties of regression co-efficient.


b. Let us Sum up
c. Unit – End exercises
d. Answer to check your progress
5.8 Suggested Readings

5.1 Introduction to Correlation

Correlation : Correlation is a statistical technique that is used to measure


and describe a relationship between two variables. Usually the two
variables are simply observed as they exist naturally in the environment -
there is no attempt to control or manipulate the variables.

For example, a researcher interested in the relationship between


nutrition and IQ could observe and record dietary patterns for a group of
pre-school children and then measures IQ scores for the same group.
Notice that the researcher is not trying to manipulate the children‘s diet or
IQ but is simply observing what occurs naturally. Correlation requires two
scores for each individual (one score from each of the two variables).
These scores are normally identified as X and Y. The pairs of scores can be
tested in a table or can the presented graphically in a scatter pot.

In the scatter plot, the X values are placed on the vertical axis. Each
individual is then identified by a single point on the graph so that the, co-
ordinates of the point (the X & Y values) match the individuals X score
and Y score.

5.1.1The characteristics of a relationship

A correlation measures three characteristics of the relationship


between X and Y. The three characteristics are as follows

47 Self-instructional Material
Correlation
1. The direction of the relationship: Correlations can be classified
into two basic categories positive and negative.
NOTES
In positive correlation, the two variables tend to move in the same
direction. When the X variable increases, the Y variable also increases, if
the X variable decreases, the Y variable also decreases.

In negative correlation, the two variables tend to go in opposite


direction. As the X variable increases, the Y variable decreases that is, it is
an inverse relationship.

The direction of a relationship is identified by the sign of the


correlation. A positive value (+) indicates a positive relationship; a
negative value (-) indicates a negative relationship.

2. The Form of Relationship


The most common use of correlation is to measure straight - line
relationships. However, others forms of relationships do exist and that
there are special correlations used to measure them. For example the
relationship between reaction time and age. Reaction time improves with
age until the late teens, when it reaches a peak, after that reaction time
starts to get worse. This shows a curved relationship.
3. The Degree or the strength of Relationship
Finally, a correlation measures how well the data fit the specific
form being considered. For example, a linear correlation measures how
well the data points fit on a straight line. A perfect correlation always is
identified by a correlation of 1.00 and indicates a perfect fit. At the other
extreme, a correlation of 0 indicates no fit at all. Intermediate values
represent the degree to which the data points approximate the perfect fit.
The numerical value of the correlation also reflects the degree to which
there is a consistent, predictable relationship between the two variables.
Again, a correlation of 1.00 (or - 1.00) indicates a perfect consistent
relationship.
5.2: The scatter diagram:

A Scatter (XY) Plot has points that show the relationship between two sets
of data. Pairs of numerical data, with one variable on each axis are plotted
to look for a relationship between them. If the variables are correlated, the
points will fall along a line or curve. The better the correlation, the tighter
the points will hug the line. Below are examples of scatter plots showing
positive , negative and zero correlation.

Self-instructional Material 48
Correlation

NOTES

5.3 Methods of computing Co-efficient of correlation


For expressing the degree of relationship quantitatively between
two sets of measures or variables, we usually take the help of an index that
is known as co-efficient of correlation. It is a ratio which expresses the
extent to which changes in one variable are accompanied by chance in the
other variable. It involves no units and varies from - 1 to +1.In case the co-
efficient of correlation is zero. it indicates zero correlation between two
sets of measures.

Check your Progress -1

1. For each of the following, indicate whether you would expect a positive
or a negative correlation.
a. Model year and price for a used Honda
b. IQ and grade point average for high school students
c. Daily high temperature and daily energy consumption for 30 winter days
in New York City

2. The data points would be clustered more closely around a straight line
for a correlation
of –0.80 than for a correlation of _0.05. (True or false?)

5.3.1 Pearson Product moment correlation

The Pearson‘s correlation measures the degree and the direction of


linear relationship between two variables.

The Pearson‘s correlation is identified by the letter r and computed


by the formula.

deg ree to which x and y var y together


r
deg ree to which x and y var yseperately

When there is a perfect linear relationship, every change in the X variable


is accompanied by a corresponding change in the Y variable.

To calculate the Pearson‘s correlation, it is necessary to compute


the sum of products of deviations. The sum of products provides a parallel
49 Self-instructional Material
Correlation
procedure for measuring the amount of co - variability between two
variables. The Pearson‘s correlation consists of a ratio comparing the co-
NOTES variability of X and Y (the numerator) with the variability of X and Y
separately (the denominator).

Computation correlation of Pearson Product moment

Individuals X Y X Y Xy x2 y2
A 15 60 -10 10 -100 100 100
B 25 70 0 20 0 0 400
C 20 40 -5 -10 50 25 100
D 30 50 5 0 0 25 0
E 35 30 10 -20 -200 100 400

xy = 250 x2=250


y2=1000

Mean of series x, x = 25

Mean of series y, y = 50

xy 250
r =
x 2 y 2 250  1000

= 0.5

xy- sum of product of deviations from the mean.

Understanding and interpreting the Pearson correlation.

There are three additional consideration that we need to


keep in mind while dealing with Pearson correlation.

1. Correlation simply describes a relationship between two variables.


It does not explain why the two variables are related. The
correlation cannot be interpreted as proof of a cause and effect
relationship between two variables.

2. The value of a correlation can be affected greatly by the range of


scores represented in the data.

3. When judging ―how good‖ a relationship is, it is tempting to focus


on the numerical value of the correlation for example, a correlation
of +.5 is halfway between 0 and 1.00 and therefore appears to
represent a moderate degree of relationship. However correlation
should not be interpreted as a proportion. Although a correlation of
1.00 does mean that there is a 100% perfectly predictable
Self-instructional Material 50
Correlation
relationship between X and Y, a correlation of 0.5 does not mean
that you can make predictions with 50% accuracy. To describe how
accurately one variable predicts the other, you must square the NOTES
correlation. Thus, a correlation 0.5 provides only r 2 = 0.52 = 0.25 or
25% accuracy. The value of r2 is called as the co-efficient of
determination because it measures the proportion of variability in
one variable that can be determined from the relationship with the
other variable. Computed correlation co-efficient may be
understood by summarizing it as follows.

The range of computed


Interpretation
correlation co-efficient
1. 0 (zero value) Zero correlation, absolutely no
relationship
2. From 0.00 to + 0.20 Slight, almost negligible
relationship
3. From + 0.21 to + 0.40 Low correlation, definite but small
relationship
4. From + 0.41 to + 0.70 Moderate correlation, substantial
but small relationship
5. From + 0.71 to + 0.90 High correlation, marked
relationship
6. From + 0.91 to 0.99 Very high correlations, quite
dependable relationship
7. +1 Perfect correlation, almost identical
and opposite relationship
The use of the product moment method for the computation of correlation
co-efficient between two variable is based on the following assumptions.

1. Linearity of relationship: The relationship should be linear


(described by a straight line).

2. Homoscedasticity: The standard deviation of the scores in the


different columns and rows should be equal and fairly homogenous.

3. Continuity of the variables: The two variables for which correlation


is to be computed should be continuous variables in terms of
measurement.

4. Normality of the distribution: Distributions in two variables should


be fairly symmetrical and unimodal. In other words it should not be
badly skewed.

51 Self-instructional Material
Correlation
5.3.2 Spearman’s Rank-Order correlation

For computing the co-efficient of correlation between two sets of


NOTES
scores achieved by individuals with the help of this method, we require
rank; ie position of merit of these individuals in possession of certain
characteristics. The co-efficient of correlation computed by this method, is
known as the rank correlation co-efficient as it considers only the ranks of
the individuals in the characteristics A and B. It is designated by the Greek
letter  (rho). Sometimes, it is also known as Spearman‘s co-efficient of
correlation after the name of its inventor.

It we do not have scores and have to work with data in which


differences between the individuals can be expressed by ranks, rank
correlation co-efficient is the only way. But this does not mean that it
cannot be computed from the usual data given in raw scores. In case the
data contains scores of individuals, we can compute  by converting
individual scores into ranks.

Computation of Rank correlation co-efficiencies:

Difference
Marks Ranks Ranks ranks,
Marks Difference
in in in irrespective
Individuals in civics squared
History History Civics of +ve or - 2
(Y) (d )
(X) (R1) (R2) ve signs R1-
R2= d
A 80 82 2 3 1 1
B 45 86 11 2 9 81
C 55 50 10 10 0 0
D 56 48 9 11 2 4
E 58 60 8 9 1 1
F 60 62 7 8 1 1
G 65 64 6 7 1 1
H 68 65 5 6 1 1
I 70 70 4 5 1 1
J 75 74 3 4 1 1
K 85 90 1 1 0 0
d2=92

6d 2
 = 1
N(N 2  1)

Self-instructional Material 52
Correlation
6x92
= 1
11(112  1)
NOTES
23
= 1 = 1-0.42 = 0.58
55
Check your progress – 2

3. When is a spearman‘ s rho used?

4. What is measured by pearson correlation?

5. How is co-efficient of correlation different from coefficient of


determination?

5.4 Regression:
The statistical technique for finding the best-fitting straight line for a set of
data is called regression, and the resulting straight line is called the
regression line. Regression is used to denote the estimation and prediction
of the average value of one variable for a specified value of the other
variable. This estimation is done by deriving a suitable equation on the
basis of available bivariate data. This equation is called Regression
equation and its geometrical representation is called Regression curve. The
regression equation requires the Regression coefficient.

5.4.1Regression Line:
When the bivariate data are plotted on graph paper, the concentration of
points shows certain pattern showing the relationship. When the trend
points are found to be linear then by least square method we can obtain the
regression line.

If two variables are linearly related then the relation can be expressed as y
= bx + a, where ‗b‘ is the slope of the line and ‗a‘ is the intercept of that
line. This line serves several purposes. 1.The line makes the relationship
53 Self-instructional Material
Correlation
between the two variables easier to see. 2. The line identifies the center, or
central tendency, of the relationship, just as the mean describes central
NOTES tendency for a set of scores. 3. Finally, the line can be used for prediction.
5.4.2 Regression Equation:

In general, a linear relationship between two variables X and Y can be


expressed by the equation
Y = bX+ a (where a and b are fixed constants)
For example, a local video store charges a membership fee of $5 per
month, which allows you to rent videos and games for $2 each. With this
information, the total cost for 1 month can be computed using a linear
equation that describes the relationship between the total cost (Y) and the
number of videos and games rented (X).
Y =2X + 5

In the general linear equation, the value ofb is called the slope. The slope
determines how much the Y variable changes when X is increased by 1
point. For the video store example, the slope is b = 2 and indicates that
your total cost increases by $2 for each video you rent. The value of a in
the general equation is called the Y-intercept because it determines the
value of Y when X = 0. (On a graph, the avalue identifies the point where
the line intercepts the Y-axis.).In other words, Constant a is distance
between the point of origin and the point where the regression line touches
Y axis. Constant b is the steepness or slope of line and is also called
coefficient of regression. In the video store example, a = 5; there is a $5
membership charge even if you never rent a video.

When X= 3 When X=8


Y = bX+ a Y = bX+ a
= $2(3) + $5 = $2(8) _ $5
= $6 _ $5 = $16 _ $5
= $11 = $21

Regression Equation through Least Square Method:


Regression equation can be derived through least square method. Linear
regression equations are the equations of those lines of best fit which are
made on the basis of least square method. To determine how well a line
fits the data points, the first step is to define mathematically the
distance between the line and each data point.

Self-instructional Material 54
Correlation

NOTES

Figure: The X and Y data points and the regression line for the n = 8
pairs of scores

For every X value in the data, the linear equation determines a Y


value on the line. This value is the predicted Y and is called 𝑌(―Y
hat"). The distance between this predicted value and the actual Y
value in the data is determined by

distance = Y- 𝑌

Note that we simply are measuring the vertical distance


between the actual data point. This distance measures the error
between the line and the actual data. Because some of these
distances are positive and some are negative, the next step is to
square each distance to obtain a uniformly positive measure of error.
Finally, to determine the total error between the line and the data, we
add the squared editors for all of the data points. The result is a
measure of overall squared error between the line and the data:

total squared error = (Y - 𝑌)2

Now we can define the best-fitting line as the one that has the
smallest total squared - error. For obvious reasons, the resulting line is
commonly called the least-squared-error solution. In symbols, we are
looking for a linear equation of the for

𝑌 = bX + a

55 Self-instructional Material
Correlation
For each value of X in the data, this equation determines the
point on the line (Y) that gives the best prediction of Y. The problem
NOTES
is to find the specific values for a andb that make this the best-fitting
line.

The calculations that are needed to find this equation require


calculus and some sophisticated algebra, so we do not present the
details of the solution. The results, however, are relatively
straightforward, and the solutions for b and aare as follows:

𝑆𝑃
b=r
𝑆𝑆𝑥

whereSP is the sum of products and SSXis the sum of squares


for the X scores.

A commonly used alternative formula for the slope is based


on the standard deviations for X and Y. The alternative formula is

𝑆𝑌
b=r
𝑆𝑥

wheresyis the standard deviation for the Y scores, sxis the


standard deviation for the X scores, and r is the Pearson correlation
for X and Y. The value of the constant ain the equation is determined
by

a = My – bMx

Note that these formulas determine the linear equation that


provides the best prediction of Y values. This equation is called the
regression equation for Y

𝑌 = bX + a

The scores in the following table are used to demonstrate the


calculation and use of the regression equation for predicting Y.

Self-instructional Material 56
Correlation
:
X Y X - Mx Y - My (X-Mx)2 (Y-My)2 (X- Mx)(Y -My)
2 3 -2 -5 4 25 M,10
Y) NOTES
Mx)2
6 11 2 3 4 9 6

0 6 -4 -2 16 4 8

4 6 0 -2 0 4 0

7 12 3 4 9 16 12

5 7 1 -1 1 1 -1

5 10 1 2 1 4 2

3 9 -1 1 1 1 -1

SSx =36 SSy =64 SP = 36

For these data, ∑X = 32. soMx= 4. Also, ∑Y = 64, so My = 8.


These values have been used to compute the deviation scores for each X
and Y value. The final three columns show the squared deviations for X
and for Y, and the products of the deviation scores.

Our goal is to find the values for b and a in the regression equation.
Using Equations, the solutions for b and a are

𝑆𝑃 36
b= = = 100
𝑆𝑆𝑥 36

a = My - bMx = 8 - 1(4) = 4.00

The resulting equation is

𝑌= X + 4
\

USING THE REGRESSION EQUATION FOR PREDICTION

As we noted at the beginning of this section, one common use of


regression equations is for prediction. For any given value of X, we can use
the equation to compute a predicted value for Y. For the equation from

57 Self-instructional Material
Correlation
Example, an individual with a scope ofX = 1 would be predicted to
have a Y score of
NOTES
𝑌= X + 4 = 1 + 4 = 5

Although regression equations can be used for prediction, a few


cautions should be considered whenever you are interpreting the predicted
values:

USING THE REGRESSION EQUATION FOR PREDICTION

1. The predicted value is not perfect (unless r = +1.00 or -1.00). Although


the amount of error varies from point to point, on average the errors are
directly related to the magnitude of the correlation. With a correlation
near 1.00 (or - 1.00), the data points generally are clustered close to the
line and the error is small. As the correlation gets nearer to zero, the
points move away from the line and the magnitude of the error increases.

2. The regression equation should not be used to make predictions for X


values that fall outside of the range of values covered by the original
data.

5.5 Properties of regression co-efficient:

1. It is denoted by b.
2. It is expressed in terms of original unit of data.
3. Between two variables (say x and y), two values of regression
coefficient can be obtained. One will be obtained when we consider
x as independent and y as dependent and the other when we
consider y as independent and x as dependent. Both regression
coefficients must have the same sign.
4. If one regression coefficient is greater than unity, then the other
regression coefficient must be lesser than unity.

Check your progress -3


6..A local gym charges a $25 monthly membership fee plus $2 per hour for
aerobics classes. What is the linear equation that describes the relationship
between the total monthly cost (Y) and the number of class hours each
month (X)?

Self-instructional Material 58
Correlation
7. For the following linear equation, what happens to the value of Y each
time X is increased by 1 point?
Y= 3X + 7 NOTES
8.If the slope constant (b) in a linear equation is positive, then a graph of
the equation is a line tilted from lower left to upper right. (True or false?)
5.7 Let us Sum up:
A correlation measures the relationship between two variables, X and Y.
The relationship is described by three characteristics: Direction, Form and
strength of the relationship. The most commonly used correlation is the
Pearson correlation, which measures the degree of linear relationship. The
Pearson correlation is identified by the letter r A correlation between two
variables should not be interpreted as implying a causal relationship.
Simply because X and Y are related does not mean that X causes Y or that Y
causes X. To evaluate the strength of a relationship, you square the value of
the correlation. The resulting value, r2, is called the coefficient of
determination because it measures the portion of the variability in one
variable that can be predicted using the relationship with the second
variable.
The Spearman correlation measures the consistency of direction in the
relationship between X and Y—that is, the degree to which the relationship
is one-directional, or monotonic. A correlation always has a value from
+1.00 to –1.00. If you obtain a correlation outside this range, then you have
made a computational error. When interpreting a correlation, do not
confuse the sign (+ or –) with its numerical value. The sign indicates the
direction of the relationship between X and Y. On the other hand, the
numerical value reflects the strength of the relationship. We can sketch a
scatter plot of the data and make an estimate of the correlation.
When there is a general linear relationship between two variables, X and Y,
it is possible to construct a linear equation that allows you to predict the Y
value corresponding to any known value of X. Predicted Y value = Yˆ= bX+
a.The technique for determining this equation is called regression. By
using a least-squares method to minimize the error between the predicted Y
values and the actual Y values, the best-fitting line is achieved

5.7 Unit – End exercises

1. For the following data, compute the Pearson correlation.

X 2 1 3 2 4
Y 5 6 3 7 4

2.Rank the following scores and compute the Spearman correlation:


X 2 12 9 10 11
Y 7 35 6 19 8

59 Self-instructional Material
Correlation

3.List assumptions that our data needs to satisfy to compute pearson‘s


NOTES product moment correlation.

4. Compute linear regression for the following set of scores


X 0 2 8 6 4
Y 4 1 10 9 6

5.8 Answer to check your progress

1. a. Positive: Higher model years tend to have higher prices.

b. Positive: More intelligent students tend to get higher grades.

c. Negative: Higher temperature tends to decrease the need for heating.

2.True. The numerical value indicates the strength of the relationship. The
sign only indicate the direction of the relationship.

3. It is used when the researcher wants to establish relationship between


two sets of scores measured on an ordinal scale.

4. The Pearson correlation measures the degree and direction of the linear
relationship between two variables.

5. While coefficient of correlation tells whether changes in one variable are


accompanied by changes in the other, co-efficient of determination
measures the proportion of variability in one variable that can be
determined from the relationship with the other variable.

6.Y= 2X + 25

7. The slope is - 3, so Y decreases by 3 points each time X increases by 1


point.

8.True. A positive slope indicates that Y increases (goes up in the graph)


when X increases
5.9 Suggested Readings

1. Garrett.H.E.(1981). Statistics in Psychology and Education.Vakils,


Feffer and Simons Ltd.
2. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural
sciences (9th Ed.).Cengage Learning India Private Ltd.
3.https://www.simplypsychology.org/correlation.html
Self-instructional Material 60
Concept of Probability
UNIT 6: CONCEPT OF PROBABILITY
NOTES
6.1 Introduction
6.2 Approaches to Probability
6.3 Random Sampling
6.4 Basic Rules of Probability
6.4.1Addition theorem
6.4.2Multiplication theorem for dependent and independent
events
6.5 Let us Sum up
6.6 Unit – End exercises
6.7 Answer to check your progress
6.8 Suggested Readings

6.1 Introduction:

The role of inferential statistics is to use the sample data as the basis for
answering questions about the population. To accomplish this goal,
inferential procedures are typically built around the concept of probability.
Specifically, the relationships between samples and populations are usually
defined in terms of probability. Suppose, for example, that you are
selecting a single marble from a jar that contains 50 black and 50 white
marbles. (In this example, the jar of marbles is the population and the
single marble to be selected is the sample.) Although you cannot guarantee
the exact outcome of your sample, it is possible to talk about the potential
outcomes in terms of probabilities. In this case, you have a 50-50 chance of
getting either colour. Now consider another jar (population) that has 90
black and only 10 white marbles. Again, you cannot predict the exact
outcome of a sample, but now you know that the sample probably will be a
black marble. By knowing the makeup of a population, we can determine
the probability of obtaining specific samples. In this way, probability gives
us a connection between populations and samples, and this connection is
the foundation for the inferential statistics
A random experiment is an action or process that leads to one of many
possible outcomes. Examples:

Experiment Outcomes
Flip a coin Heads , Tails
Roll a dice 1,2,3,4,5,6
Exam Marks 0-100
Course Grades A,B,C,D,

The list of possible outcomes of a random experiment must be exhaustive


and mutually exclusive. The set of all possible outcomes of an experiment
is called the sample space. An individual outcome in the sample space is
61
Self-instructional Material
Concept of Probability
called a simple event, while an event is a collection or set of one or more
simple events in a sample space.
NOTES
Given a sample space S = {O1, O2, . ..}, the probabilities assigned to
events must satisfy these requirements:

1. The probability of any event must be nonnegative, e.g., P(Oi) ≥ 0 for


each i.

2. The probability of the entire sample space must be 1, i.e., P(S) = 1.

3. For two disjoint events A and B, the probability of the union of A and B
is equal to the sum of the probabilities of A and B, i.e., P(A ∪ B) = P(A) +
P(B) .

6.2 Approaches to probability:

a) Classical approach to assigning values to events: If there are a finite


number of possible outcomes of an experiment, all equally likely and
mutually exclusive, then the probability of an event is the number of
outcomes favourable to the event, divided by the total number of
possible outcomes. Customary examples include tossing an unbiased coin
or throwing a balanced dice.

Example 1: What are the chances of getting a ' Head' in tossing an


unbiased coin? There are only two equally likely outcomes, namely head
and tail. In our day to day language, we say that the coin has chance 1 in 2
of showing up a head. In technical language, we say that the probability of
getting a head is 1 / 2.

Example 2: Similarly, in the experiment of rolling a dice, there are six


equally likely outcomes 1, 2,3,4,5 or 6. The face with number '1' has
chance 1 in 6 of appearing on the top. Thus, we say that the probability of
getting 1 is 1/ 6.

Probability can be defined as a proportion, or a part of the whole. This


definition makes it possible to restate any probability problem as a
proportion problem.

Example 3: The probability problem ―What is the probability of selecting


a king from a deck of cards?‖ can be restated as ―What proportion of the
whole deck consists of kings?‖ In each case, the answer is 4’52, or ―4 out
of 52.‖ This translation from probability to proportion may seem trivial
now, but it is a great aid when the probability problems become more
complex

.
b) Relative frequency Approach: Most situations, however, do not
involve equally likely outcomes. Nor does this definition explain what

62
Self-instructional Material
Concept of Probability
probability is, it just states how to assign a numeric value to
this primitive idea in certain simple cases. Therefore there is another
approach of assigning probable values. In this approach, the probability of NOTES
an event denotes the relative frequency of occurrence of that event in the
long run. This means that to estimate the probability the experiment has to
be repeated indefinitely under identical conditions, atleast in principle.

A relative frequency distribution shows the proportion of the total


number of observations associated with each value or class of values and is
related to a probability distribution, which is extensively used in
statistics.
Let A be an event of interest, and assume that you have performed the
same experiment n times so that n is the number of times A could have
occurred. Further, let nA be the number of times that A did occur. Now,
consider the relative frequency nA/n. Then, in this method, we ―attempt‖ to
define P (A) as: P(A) = nA /n . The above can only be viewed as an
attempt because it is not physically feasible to repeat an experiment an
infinite number of times. Another important issue with this definition is
that two sets of n experiments will typically result in two different ratios.
However, we expect the discrepancy to converge to 0 for large n. Hence,
for large n, the ratio nA/n may be taken as a reasonable approximation for
P (A).

Example 1: Roll of a Dice S = {1, 2, · · · , 6} Probabilities: Roll the given


dice 100 times (say) and suppose the number of times the outcome 1 is
observed is 15. Thus, A = {1}, nA = 15, and n = 100. Therefore, we say
that P(A) is approximately equal to 15/100 = 0.15.

Example 2: Computer Sales A computer store tracks the daily sales of


desktop computers in the past 30 days. The resulting data is:

Desktop Sold No: of Days


0 1
1 2
2 10
3 12
4 5
5 or more 0
Therefore the approximate possibilities are:

Desktop Sold No: of Days Probability


0 1 1/30 = 0.03
1 2 2/30 = 0.07
2 10 10/30 = 0.33
3 12 12/30 = 0.40
4 5 5/30 = 0.17
5 or more 0 0
63
Self-instructional Material
Concept of Probability
Thus, for example, there is a 40% chance that the store will sell 3 desktops
on any given day.
NOTES
6.3 Random Sampling :
For the definition of probability to be accurate, it is necessary that
the outcomes be obtained by a process called random sampling. A
random sample requires that each individual in the population has an
equalchance of being selected. A second requirement, necessary for
many statistical formulas, states that if more than one individual is
being selected, the probabilities must stay constant from one selection
to the next. Adding this second requirement produces what is called
independent randomsampling. The term independent refers to the fact
that the probability of selecting any particular individual is independent
of those individuals who have already been selected for the sample. For
example, the probability that you will be selected is constant and does
not change even when other individuals are selected before you are.

Each of the two requirements for random sampling has some interesting
consequences. The first assures that there is no bias in the selection
process. For a population with N individuals, each individual must have the
same probability, p =1 ÷ N, of being selected. This means, for example,
that you would not get a random sample of people in your city by selecting
names from a yacht-club membership list. Similarly, you would not get a
random sample of college students by selecting individuals from your
psychology classes. You also should note that the first requirement of
random sampling prohibits you from applying the definition of probability
to situations in which the possible outcomes are not equally likely.
Consider, for example, the question of whether you will win a million
dollars in the lottery tomorrow. There are only two possible alternatives.

1. You will win.


2. You will not win.

According to our simple definition, the probability of winning would be


one out of two. However, the two alternatives are not equally likely, so the
simple definition of probability does not apply. The second requirement
also is more interesting than may be apparent at first glance. Consider, for
example, the selection of n = 2 cards from a complete deck. For the first
draw, the probability of obtaining the jack of diamonds is p ( jack of
diamonds) = 1÷52.After selecting one card for the sample, you are ready
to draw the second card. What is the probability of obtaining the jack of
diamonds this time? Assuming that you still are holding the first card, there
are two possibilities:

p(jack of diamonds) = 1 51 (if the first card was not the jack of diamonds)
or
64
Self-instructional Material
Concept of Probability
p(jack of diamonds) = 0 ( if the first card was the jack of diamonds)

In either case, the probability is different from its value for the first draw. NOTES
This contradicts
the requirement for random sampling, which says that the probability must
stay constant. To keep the probabilities from changing from one selection
to the next, it is necessary to return each individual to the population before
you make the next selection.This process is called sampling with
replacement.
Check your progress -1

1. A survey of the students in a psychology class revealed that there were


19 females and 8 males. Of the 19 females, only 4 had no brothers or
sisters, and 3 of the males were also the only child in the household. If a
student is randomly selected from this class,
a. What is the probability of obtaining a male?
b. What is the probability of selecting a student who has at least one
brother or sister?
c. What is the probability of selecting a female who has no siblings?

2. A jar contains 10 red marbles and 30 blue marbles.


a. If you randomly select 1 marble from the jar, what is the probability of
obtaining a red marble?
3.Which of these number cannot be a probability?

a)-0.00001
b)0.5
c)0
e) 1

6.4 Basic Rules of probability:


All definitions of probability must follow the same rules.

6.4.1: Union/Addition:

Let A and B be two events. Then, P (A ∪ B) = P(A) + P(B)−P(A ∩ B) .


The subtraction of P (A ∩ B) is necessary because A and B may ―overlap.‖
If A and B are mutually exclusive, i.e., A ∩ B = ∅, then P (A ∪ B) = P(A)
+ P(B) .

Example: Roll of a Dice

P(even) = 3/6 and P(odd) = 3/6

P(even and odd) = P({2}) = 1/6

P(even or odd) = 3/6 + 3/6 − 1/6 = 5/6

P({1} or {6}) = 1/6 + 1/6 − 0 = 2/6


65
Self-instructional Material
Concept of Probability
Conditional Probability:

NOTES Let A and B be two events. Then, the conditional probability of A given
that B has occurred, P (A | B), is defined as: P(A | B) = P(A ∩ B)/ P(B) .
(1) The reasoning behind this definition is that if B has occurred, then only
the ―portion‖ of A that is contained in B, i.e., A ∩ B, could occur;
moreover, the original probability of A ∩ B must be recalculated to reflect
the fact that the ―new‖ sample space is B

Example: Pick a Card from a Deck

Suppose a card is drawn randomly from a deck and found to be an Ace.


What is the conditional probability for this card to be Spade Ace?

A = Spade Ace

B = an Ace

A ∩ B = Spade Ace

P(A) = 1/52

P(B) = 4/52; and P(A ∩ B) = 1/52

Hence,.P(A | B) = 1/52 divided by 4/52 = 1/ 4

Note that probability is defined as a proportion, or a part of the whole. This


definition makes it possible to restate any probability problem as a
proportion problem. For example, the probability problem ―What is the
probability of selecting a king from a deck of cards?‖ can be restated as
―What proportion of the whole deck consists of kings?‖ In each case, the
answer is 4’ 52, or ―4 out of 52.‖ This translation from probability to
proportion may seemtrivial now, but it is a great aid when the probability
problems become more complex

Check your progress -2

4.Find the probability of rolling a ‗3 with a dice.‘

5.Draw a random card from a pack of cards. What is the probability that
the card drawn is a face card?
6.4.2MultiplicationThe multiplication rule is used to calculate the joint
probability of two events. It is simply a rearrangement of the conditional
probability formula. When we know that a particular event B has occurred,
then instead of Sample space, we concentrate on B for calculating the
probability of occurrence of event A given B

Formally, P(A ∩ B) = P(A | B)P(B) ; or, P(A ∩ B) = P(B | A)P(A) .

66
Self-instructional Material
Concept of Probability
Example 1: Drawing a Spade Ace

A = an Ace NOTES
B = a Spade

A ∩ B = the Spade Ace

P(B) = 13/52; P(A | B) = 1/13 Hence, P(A ∩ B) = P(A | B)P(B) = 1 /13 X


13/ 52 = 1 /52

Example 2: Selecting Students

A statistics course has seven male and three female students. The professor
wants to select two students at random to help her conduct a research
project. What is the probability that the two students chosen are female?

A = the first student selected is female

B = the second student selected is female

A ∩ B = both chosen students are female

P(A) = 3/10; P(B | A) = 2/9

Hence, P(A ∩ B) = P(B | A)P(A) = 2 /9 x 3 /10 = 1 /1

Multiplication Theorem for Independent Events

The multiplication theorem on probability for dependent events can be


extended for the independent events. From the theorem, we have, P(A ∩ B) =
P(A) P(B | A). If the events A and B are independent, then, P(B | A) = P(B).
The above theorem reduces to

P(A ∩ B) = P(A) P(B).

This shows that the probability that both of these occur simultaneously is the
product of their respective probabilities.

Problem: A box contains 5 black, 7 red and 6 green balls. Three balls are
drawn from this box one after the other without replacement. What is the
probability that the three balls are

1. all black balls

2. of different colors

3. two black and one green black.


Solution: The total number of the balls in the box is 5 + 7 + 6 = 18. Let events
67
Self-instructional Material
Concept of Probability
B: drawing black balls.
R: drawing red balls.
NOTES G: drawing green balls.

The balls are drawn without replacement. For the first draw, there are 18 balls
to choose from. The number of balls gets lessened by 1 for the second draw
i.e., 18 − 1 = 17 and 16 for the third draw.

1. Probability that the three balls are all black = P(B1) P(B2 | B1) P(B3 |
B1 ∩ B2) = 5⁄18 × 4⁄17 × 3⁄16 = 5⁄408.

2. The probability that the three balls are all different in color = P(B1)
P(R1 | B1) P(G1 | B1 ∩ R1) = 5⁄18 × 7⁄17 × 6⁄16 = 35⁄816.

3. Probability that two black and one green balls are drawn = P(B1)
P(B2 | B1) P(G1 | B1 ∩ B2) = 5⁄18 × 4⁄17 × 6⁄16 = 5⁄204.
It does not matter which colour ball is drawn first.

6.5 Let us Sum up:

Probability is a branch of mathematics that deals with the


occurrence of a random event. This basic theory is also used in
the probability distribution. The Addition rule states that when two
events, A and B, are mutually exclusive, the probability that A or B will
occur is the sum of the probability of each event, minus the probability of
the overlap. P(A or B) = P(A) + P(B) - P(A and B). The Multiplication
Rule of probability states that, If events A and B come from the same
sample space, the probability that both A and B occur is equal to
the probability the event A occurs times the probability that B occurs,
given that A has occurred.

6.6 Unit – End exercises:

1. State the addition and multiplication rule of probability.

2. Tickets numbered 1 to 20 are mixed up and then a ticket is drawn at


random. What is the probability that the ticket drawn has a number which
is a multiple of 3 or 5?
3. A bag contains 2 red, 3 green and 2 blue balls. Two balls are drawn at
random. What is the probability that none of the balls drawn is blue?

4. State why it is important to sample with replacement.

6.7 Answer to check your progress

1. a. p = 8/27

68
Self-instructional Material
Concept of Probability
b. p = 20 /27
c. p = 4/27
NOTES
7 p = 10/40 = 0.25

8 a)-0.00001(A probability is always greater than or equal to 0 and less


than or equal to 1)

9 Sample Space = {1, 2, 3, 4, 5, 6}


Number of favourable event = 1
Total number of outcomes = 6
Thus, Probability, P = 1/6
5. A standard deck has 52 cards.
Total number of outcomes = 52
Number of favourable events = 4 x 3 = 12 (considered Jack, Queen
and King only)
Probability, P = Number of Favourable Outcome/Total Number of
Outcomes
= 12/52= 3/13.

6.8 Suggested Readings:

1. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for


behavioural sciences (9th Ed.).Cengage Learning India
Private Ltd.
2. https://www.toppr.com/guides/maths/probability/multiplicat
ion-theorem-on-probability/

69
Self-instructional Material
Normal distribution
UNIT 7: NORMAL DISTRIBUTION

NOTES 7.1 Meaning and Importance


7.2 Characteristics of Normal Probability Curve
7.8.1 Probability and Normal Distribution
7.3 Deviations of Normal Probability Curve
7.3.1 Skewness
i. Kurtosis
7.4 Applications of the Normal Probability Curve.
7.5 Let us Sum up
7.6 Unit – End exercises
7.7 Answer to check your progress
7.8 7.8 Suggested Readings

7.1 Meaning and Importance of Normal Probability Curve:


The literal meaning of the term normal is average. The data from a certain
coin or a dice throwing experiment involving a chance success or
probability, if plotted on a graph paper, give a frequency curve which
closely resembles the normal curve. It is because of this reason and
because of its origin from a game of chance the normal curve is called
normal probability curve. The normal curve takes into account the law
which states that the greater a deviation from the mean or average value in
a series the less frequently it occurs. This is satisfactorily used for
describing many distributions which arise in the fields of education,
psychology and sociology. The normal probability curve is a mathematical
distribution.

Researchers often compare the actual distribution of variables they are


studying to the normal curve. They do not expect the distributions of their
variables to match the normal curve (perfectly) but often check whether the
variables approximately follow the normal curve. It is also called a
Gaussian distribution, however its discovery is attributed to Abraham De
Moivre.

7.2 Characteristics of a normal curve:

1. For this curve, mean, median and mode are the same.

2. The curve is perfectly symmetrical. In other words it is not skewed.


The value of measure of skewness computed for this curve is zero.

3. The normal curve serves as a model for describing the peakedness


or flatness of a curve through the measure of kurtosis. For the
normal curve the value of kurtosis is 0.263. If for a distribution the
value of kurtosis is more than 0.263, the distribution is said to be
70
Self-instructional Material
Normal distribution
more flat at the top than the normal curve. But in case the value of
kurtosis is less than 0.263, the distribution is said to be more
peaked than the normal.

4. The curve is asymptotic. It approaches but never touches the


baseline at the extremes because of the possibility of locating in the NOTES
population a case which score still higher than our highest score or
lower than our lowest score. Therefore, theoretically it extends
from minus infinity to plus infinity.

5. As the curve does not touch the baseline, the mean is used as a
starting point for working with the normal curve.

6. The curve has its maximum height or ordinate at the starting point
i.e, the mean of the distribution. In a unit normal curve, the value of
this ordinate is equal to .3989.

7. To find the deviation from the point of departure standard deviation


of the distribution is used as a unit of measurement.

8. The curve extends on both sides – 3 distance on the left to +3


distance on the right.

9. The points of inflection of the curve occur at +1 standard deviation


unit above or below the mean. Thus the curve changes from convex
to concave in relation to the horizontal axis at these points.

10. The total area under the curve extending from -3 to +3 is taken
arbitrarily to be 10,000 because of the greater ease in the
computation of the fractional parts of the total area found for the
mean and the ordinates erected at various distances from the mean.

11. We may find that 3413 cases out of 10,000 or 34.13% of the entire
area of the curve lie between the mean and +1 on the baseline of
the normal curve. Similarly, another 34.13% cases lie between the
mean and -1 on the baseline. Consequently, 68.26% of the area of
the curve falls within the limits +1 standard deviation unit from the
mean. Going further it may be found that 95.44% cases lie from -
2 to +2 and 99.74% cases the from -3 to +3. Consequently
only 26 cases in 10,000 (10000 - 9974) should be expected to the
beyond the range +3 in a large sample. Since it‘s a mathematical
curve these proportions are worked out based on the principles of
probability.

71
Self-instructional Material
Normal distribution

NOTES

12. In this curve, the limits of the distance 1.96 includes 95% and the
limit +2.58 includes 99% of the total area of curve, 5% and 1% of
the area, respectively falling beyond these limits.

7.2.1.Probabilty and normal distribution:


As you have seen in the figure above the normal distribution is
symmetrical, with the highest frequency in the middle and frequencies
tapering off as you move toward either extreme. The exact shape can also
be described by the proportions of area contained in each section of the
distribution. Statisticians often identify sections of a normal distribution by
using z-scores, which you see on the baseline of the curve.
The z score is a standard score that is used to identify and describe
the location of each scores in a distribution. The original unchanged scores
known as raw scores by itself does not provide much information about its
position in the distribution, hence the transformation of X values(the raw
scores) into z scores tell exactly where the original scores are located. The
sign of the z score tells whether the score is located above (+) or below (-)
the mean and the number tells the distance between the score and the mean
in terms of number of standard deviation. In other words, z-scores measure
positions in a distribution in terms of standard deviations from the mean.
(Thus, z =+ 1is 1 standard deviation above the mean, z =+2 is 2 standard
deviations above the mean, and so on.). One major advantage of z score is
that, even when different distribution have different means and standard
deviations, the location identified by the z scores are the same for all
distributions.

72
Self-instructional Material
Normal distribution
The normal curve table is worked by statisticians and it provides
the percentage of scores between the mean and any other Z score. This
provides psychologists more accurate information in many research and
applied situation. The first column in the table tells the z score. The second
column, labeled ―% Mean to z‖ - gives the percentage of scores below the
NOTES
mean and that z score. The third column, labeled ―z in Tail‖ gives the
percentage of scores in the tail for that z score. The table lists only positive
z scores. This is because the normal curve is perfectly symmetrical.

Check your Progress – 1

1. The term normal in the normal distribution means ___________


2. The base line of the normal curve is divided into ___________ units.
3. What does the term ―asymptotic‖ mean?
4. What does the normal curve table provide?
5. Describe the location of the distribution for each of the following z
scores?
a) z= -1.50 b) z= 0.25
6. For a population with µ= 30 and σ= 8, Find the z score of a X= 32

7..3. Deviation from the normal curve:

7.3.1Skewness: The normal curve is symmetrical i.e the left half of the
curve is equal to the right half of it. Also for this curve, mean, median and
mode are the same.

In many distribution which deviate from normal, the value of mean,


median mode are different and there is no symmetry. Lack of symmetry is
referred as skewness. Such distributions are said to be skewed, being
inclined more towards the left or the right to the centre of the curve.

The distribution is said to be skewed negatively when there are


many individual in a group with their scores higher than the average score
of the group. Similarly, the distributions are said to be skewed positively
when there are more individuals in a group who score less than the average
score for their group.

Skewness in any given distribution may be computed by the following


formula:

3  Mean  Median 
Skewness 
SD
3  M  Md 
Sk 
SD

73
Self-instructional Material
Normal distribution
In case when the percentiles are known, the value of skewness may
be computed from the following formula:

P90  P10
Sk   P50
2
NOTES
7.3.2Kurtosis:

When there are very few individuals whose scores are near to the
average score for their group (too few cases in the central area of the
curve) the curve representing such a distribution becomes ‗flattened in the
middle. On the other hand, when there are too many cases in central area,
the distribution curve becomes too ‗peaked in comparison to normal‘. Both
these characteristics of being flat on peaked, are used to describe the term
Kurtosis.

Kurtosis is usually of three types.

Platykurtic: A frequency distribution is said to be platykurtic, when it is


flatter than the normal.

Leptokurtic: A frequency distribution is said to be leptokurtic, when it is


more peaked than the normal.

Mesokurtic: A frequency distribution is said to be mesokurtic, when it


almost resemble the normal curve (neither flattened nor too peaked)

The value of kurtosis for a given curve may be computed through


the following formula.

Quantile deviation
Kurtosis 
90th percentile  10th percentile

In the case of a normal curve, this value is equal to 0.263, the


distribution is said to be platykurtic, if less than 0.263, the distribution is
leptokurtic.

Check your Progress- 2

7. When do we say that the normal curve is skewed?


8. When do we observe a leptokurtic curve?

7.4 Applications of the Normal Curve

Normal curve has wide significance. Some of the main applications


are:-

74
Self-instructional Material
Normal distribution
1. Used as a Model

Normal curve represents a model distribution. It can be used as a


model to

(a) Compare various distributions with it i.e.to say, whether the NOTES
distribution is normal or not and if not, in what way it
deviates from the normal.

(b) Compare two or more distribution is terms of overlapping.

(c) Evaluate students‘ performance from this score.

2. Computing percentiles and percentile Ranks

3. Ability grouping

A group of individuals may be conveniently grouped into certain


categories are A, B, C, D, E in terms of some that with the help of a normal
cure.

4. Transforming and combining qualitative data

Under the assumptions of normality of the distributed variable, the


sets of qualitative data such as ratings, letter grades and categorical ranks
on a scale may be conveniently transformed and combined to provide an
average rating for each individual.

5. Converting Raw score into comparable Standard normalized


scores

With the help of the normal curve, we can convert the raw scores
belonging to different tests into standardized normalized scores like the 
scores and T scores. For converting a given raw scores into a Z score, we
subtract the mean of the scores of the distribution from the respective raw
scores and divide it by the standard deviation of the distribution.

X M
Z

6. Determining relative difficulty of test items

Normal curve provides the simplest rational method of scaling test


items for difficulty and therefore, may be conveniently employed for
determining the relative difficulty of the test questions, problems and other
test items.

75
Self-instructional Material
Normal distribution
Some illustrations of the applications of the normal curve

Case I: Comparing scores on two different tests

A student obtains 80 marks in Maths and 50 in English. If the mean


NOTES and SD for the scores in Maths are 70 and 20 and for the scores in English
are 30 and 10 find out in which subject, Maths or English he did better.

Solution: Direct comparisons of marks obtained on both subjects are not


possible as they do not belong to the same scale of measurement. Therefore
to compare convert raw scores into z scores.

Raw scores in Maths (X1) = 80 M1 = 70 and 1 = 20

X1  M1 80  70
Z score in Maths =   0.5
1 20

Raw scores in English X2 = 50 , M2 = 30, 2 = 10

Therefore

X2  M2 50  30
Z score in English =  2
2 10

We can thus conclude that the student did better in English than in
Maths.

Case II: To determine percentile rank ie percentages of cases lying before


a given score point.

Given a normal distribution N = 1000, mean 80 and SD = 16 find


the percentile equal of the individual scoring 90.

Solution (i) the percentile rank is essentially a rank or position of an


individual (on a scale of 100) decided on the best of the individuals scores.
In other words, we have to determine the percentage of cases lying below
the score point 90. The first step is to convert the raw scores into z score.

X M
z

90  80
 0.625
10

76
Self-instructional Material
Normal distribution

NOTES

From the table of normal curve we can say that 23.41% of cases lie
between M +0.625 distance. But 50% of cases lie up to the mean.
Therefore, it may be concluded that there are 50 + 23.41 = 73.41% of the
individuals whose scores lie below the score point 90 or we may say the
percentiles rank of the individual scoring 90 is 73.

Case III: To determine percentile points or limits in terms of score which


include the lowest given percentage of cases.

Given a normal distribution of N=1000, M = 80 and  = 16,


determine the percentile P30.

77
Self-instructional Material
Normal distribution

NOTES

In determining the percentile P30 we have to look for a score point


on the scale of measurement below which 30% of the case lie. Such a score
with have 20% of the cases lying on the left side of the mean. From the
table of the normal curve, we find out the corresponding  distances from
the mean for 20% of the cases. We find that distance to be 0.525.
Therefore the required score here will be

M - 0.525 or

80 - 0.525  16 = 71.6 or 72

Case IV: To determine the relative difficulty value of the items

Four problems A, B, C and D have been solved by 50%, 60%, 70%


& 80% respectively of a large group. Compare the difference in difficulty
between A &B with the difference between C & D.

78
Self-instructional Material
Normal distribution

NOTES

In the case of a large group, the assumption about normal


distribution of the ability of a group in terms of the achievement on a test
holds good. The percentage of students who are able to solve a particular
problem are counted from extreme right. Therefore, while starting on the
base line of the curve from the extreme eight, up to the point M, we may
cover 50% of the case who can solve the problem. For the rest 60, 70 and
80%. We have to proceed on the left side of the base line of the curve.

For problem A, we see that it has been solved by 50% of the group.
It is also implied that 50% of the group has not been able to solve it.
Therefore, we may say that it was an average problem having of zero
difficulty value.

In the case of problem B, we see that it has been solve by 60% of


the group. It has been a simple problem in comparison to A as 10% more
individuals in the group are able to solve it. For determining the difficulty
value of this problem, we find the  distance from the mean of these 10%
of individuals. From the normal curve table we see that 10% (1000 out of
10,000) cases fall at a sigma distance of 0.253 from the mean after
interpolation. Therefore the difficulty value of problem B will be taken as -
0.253.

Similarly, we may determine the difficulty value of problem C


passed by 70% of the group (20% more individuals of the group than the
average) and problem D by 80% (50% more individuals than the average).

79
Self-instructional Material
Normal distribution
From the table the we know the sigma distance that corresponds to 20% &
30% are 0.525& -0.84. Hence these are the difficulty value of these
items.

Case V: To divide a group into categories according to an ability or trait


NOTES
assumed to be normally distributed.

Example: There is a group of 200 students who has to be classified into


five categories. A, B, C, D & E, according to ability, the range of ability
being equal in each category. If the trait counted under ability is normally
distributed tell how many students should be placed in each category A, B,
C, D& E.

Solution:

As the trait under measurement is normally distributed, the whole


group is divided into five equal categories as shown the above figure. It
shows that the base line of the curve considered to extent from -3 to
+3i.e over a range of 6, may be divided into five equal parts. It gives
1.2 as the portion of the base line to be allotted to each category. Here
group A covers the upper 1.2 segment (falling between 1.8 and 3),
group B the next 1.2, group C lies 0.6 to the right and 0.6 to the left of
the mean. Groups D and E covers the same relative position on the left side

80
Self-instructional Material
Normal distribution
of the mean as covered by the groups B and A on the right side of the
mean.

After the area of the curve covered by the respective categories is


demarcated, the next step is to find out from the normal curve the
percentage of cases lying within each of these area. NOTES

For example area A extends from 1.8 to 3. The percentage of


cases lying between mean and 3 is 49.86% and mean and 1.8 is 46.41%.
The difference (49.86% - 46.41% = 3.45%) will yield the required
percentage of the whole group belonging to category A.

Similarly, group B will cover the cases lying between 0.6 and
1.8. We find from the table that the percentage of the whole group lying
between mean and 1.8 is 46.41 and between M & 0.6 is 22.57 therefore
group B may be said to comprise of 46.41% - 22.57% = 23.84% of the
entire group.

Group C extends from -0.6 to 0.6 on both sides of the mean. The
normal curve table tells us that 22.57% cases to between M & 0.6 and a
similar percentage of cases i.e, 22.57, lies between M2 - 0.6. Therefore,
group C may be said to compute 22.57 2 = 45.14 or 45% of the entire
group.

Group D & E are identical to groups B & A. Therefore, may be


found to consist of the same percentage of cases as covered in groups B
&A respectively.

7.5 Let us Sum up:

The literal meaning of the term normal is average. Though normal


distribution is a mathematical distribution and not the law of nature, we do
find it as a convenient model to explain many physical and psychological
attributes such as intelligence, ability, weight, aptitude etc. When we study
each of the above mentioned attributes we do find that there are a number
of people who deviate from average but only very few who markedly differ
from the average. If we graphically plot the scores that were obtained by
measuring these attributes, we are likely to get a typical curve resembling
the vertical cross section of the bell. This bell shaped curve is called as the
normal curve and has several interesting properties. Probability forms a
direct link between samples and the populations from which they come.
This link is the foundation for the inferential statistics in future chapters.
Deviations from the normal curve are measured in terms of skewness and
Kurtosis. The normal curve as seen in the text above has wide significance
and applicability.

81
Self-instructional Material
Normal distribution
7.6 Unit – End exercises

1. Given the following data regarding two distributions

Mathematics Physics
NOTES M 60 33
S.D 8 9
Achievement scores 70 67
of students

Find out whether the student did better in Maths or Physics

4. Bring out the utility of the normal curve in the field of psychology.

5. Given a normal distribution with the mean of 50 and standard


deviation of 15

a) What percentage of the cases lie between the scores of 47 and 60?
b) What percentage of cases will lie between 40 and 47?
c) What percentage of group is expected to have scores greater than 6?

4. What is a z score?
7.7 Answer to check your progress

1 .Average
2. 6 σ units.
3. The term ―asymptotic‖ in the normal curve refers to the fact that the
curve never touches the baseline due to the possibility of locating in a
population a case which score either higher than the highest scores or
lower than the lowest score.
4. It shows the exact percentages between a score and the mean, which is
worked out by statisticians based probability
5. a) Below the mean 1 ½ standard deviation
b) Above the mean ¼ standard deviation
6. z= 0.25
7. We state that the normal curve is skewed when the tail on one side is
longer than the others. Such asymmetry is observed when there are too
many individuals in either the right or left of the distribution
8. We observe a leptokurtic curve when there are too many cases in central
area, thus making the distribution curve too ‗peaked in comparison to
normal‘.
7.8 Suggested Readings:
1. Mangal S.K. (1987). Statistics in Psychology and Education. New
Delhi: PHI Learning Private Ltd.
2. Haslam,S.M.,&McGarty.C.(2003).Research Methods and statistics
in Psychology.Sage Publications Ltd.
82
Self-instructional Material
Normal distribution
UNIT 8: BINOMIAL DISTRIBUTIONS
8.1 Introduction
8.2 The Binomial Distribution
8.3 The Binomial Test NOTES
8.3.1Hypotheses for the binomial test
8.3.2The data for the binomial test
8.3.3The test statistic for the binomial test
8.4 Let us sum up
8.5 Unit End Exercises
8.6 Answers to check your progress
8.7 Suggested Readings

8.1 Introduction

When a variable is measured on a scale consisting of exactly


two categories, the resulting data are called binomial. The term
binomial can be loosely translated as "two names," referring to the
two categories in the measurement scale.
Binomial data can occur when a variable naturally exists with
only two categories. For example, people can be classified as male or
female, and a coin toss results in either heads or tails. It also is
common for a researcher to simplify data by collapsing the scores
into two categories. For example, a psychologist may use personality
scores to classify people as either high or low in aggression.
In binomial situations, the researcher often knows the
probabilities associated with each of the two categories. With a
1
balanced coin, for example, p(heads) = p(tails)= . The question of
2
interest is the number of limes each category occurs in a series of tri-
als or in a sample of individuals.For example: What is the probability
of obtaining 15 heads in 20 tosses of a balanced coin? What is the
probability of obtaining more than 40 introverts in a sampling of 50
college freshmen?
As we shall see, the normal distribution serves as an excellent
model for computing probabilities with binomial data.

83
Self-instructional Material
Normal distribution
8.2 The Binomial Distribution

To answer probability questions about binomial data, we must


examine the binomial distribution. To define and describe this
NOTES
distribution, we first introduce some notation.

1. The two categories are identified as A and B.


2. The probabilities (or proportions) associated with each
category are identified as
p = p(A) = the probability of A

q = p{B) = the probability of B

Notice that p + q = 1.00 because A and B are the only two


possible outcomes,

3. The number of individuals or observations in the


sample is identified
byn.
4. The variable X refers to the number of times category A
occurs in the sample.
Notice that X clan have any value from 0 (none of the sample
is in category A) to n (all of the sample is in category A).

Using the notation presented here, the binomial distribution


shows the probability associated with each value of X from X = 0 to X
= n.

A simple example of a binomial distribution is presented next.

Figure below shows the binomial distribution for the number


of heads obtained in 2 tosses of a balanced coin. This distribution
shows that it is possible to obtain as many as 2 heads or as few as 0
heads in 2 tosses. The most likely outcome (highest -probability) is
to obtain exactly 1 head in 2 tosses. For this example, the event we
are considering is a coin toss. There are two possible outcomes,
heads and tails. We assume the coin is balanced, so

1
P = p (heads) =
2
84
Self-instructional Material
Normal distribution
Figure The binomial distribution showing the probability for the
number of head sin 2 tosses of a balanced coin

NOTES
0.50

0.25

0 1 2
Number of heads in 2 coin tosses

1
q = p (tails) =
2

We are looking at a sample of n = 2 tosses, and the variable


of interest is

X = the number of heads;

To construct the binomial distribution, we look at all of the


possible outcomes from tossing a coin 2 times. The complete set of 4
outcomes is listed in the following table.

1st Toss 2nd Toss


Heads Heads (Both heads)
Heads Tails (Each sequence has exactly 1 head)
Tails Heads
Tails Tails (No heads)

Notice that there are 4 possible outcomes when you toss a coin
2 times. Only 1 of the 4 outcomes has 2 heads, so the probability of
1
obtaining 2 heads is p = . Similarly, 2 of the 4 outcomes have exactly
4
2 1
1 head, so the probability of 1 head is p = = . Finally, the
4 2
1
probability of no heads (X = 0) is p = .These are the probabilities
4

shown in Figure
85
Self-instructional Material
Normal distribution
Note that this binomial distribution can be used to answer
probability questions. For example, what is the probability of
obtaining at least 1 head in 2 tosses? According to the distribution,
3
NOTES the answer is
4

Check your Progress -1

1. What is a binomial data? When a variable is measured on a scale


consisting of exactly two categories, the resulting data are called binomial

2. What does the binominal distribution give? The binomial distribution


gives the probability for each value of X, where X equals the number of
occurrences of category A in a sample of n events.

8.3 The Binomial Test:

In this chapter, we examine the statistical process of using binomial


data for testing hypotheses about the values of p and q for the
population. This type of hypothesis test is called a binomial test. A
binomial test uses sample data to evaluate hypotheses about the
values of p and q for a population consisting of binomial data.

Consider the following two situations:

1. In a sample of n = 34 colour-blind students, 30 are male, and


only 4 are female. Does this sample indicate that colour
blindness is significantly more common for males in the
general population?

2. In 2005, only 10% of American families had incomes below


the poverty level. This year, in a sample of 100 families, 19
were below the poverty level. Does this sample indicate that
there has been a significant change in the population
proportions?

Notice that both of these examples have binomial data


(exactly two categories). Although the data are relatively simple, we
are asking a statistical question about significance that is appropriate

86
Self-instructional Material
Normal distribution
for a hypothesis test: Do the sample data provide sufficient evidence
to make a conclusion about the population?

8.3.1HYPOTHESES FOR THE BINOMIAL TEST I


NOTES
n the binomial test, the null hypothesis specifies exact values for the
population proportions p and q. Theoretically, you could choose any
proportions for H0, but usually there is a clear reason for the values
that are selected. The null hypothesis typically falls into one of the
following two categories:

1. Just Chance. Often the null hypothesis states that the two
outcomes, A and B, occur in the population with the
proportions that would be predicted simply by chance. If you
were tossing a coin, for example, the null hypothesis might
1 1
specify p(heads) = and p{tails) = Notice that this
2 2

hypothesis states the usual, chance proportions for a balanced


coin. Also notice that it is not necessary to specify both
proportions. Once the value of p is identified, the value of q is
determined by 1 - p. For the coin toss example, the null
hypothesis1 would simply state

1
H0: p = p(heads) = (The coin is balanced.)
2

Similarly, if you were selecting cards from a deck and trying


to predict the suit on each draw, the probability of predicting
1
correctly would be p = for any. given trial. (With four suits, you
4

have a 1-out-of-4 chance of guessing correctly.) In this case, the null


hypothesis would state

1
H0 – P = p (guessing correctly) = (The outcome is
4

simply the result of chance.)

In each case, the null hypothesis states that there is nothing


unusual about the proportions of the population; that is, the outcomes
are occurring by chance.
87
Self-instructional Material
Normal distribution
2. No Change or No Difference: Often you may know the
proportions for one population and want to determine whether
the same proportions apply to a different population. In this
NOTES case, the null hypothesis would simply specify that there is no
difference between the two populations. Suppose that
nationalstatistics indicate that 1 out of 12 drivers will be
involved in a traffic accident during the next year. Does this
same proportion apply to 16-year-olds who are driving for the
first |time? According to the null hypothesis,

1
H0 : For 16-year-olds, p = p(accident) = (Not different
12

from the general population)

Similarly, suppose that last year, 30% of the freshman class


failed the college writing test. This year, the college is requiring all
freshmen to take a writing course. Will the course have any effect on
the number who fail the test? According to the null hypothesis,

H0: For this year, p = p(fail) = 30% (Not different from last year's
class)

8.3.2THE DATA FOR THE BINOMIAL TEST

For the binomial test, a sample of n individuals is obtained and


you simply count how many are classified in category A and how
many are classified in category B. We focus attention on category A
and use the symbol X to stand for the number of individuals classified
in category A. X can have any value from 0 to n and that each value of
X has a specific probability. The distribution of probabilities for each
value of X is called the binomial distribution.

8.3.3 THE TEST STATISTIC FOR THE BINOMIAL TEST

When the values pnand qnare both equal to or greater than 10,
the binomial distribution approximates a normal distribution. This
fact is important because it allows us to compute z-scores and use the
unit normal table to answer probability questions about binomial
88
Self-instructional Material
Normal distribution
events. In particular, when pnand qnare both at least 10, the binomial
distribution has the following properties:

1. The shape of the distribution is approximately normal.


NOTES
2. The mean of the distribution is µ = pn.

3. The standard deviation of the distribution is

 = 𝑛𝑝𝑞

With these parameters in mind, it is possible to compute a z-score


corresponding to each value of X in the binomial distribution.

𝑋−µ 𝑋−𝑝𝑚
z= =
 𝑛𝑝𝑞

This is the basic z-score formula that is used for the binomial test.
However, the formula can be modified slightly to make it more
compatible with the logic of the binomial hypothesis te.st. The
modification consists of dividing both the numerator and the denominator
of the z-score by n. (You should realize that dividing both the numerator
and the denominator by the same value does not change the value of the z-
score.) The resulting equation is

𝑋/𝑛−𝑃
z=
𝑛𝑝𝑞

The logic underlying the binomial test is exactly the same as with
the original z-score hypothesis. The hypothesis test involves comparing the
sample data with the hypothesis. If the data are consistent with the
hypothesis, then we conclude that the hypothesis is reasonable. But if there
is a big discrepancy between the data and the hypothesis, then we reject
the hypothesis. The value of the standard error provides a benchmark for
determining whether the discrepancy between the data and the hypothesis
is more than would be expected by chance. The alpha level for the test
89
Self-instructional Material
Normal distribution
provides a criterion for deciding whether the discrepancy is significant.
The hypothesis-testing procedure is demonstrated in the following
section.
NOTES
STEP 1State the hypotheses. In the binomial test, the null
hypothesis specifies values for the population proportions p
and q. Typically, H(ispecifies a value only for p, the proportion
associated with category A. The value of q is directly
determined from p by the relationship q = I - p. Finally, you
should realize that the hypothesis, as always, addresses the
probabilities or proportions for the population. Although we
use a sample to test the hypothesis, the hypothesis itself
always concerns a population.

[STEP 2Locate the critical region. When both values for pnand
qnarc greater than or equal to 10, then the scores form an
approximately normal distribution. Thus, the unit normal table
can be used to find the boundaries for the critical region. With
a = .05, for example, you may recall that the critical region is
defined as z-score values greater than +1.96 or less than -1.96.

STEP 3Compute the test statistic (z-score). At this time, you obtain a
sample of n individuals (or events) and count the number of
times category A occurs in the sample. The number of
occurrences of A in the sample is the X value for Equation
given below. Because the two z-score equations are
equivalent, you may use either one for the hypothesis test.

STEP 4Make a decision. If the z-score for the sample data is in the
critical region, then you reject H0and conclude that the
discrepancy between the sample proportions and the
hypothesized population proportions is significantly greater
than chance. That is, the data are not consistent with the null
hypothesis, so H0 must be wrong. On the other hand, if the z-
score is not in the critical region, then you fail to reject H0.The
following example demonstrates a complete binomial test.
90
Self-instructional Material
Normal distribution
E X A M P L E : Avisual cliff experiment was designed to examine
depth perception in infants. An infant is placed on a wide board that
appears to have a deep drop on one side and a relatively shallow drop
on the other. An infant who is able to perceive depth should avoid the NOTES
deep side and move toward the shallow side. Without depth
perception, the infant should show no preference between the two
sides. Of the 27 infants in the experiment, 24 stayed exclusively on
the shallow side and only 3 moved onto the deep side. The purpose
of the hypothesis test is to determine whether these data demonstrate
that infants have a significant preference for the shallow side.

This is a binomial hypothesis-testing situation? The two


categories are

A = move onto the deep side

B = move onto the shallow side

STEP1The null hypothesis states that, for the general population of


infants, there is no preference between the deep and the
shallow sides; the direction of movement is determined by
chance. In symbols,

1 1
H0 :
p = p(deep side) = 𝑎𝑛𝑑 𝑞 =
2 2

1
H1 : p ≠ (There is a preference.)
2

We use α = .05

STEP2With a sample of n =27, pn= 13.5 and qn = 13.5. Both values


are greater than 10, so the distribution of z – scores is
approximately normal. With a = .05, the critical region is
determined by boundaries of z = ± 1.96.

STEP3 For this experiment, the data consist of X = 3 out of n = 27.


Using Equation given below, these data produce a z-score
value of

91
Self-instructional Material
Normal distribution
𝑋−𝑝𝑛 3−13.5 − 10.5
z = = = = - 4.04
𝑛𝑝𝑞 27
1 1 2.60
2 2

To use the other Equation , you first compute the sample


NOTES
proportion, X/n = 3/27 = 0.11. The z – score is then

𝑋 / 𝑛−𝑝 0.111−05 − 0.389


z= = = = - 4.05
𝑝𝑞 /n 1 1 0.096
2 2
/27

Within rounding error, the two equations produce the same


result.

STEP 4 Because the data are in the critical region, our decision is to
reject H0. These data do provide sufficient evidence to
conclude that there is a significant preference for the shallow
side. Gibson and Walk (1960) interpreted these data as
convincing evidence that depth perception is innate.

Check your Progress -2

3. For the research study using a visual cliff given above State the
null hypothesis for this study in words and as a probability value
(p) that an infant will crawl off the deep side.
4. If the visual cliff study had used a sample of n = 15 infants,
would it be appropriate to use the normal approximation to the
binomial distribution? Explain why or why not.
5. For a binomial test, the null hypothesis always states that p - 1/2.
(True or false?)
8.4 Let us sum up:

The binomial test is used with dichotomous data—that is, when each
individual in the sample is classified inone of two categories. The two
categories are identified as A and B. The binomial test uses sample data to
test hypotheses about the binomial proportions, p and q, for a population.
The null hypothesis specifies p and q, and the binomial distribution (or the
normal approximation) is used to determine the critical region.
8.5 Unit End Exercises
1. To investigate the phenomenon of ―home team advantage,‖ a
researcher recorded the outcomes from 64 college football games
on one Saturday in October. Of the 64 games, 42 were won by
home teams. Does this result provide enough evidence to conclude

92
Self-instructional Material
Normal distribution
that home teams win significantly more than would be expected by
chance? Use a two-tailed test with α= .05.

2. A researcher would like to determine whether people really can tell


the difference between bottled water and tap water. Participants are
NOTES
asked to taste two unlabelled glasses of water, one bottled and one
tap, and identify the one they thought tasted better. Out of 40
people in the sample, 28 picked the bottled water. Was the bottled
water selected significantly more often than would be expected by
chance? Use a two-tailed test with α= .05.

8.6 Answers to check your progress


1. When a variable is measured on a scale consisting of exactly two
categories, the resulting data are called binomial

2. The binomial distribution gives the probability for each value of X,


where X equals the number of occurrences of category A in a sample of n
events.

3.The null hypothesis states that the probability of choosing between the
deep side and the shallow side is just chance: p(deep side) = ½.

4. The normal approximation to the binomial distribution requires that both


pnand qnare atleast 10. With n = 15, pn= qn= 7.5. The normal
approximation should not be used.

5.False.

8.7Suggested Readings

1. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural sciences


(9th Ed.).Cengage Learning India Private Ltd.

93
Self-instructional Material
Analysis of variance
UNIT 9: ANALYSIS OF VARIANCE
9.1 Purpose and Logic of Analysis of Variance
9.2 Assumptions of Analysis of Variance.
NOTES
9.3 One Way Analysis of Variance.
9.3.1 Illustration
9.4 Two Way Analysis of Variance
9.4.1 Illustration
9.5 Let us Sum up
9.6 Unit – End exercises
9.7 Answer to check your progress
9.8 Suggested Readings

9.1 PURPOSE FOR ANALYSIS OF VARIANCE (ANOVA) :

Analysis of Variance is carried out when we want to determine the


significances of difference between more than two group means. In
ANOVA, the variable (independent or quasi-independent) that designates
the groups being compared is called a factor. In addition, the individual
groups or treatment conditions that are used to make up a factor are called
the levels of the factor. The technique of ANOVA as a single composite
test of significance, for the difference between several group means
demands derivation of two independent estimates of the population
variance, one based on the variance of group means and other variance
within groups.

THE LOGIC OF ONE-WAY ANALYSIS OF VARIANCE


Variation within and between groups: Suppose if we have selected three
samples of 10 cases each at random and have given them three different
treatment conditions. If we wish to test our null hypothesis for possible
difference in treatment conditions, we have to deal with two types of
variation.

Within group Variation: Within each sample the individual vary about
the sample mean. We call this within group variation. It is a direct
reflection of the inherent variation among individual given the same
treatment. We can present exactly the same stimulus to everyone in a group
and still observe variation in reaction times.

Between Group Variations: The means of the samples vary among


themselves. It is important to realize that between groups variation is also a
reflection of the inherent variation among individuals. The greater the
inherent variation among individuals, the greater the opportunity for
chance to produce sample means that vary from one another.

94
Self-instructional Material
Analysis of variance
Now the question is what the variation in scores within and among
groups has to do with testing hypothesis about means? The answer is that
these two kinds of variation provides the basis for testing the null
NOTES
hypothesis.

For example, In a study, there were to forty boys in all belonging to


four different locations. It we add the 1Q score of these 40 boys and divide
the sum by 40, we get the value of the grand or general mean which is the
best estimate of the population mean. There are 4 groups of 10 boys each.
The mean and IQ scores of 10 boys in each groups is called the group
mean and in this way, there will the 4 group means, which will vary
considerably from each other.

Now the question arises as to how far does an IQ score of a


particular boy belonging to a particular sample deviate from the grand
mean of 40 boys. We may observe that the deviation of the score from the
mean of that particular group and the deviation of the mean of the group
from the grand mean. For deriving useful results we can use variance as a
measure of dispersion in place of standard deviation. As variance of an
individual‘s score from the grand mean may be broken into two part i.e.
within group variance and between group variance

If we wanted to compare the variance of the two variable one


obvious way is to form a ratio. To do this all we need to do is to divide the
variance of one variable by the variance of the other. It is convenient to
divide the larger variance by a smaller one. A ratio of this form gives us
the information about whether one variance is larger than another and by
what amount. If the difference is big we would also like to know if it is big
enough to be meaningfully different.

The related question in this case is: how plausible is it that the
variances we have observed are really the estimates of the same population
variance? i.e can they be treated as if they were two different sample
drawn in the same way from the same population . If the variances are both
estimates of the same population variance then they should be pretty
similar and any difference between them will be due to chance variations.
In order to examine this issue further we need to know how this ratio based
statistic behave. In the case of computing variances, we need to know
about the sampling distribution of the ratio of variance. As it happens, the
behaviour of the sampling distribution is quite well understood and
providing that certain assumptions hold, it follows a particular pattern. This
pattern was called
F- ratio denoted by

between group variance


F
within group variance

95 Self-instructional Material
Analysis of variance is used as a critical ratio for determining the significance of difference
between group means at a given level of significance. If we assume that
these two variances are drawn from the same population after calculating F
ratio, what we now need to do is check what proportions of the F
NOTES distribution is cut off by this value. The critical value in the F table will
suggest whether or not the obtained ratio is particularly unusual one to be
obtained if the samples really were both drawn from the same populations
with the same variance.

9.2 ASSUMPTIONS IN ANALYSIS OF VARIANCE

The following are the fundamental assumption for the use of analysis
of variance technique:
1. The dependent variable which is measured should be normally
distributed in the population.
2. The individuals being observed should be distributed randomly in
the groups
3. Within-groups variances must be approximately equal.
4. The contributions to variance in the total sample must be additive.

Check your Progress – 1

1.ANOVA is a statistical procedure that compares two or more treatment


conditions for differences in variance. (True or false?)

2.In ANOVA, the total variability is partitioned into two parts. What are
these two variability components called, and how are they used in the F-
ratio? The two components are between-treatments variability and within-
treatments variability. Between-treatments variance is the numerator of the
F-ratio, and within-treatments varianceis the denominator.

3. What happens to the value of the F-ratio if differences between


treatments are increased? What happens to the F-ratio if variability inside
the treatments is increased? As differences between treatments increase,
the F-ratio increases. As variability within treatments increases, the F-ratio
decreases.

9.3 ONE WAY ANALYSIS OF VARIANCE

The Null Hypothesis: One way analysis of variance allows us to compare


the means of two or more groups simultaneously. It is closely related to‗t‘
test. In one-way analysis of variance, there may be two or more treatment
conditions, often referred to as different levels of the independent variable.
The Null Hypothesis in ANOVA states that if the different treatments have
no different effect on the variable under observation, then we expect the
96
Self-instructional Material
Analysis of variance
population means to be equal. To inquire whether variation in treatment
conditions makes a difference, we test the Null hypothesis. The Null
hypothesis in ANOVA is often referred to as an omnibus hypothesis (i.e.
NOTES
covering many situations at once) and ANOVA itself as an omnibus test.

As already pointed out, the deviation of an individual‘s score


belonging to a sample or a group of population from the grand mean can be
divided into two parts: (i) deviation of the individual‘s score from his
group mean; and (ii) deviation of the group mean from the grand mean.
Consequently, the total variance of the scores of all individuals included in
the study may be partitioned into within-group variance and between-
groups variance. The formula used for the computation of variance (2) is
 x2/N, i.e., sum of the squared deviation from the mean value divided by
total frequencies. Hence, by taking N as the common denominator, the
total sum of the squared deviation of scores around the grand or general
mean for all groups combined can be made equal to the sum of the two
partitioned, between-groups and within-groups sum of squares.
Mathematically,

Total sum of squares (around the general mean) = between-groups sum of


squares + within-groups sum of squares

or

St2  Sb2  Sw2

Hence the procedure for the analysis of variance involves the following
main tasks:

(i) Computation of total sum of squares  St2 


(ii) Computation of between-groups sum of squares  Sb2 
(iiii) Computation of within-groups sum of squares  Sw2 
(iv) Computation of F-ratio.
(v) Use of t test (if the need for further testing arises).

All these tasks may be carried out in a series of systematic steps.


Let us try to understand these steps by adopting the following terminology:

X = Raw score of any individual included in the study


(any score entry in the given table)

X = Grand Sum

X
= Grand or general mean
N

X1, X2, … denote scores within first group, second group, …

97 Self-instructional Material
Analysis of variance n1, n2, n3= No. of individuals in first, second and third groups

X 1 X 2
, ,... denote means of the first group, second group, …
N N
NOTES N = total No. of scores or frequencies

Let us now outline the steps.

Step 1.Arrangement of the given table and computation of some initial


values. In this step, the following values needed in computation are
calculated from the experimental data arranged in proper tabular form:

(i) Sum of squares,  X1, X2, … and the grand sum,  X

X 1 X 2 X
, ,...
(ii) Group means, N N , and the grand mean N

(iii) Correction term C computed by the formula

(X )2 Square of the grand sum


C 
N Total No. of cases

Step 2.Arrangement of the given table into squared-form table and


calculation of some other values. The given table is transformed into a
squared-form table by squaring the values of each score given in the
original table and then the following values are computed;

(i)
X12 , X 22 , X 32 ,...

(ii) X 2

Step 3.Calculation of total sum of squares. The total sum of squares around
the general mean is calculated with the help of the following formula:

St2  X 2  C

(X )2
 X  2

N
Step 4.Calculation of between-group sum of squares. The value of the
between-groups sum of squares may be computed with the help of the
following formula:

(X1 )2 (X 2 )2 (X 3 )2 (X n )2


S 
2
b    ...  C
n1 n2 n3 nn

Step 5.Calculation of within-group sum of squares. Between


groups and within - groups sum of squares constitute the total sum of

98
Self-instructional Material
Analysis of variance
squares. Therefore, after steps 3 and 4, the value of within-groups sum of
squares (sum of squares of deviation within each group about their
respective groups means) may be calculated by subtracting the value of
NOTES
between-groups sum of squares from the total sum of squares. Its formula,
therefore, goes thus :

Sw2  Sl2  Sb2

Step 6.Calculation of the number of degrees of freedom. All these sums of


squares calculated in step 3-5 possess different degrees of freedom given
by

Total sum of squares ,


(Sl2 )  N 1

Between - groups sum of squares, ( Sb )  K  1


2

Within groups sum of squares, (Sw )  ( N 1)  ( K  1)  N  K


2

where N represents the total number of observations , scores or


frequencies and K, the number of groups in the research study.

Step 7.Calculations of F-ratio. The value of F-ratio furnishes a


comprehensive or overall test of significance of the difference between
means. For its computation, we have to arrange the data and computation
work in the following manner.

Source of Means square


Sum of squares df
variance variance
Between - groups S 2 K-1 Sb2
b (computed in step 4)
K 1
Within - groups Sw2 (computed in step 5) N-1 Sw2
N K

Mean square variancebetween  groups


F
Mean square variance within  groups

Step 8: Interpretation of F-ratio: F- ratio are interpreted by the use of the


critical value of F - ratios. This table has the number of degrees of freedom
for the greater mean square variance across the top and the number of
degrees of freedom in the smaller mean square variance on the left - hand
side. If our computed value of F is equal to or greater than the critical
tabled value of F at a given level of significance 0.05 or 0.01, it is assumed
to be significant and consequently we reject the null hypothesis of no
99 Self-instructional Material
Analysis of variance difference among these means at that level of significance. However, a
significant F does not tell us which of the group means differ significantly;
it merely tells us that at least one of the means is relatively different from
some other. Consequently, there arises a need for further testing to
NOTES determine which of the difference between means are significant.

In case our computed value of F is less than the critical tabled value
of F at a given level of significance, it is taken as non - significant and
consequently, the null hypothesis cannot be rejected. Then there is no
reason for further testing (as none of the difference between means will be
significant). In a summarized form, the above analysis may be represented
as follows :

F Significant  Null hypothesis rejected  Need for further


testing

FNon-significantNull hypothesis rejectedNo need for further testing

Generally as and when we get the value of F as less than 1, we


straightaway interpret it as non - significant resulting in the non - rejection
of the null hypothesis.

Step 9.Testing differences between means with the t test. When F is found
significant, the need for further testing arises. We take pairs of the group
means one by one for testing the significance of differences. The t test
provides an adequate procedure for testing the significance when we have
means of only two samples or group at a time for consideration. Therefore,
we make use of the t test to test the difference between pairs of means.

As we have seen in Chapter 10, the usual formula for computation


of t value is

D Differencebetween two means


t 
 D S tandard error of the difference betweenthe means

and  D is computed by the formula

1 1
D    
 n1 n2 

Where

 = Pooled SD of the samples drawn from the same


population.

n1, n2 = Total No. of cases in samples 1 and II, respectively.

100
Self-instructional Material
Analysis of variance
In the analysis of variance technique, within groups means square
variance provides us the value of 2, the square root of which can give us
the required pooled SD of the samples or groups included in our study. NOTES
The degrees of freedom for within groups sum of squares are given
by the formula N -K. With these degrees of freedom we can read the t
values, at 0.05 and 0.01 levels of significance. If the computed value of t is
found to be equal to or greater than the critical value of t at 0.05 or 0.01
levels, we can reject the null hypothesis at that level of significance.

9.3.1 Illustration: The aim of an experimental study was to determine the


effect of three different technique of training on the learning of a particular
skill. Three groups, each consisting of seven students of classes IX,
assigned randomly, were given training through these different techniques.
The scores obtained on a performance test were recorded as follows:

Group I Group II Group III


3 4 5
5 5 5
3 3 5
1 4 1
7 9 7
3 5 3
6 5 7
Test the difference between groups by adopting the analysis of variance
technique.

Solutions

Step 1. Original table computation

Organization of Data

Rating of the coaching experts


Group III Total
Group I (X1) Group II(X2)
(X3)
3 4 5 12
5 5 5 15
3 3 5 11
1 4 1 6
7 9 7 23
3 5 3 11
6 5 7 18
X1=28 X2=35 X3=33 X=96
Grand Total

101 Self-instructional Material


Analysis of variance n1=n2 = n3 =7, N = n1 + n2 +n3 = 21

1 28
 4
Group means = n1 7
NOTES
 2 35
 5
n2 7

3 33
  4.71
n3 7

()2 96 x96 9216


   438.85
Correction term C = N 21 21

Step 2: Squared - Table computation

X12 X 22 X 32 Total
9 16 25 50
25 25 25 75
9 9 25 43
1 16 1 18
49 81 49 179
0 25 9 43
36 25 49 110
 X12 =138  X 22 =197  X 32 =193  X 2 =518
2
Step: 3 Total sum of squares ( St ) :

(X )2
S   
t
2 2
 X 2  C  518  438.85  79.15
N
2
Step 4. Between - group sum of squares ( Sb ) :

(X1 )2 (X 2 )2 (X 3 )2


Sb2    C
n1 n2 n3

28  28 35  35 33  33
    438.85
7 7 7

784  1225  1089


  438.85
7

 442.57  438.85  3.72

2
Step 5: Within - group sum of squares ( Sw ). This is obtained as

102
Self-instructional Material
Analysis of variance
S  S  S  79.15  3.72  75.43
2
w
2
1
2
b

Step 6. Number of degrees of freedom NOTES


For total sum of squares
(S12 )  N  1  21  1  20
For between - groups sum of squares.
(Sb2 )  K 1  3 1  2
For within - groups sum of squares
(Sw2 )  N  K  21  3  18

Step 7. Calculation of F – ratio

Source of Sum of Mean square


df
variation squares variance
Between - groups S 2 2 3.72/2 = 1.86
b = 3.72

Within – groups Sw2 =75.43 18 75.43/18 = 4.19

Mean square variancebetween  group 1.86


F   0.444
Mean square variance within  group 4.19

Step 8.Interpretation of F-ratio. The F- ratio table is referred to for 2


degrees of freedom for smaller mean square variance on the left hand side,
and for 18 degrees of freedom for greater mean square variance across the
top. The critical values of F obtained by interpolation are as follows:

Critical value of F = 19.43 at 0.05 level of significance


Critical value of F = 99.44 at 0.01 level of significance
Our computed value of F (.444) is not significant at both the levels
of significance and hence, the null hypothesis cannot be rejected and we
may confidently say that the differences between means are not significant
and therefore, there is no need for further testing with the help of t test.

Check your Progress 2

4.A researcher uses an ANOVA to compare three treatment conditions


with a sample of n = 8 in each treatment. For this analysis, find dftotal,
dfbetween, and dfwithin.

5. A researcher conducts an experiment comparing four treatment


conditions with a separate sample of n = 6 in each treatment. An ANOVA

103 Self-instructional Material


Analysis of variance is used to evaluate the data, and the results of the ANOVA are presented in
the following table. Complete all missing values in the table. Hint: Begin
with the values in the dfcolumn.

NOTES Source Sum of MS df F Ratio


squares
Between - - - -
treatments
Within - 2 -
treatments
Total 58

9.4 TWO-WAY ANALYSIS OF VARIANCE

So far, we have dealt with one-way analysis of variance involving


one experimental variable. However, experiments may be conducted in the
fields of education and psychology for the simultaneous study of two
experimental variables. Such experiments involve two-way classification
based on two experimental variables. Let us make some distinction
between the need for carrying out one-way and two-way analyses of
variance through some illustrations.

Suppose that we want to study the effect of four different methods


of teaching. Here, the method of teaching is the experimental variable
(independent variable which is to be applied at four levels). We take four
groups of students randomly selected from a class. These four groups are
taught by the same teacher in the same school but by different methods. At
the end of the session, all the groups are tested through an achievement test
by the teacher. The mean scores of these four groups are computed. If we
are interested in knowing the significance of the differences between the
means of these groups, the best technique is the analysis of variance. Since
only one experimental variable (effect of the method of teaching) is to be
studied, we have to carry out one-way analysis of variance.

Let us assume, that there is one more experimental or independent


variable in the study, in addition to the method of teaching. Let it be the
school system at three levels which means that three school systems are
chosen for the experiment. These systems can be: government school,
government-aided school and public school. Now the experiment will
involve the study of 4 x 3 groups. We have 4 groups each in the three types
of schools (all randomly selected). The achievement scores of these groups
can then be compared by the method of analysis of variance by
establishing a null hypothesis that neither the school system nor the
methods have anything to do with the achievement of pupils. In this way,

104
Self-instructional Material
Analysis of variance
we have to simultaneously study the impact of two experimental variables,
each having two or more levels, characteristics or classifications and hence
we have to carry out the two-way analysis of variance.
NOTES
In the two-way classification situation, an estimate of population
variance, i.e. total variance is supposed to be broken into: (i) variance due
to methods alone, (ii) variance due to school alone, and (iii) the residual
variance in the groups called interaction variance (M x S; M=methods, S =
schools) which may exist on account of the following factors:

1. Chance

2. Uncontrolled variables like lack of uniformity in teachers

3. Relative merits of the methods (which may differ from one


school to another).

In other words, there may be interaction between methods and schools


which means that although no method may be regarded as good or bad in
general, yet it may be more suitable (or unsuitable) for a particular school
system. Hence, the presence of interaction variance is a unique feature with
all the experimental studies involving two or more experimental variables.
In such problems, requiring two or more ways of analysis of variance, we
will have to take care of the residual or interaction variance in estimating
their population variance. If the null hypothesis is true, variance in terms of
the method should not be significantly different from the interaction
variance. Similarly, the variance due to schools may also be compared with
the interaction variance. For all such purposes of comparison, the F-ratio
test is used.

The design of the above two-way classification experiment can be


further elaborated by having more experimental or independent variables,
each considered at several levels. As the number of variables increases, the
order of interaction also increases. According to the needs and
requirements of the experiment, a research worker has to select a particular
experimental design involving two or more independent or experimental
variables.

9.4.2 Illustration :In a research study, there were two experimental or


independent variable: a seven member group of player and three coaches
who were asked to rate the players in terms of a particular trait on a ten -
point scale. The data were recorded as under:

Players
Rating by three coaches
1 2 3 4 5 6 7
A 3 5 3 1 7 3 6
B 4 5 3 4 9 5 5

105 Self-instructional Material


Analysis of variance C 5 5 5 1 7 3 7

Apply the technique of analysis of variance for analysing these


NOTES data.

Step 1 : Arrangement of the data in a proper table ad computation of


essential initial value

Rating of coaches Square


Total of of the
Players By A By B By C
rows total of
(X1) (X2) (X3) rows
1 3 4 5 12 144
2 5 5 5 15 225
3 3 3 5 11 121
4 1 4 1 6 36
5 7 9 7 23 529
6 3 5 3 11 121
7 6 5 7 18 324
Total of X1 = 28 X2 = 35 X3 = 33 X = 96 (Grand
columns Total)
Square of (X1)2 = 784 (X2)2 = (X3)2 = (X)2 =
the totals 1225 1089 9216
of the
column

Here,

N  n1  n2  n3  7  7  7  21

(X )2 9216
C   438.85
N 21
Step 2: Arrangement of the table in square form and computation of some
essential values.

X12 X 22 X 32 Total

9 16 25 50
25 25 25 75
9 9 25 43
1 16 1 18
49 81 49 179

106
Self-instructional Material
Analysis of variance
9 25 9 43
36 25 49 110

X12  138 X 22  197 X 32  183 X 2  518 NOTES

2
Step 3.Calculation of the total sum of squares ( S1 ) (around grand mean).

(X )2
S  X 
t
2 2
 X 2  C
N
= 518 - 438.85 = 79.15
2
Step 4 : Calculation of the sum of squares for rows ( Sr ) (between the
means of players to be rated by three coaches).

144  225  121  36  529  121  324


Sr2  C
3

1500
  438.85  61.15
3

2
Step 5 : Calculation of the sum of squares for columns ( St ) (between the
means of coaches rating 7 players).

784 1225 1089


Sc2  C
7

= 442.57 - 438.85 = 3.72


2
Step 6: Calculation of the interaction or residual sum of squares ( St ) . The
interaction sum of squares is given by

St2  St2  (Sr2  Sc2 )

= 79.15 - (61.15 + 3.72) = 14.28

Step 7. Computation of F - ratios.

Mean
Source of Sum of
Df square
variation squares
variance
Rows (r-1)=6 61.15 61.5
 10.19
(players) 6
Columns (c-1)=2 3.72 3.72
 1.86
(coaches) 2

107 Self-instructional Material


Analysis of variance
Interaction or (r-1) (c-1) 14.28 14.28
 1.19
residual =12 12

NOTES Mean square variancebetween rows ( players )


F ( for rows) 
Mean square variance interms of interaction

10.19
  8.56
1.19

Mean square variancebetween columns (coaches )


F ( for columns) 
Mean square variance interms of interaction

1.86
  1.56
1.19

Step 8.Interpretation of F-ratios.

df for df for Critical


Judgment
greate smalle values of F about the
r r
Kind of significance Conclusi
mean means At
F at 0.05 of on
square square 0.01
level computed
varian varian level F
ce ce
F (for 2 12 3.88 6.93 Significant Null
rows) at both hypothes
levels is
replaced
F (for 6 12 3.00 4.82 Not Null
columns significant hypothes
) at both is not
levels rejected
Hence, F (for rows) is highly significant and hence null hypothesis
is rejected. It indicates that the coaches did discriminate among the players.
The second F (for columns) is insignificant and hence, the null hypothesis
cannot be rejected. It indicates that the coaches did not differ significantly
among themselves in their ratings of the players. In other words, their
ratings may be taken as trustworthy and reliable.

Check your Progress 3

6. When is Two –way analysis of variance a preferred statistical measure?

9.5 Lets sum up:


Analysis of variance (ANOVA) is a statistical technique that is used to test
the significance of mean difference among two or more treatment
108
Self-instructional Material
Analysis of variance
conditions. The null hypothesis for this test states that, in the general
population, there are no mean differences among the treatments. The
alternative states that at least one mean is different from another .The test
NOTES
statistic for ANOVA is a ratio of two variances called an F-ratio. The
variances in the F-ratio are called mean squares, or MS values. Each MS is
computed by Sum of Squares (SS)÷df. For the independent-measures
ANOVA, the F-ratio is MS between÷MS within

9.6 Unit End Exercises:


1. Explain the term factor and the levels of a factor with an example
2. Explain the logic of ANOVA
3. There is some research indicating that college students who use
Facebook while studying tend to have lower grades than non-users. A
representative study surveys students to determine the amount of Facebook
use during the time they are studying or doing homework. Based on the
amount of time spent on Facebook, students are classified into three groups
and their grade point averages are recorded. The following data show
typical pattern of results

Non – User 4 2 5 1 3 7 4 4 8 2
Rarely Use 5 7 3 4 8 6 2 7 3 5
Regularly 5 7 6 6 8 9 6 4 6 8
Use
Use ANOVA to determine whether there are significant mean differences
among the three groups.

9.7 Answers to check your progress:


1.False. Although ANOVA uses variance in the computations, the purpose
of the test is to evaluate differences in means between treatments.

2.The two components are between-treatments variability and within-


treatments variability. Between-treatments variance is the numerator of the
F-ratio, and within-treatments variance is the denominator.

3.As differences between treatments increase, the F-ratio increases. As


variability within treatments increases, the F-ratio decreases.

4. Ans :dftotal = 23; dfbetween = 2; dfwithin = 21.


5.
Source Sum of MS df F Ratio
squares
Between 18 6 3 3.00
treatments
Within 40 2 20
treatments
Total 58 23
109 Self-instructional Material
Analysis of variance

6.Two way analysis of variance is a preferred statistical measure when we


wish to examine the influence of two different categorical independent
variables on one continuous dependent variable. The two-way ANOVA not
NOTES only aims at assessing the main effect of each independent variable but
also if there is any interaction between them.

9.8 Suggested Readings


3. Mangal S.K. (1987). Statistics in Psychology and
Education. New Delhi: PHI Learning Private Ltd.
4. Gravetter,F,J., &Wallnau,L.B.(1995). Essentials of
nd
statistics. (2 Ed.).West Publishing Company.
5. King,B.M., &Bear.G.(2001). Statistical Reasoning in
Psychology and Education(3rd Ed.). John Wiley& Sons Inc.

110
Self-instructional Material
Non- parametric tests
UNIT 10: NON- PARAMETRIC TESTS
10.1.Nature and Assumptions
10.1.1 Advantages of Parametric test NOTES
10.1.1 Disadvantages of Parametric test
10.2 Chi-square
10.2.1 Assumptions of the Chi-square test
10.2.2 Null hypothesis for chisquare
10.2.3 Steps to compute Chi-square.
10.2.4 Uses of chi-square test
10.3 Contingency Co-efficient
10.4 Median test
10.4.1 Steps in the computation of the median test
10.5 Sign test
10.6 Friedman test.
10.7 Let’s sum up
10.8 Unit End Exercises
10.9 Answers to check your progress
10.10 Suggested Readings

10.1 Nature and Assumptions

A non-parametric statistical test is one which does not specify any


conditions about the parameter of the population from which the sample is
drawn. Since these statistical tests do not make any specified and precise
assumption about the form of the distribution of the population, these are
also known as distribution free statistics.

For a non - parametric statistical test, the variables under study


should be continuous and the observation should be independent. But these
condition are not too rigid & elaborate as in parametric statistics.A non -
parametric tests should be used only in the following case.

1. The shape of the distribution of the population from which a sample


is drawn is not known to be a normal one.

2. The variables have been quantified on the basis of nominal


measures (or frequency counts).

3. The variables have been quantified on the basis of ordinal measures


(or ranking)

As non-parametric statistical tests are based upon frequency counts


or rankings rather than on the measured values, they are less precise, and
are less likely to reject null hypothesis when it is fake. That is why, a non-
parametrical statistical test is used only when the parametric assumptions
cannot be met.

111 Self-instructional Material


Non- parametric tests Bradly (1968) has enumerated several advantages and
disadvantages of non-parametric statistical tests in comparison to
parametrical tests.
NOTES
10.1.1 Advantages of non-parametric tests.

1. Simplicity and facilitation in derivation: Most of the non-


parametric statistics can be derived by using simple computational
formula.

2. Wide scope of application: Since non-parametric states are based


upon fewer and less rigid and elaborate assumptions regarding the
form of the population distributions they can be easily applied to
much wide situations.

3. Susceptibility to violation of assumption: Since the assumptions


in case of non-parametric statistics are fewer and less elaborate,
they are less susceptible to violations.

4. Speed of application: When the size of the sample is small,


calculation of non-parametric statistics is faster than parametric
statistics.

5. Type of measurement required: Non parametric statistics require


measurement based upon a nominal scale and ordinal scale.

6. Impact of sample size: When sample size is 10 or less than 10,


non-parametric statistics are easier, quicker and more efficient than
parametric statistics. However, as the sample size increases, non-
parametric statistics become time consuming, laborious and less
efficient than parametric statistics.

7. Statistical efficiency: Non-parametric tests are often more


convenient than the parametric tests. If the data is such that it meets
all assumption of non-parametric statistics but not of parametric
statistics, then non-parametric statistics have statistical efficiency
equal to parametric statistics. If both parametric and non-parametric
statistics are applied to the data, which fulfills all assumptions of
parametric tests, the distribution - free statistics become more
efficient with a small sample size but they become less and less
efficient as sample size increases.

10.1.2 Disadvantages of non-parametric tests

1. According to Moses (1952), the non-parametric statistics have


lower statistical efficiency than parametric statistics when sample
size is large, preferably above 30.

Self-instructional Material 112


Non- parametric tests
2. If all the assumptions of parametric statistics are fulfilled, Siegel
(1956), and Siegel & Castellan (1988) consider the use of non-
parametric statistics as simply ‗wasteful of data‘.

Check your Progress – 1 NOTES

1. Non parametric tests make assumptions about the population


distribution-True/False

2. Data measured on a ratio scale is suitable for non – parametric tests –


True/False

10.2 Chi-square test

The chi-square is one of the most important non-parametric


statistics, which is used for several purposes. For this reason, Guilford
(1956) has called it general purpose statistics. The chi-squares applies only
to discrete data. However, any continuous data can be reduced to the
categories in such a way that they can be treated as discrete data and then
the application of chi-square is justified. The formula for computing 2 is
given below

( f0  f0 )2
 2

fe

2 = chi-square
f0 = obtained or observed frequency
fc = expected theoretical frequency

There are two types of chi-square test. The chi-square test for
goodness of fit involves testing the levels of a single nominal variable and
the chi-square test of independence is used when there are two nominal
variables, each with several categories.

10.2.1 Assumptions of chi-square statistic

Prior to using the chi square test there are certain requirements that must be
met. These are :-

1. The data must be in the form of frequencies counted in each of a set


of categories.
2. The total numbers observed must exceed 20.
3. The expected frequency under the H0 hypothesis in any one fraction
must not normally be less than 5.
4. All the observations must be independent of each other. In other
words, one observation must not have an influence upon another
observation.

113 Self-instructional Material


Non- parametric tests 10.2.2.Null Hypothesis

Null hypothesis for Chi square goodness of fit :Generally H0 falls into
NOTES one of the following categories:

1. No Preference, Equal Proportions. The null hypothesis often states


that there is no preference among the different categories. In this case, H0
states that the population is divided equally among the categories. For
example, a hypothesis stating that there is no preference among the three
leading brands of soft drinks, would mean that if 60 respondents were
surveyed the expected frequencies will be set 20 for each category

2.No Difference from a Known Population. The null hypothesis can state
that the proportions for one population are not different from the
proportions that are known to exist for another population. For example,
suppose it is known that 28% of the licensed drivers in the state are
younger than 30 years old and 72% are 30 or older. A researcher might
wonder whether this same proportion holds for the distribution of speeding
tickets. The null hypothesis would state that tickets are handed out equally
across the population of drivers, so there is no difference between the age
distribution for drivers and the age distribution for speeding tickets.

Null Hypothesis for Chi-Square test of Independence: The chi-square


test for independenceuses the frequency data from a sample to evaluate the
relationship between two variables in the population. Each individual in the
sample is classified on both of the two variables, creating a two-
dimensional frequency-distribution matrix. The frequency distribution for
the sample is then used to test hypotheses about the corresponding
frequency distribution for the population.

10.2.3 Steps for computing the chi-square statistic:


1. Determine the actual, observed frequency is in each category. The
observed frequency is likely to be a whole number as it is obtained by
counting the number of people in the sample.
2. Determine the expected frequency in each category. (This can be in
decimals). In case of hypothesis of equal probability, the expected
frequencies are obtained by dividing N by the total number of categories.
3. In each category, take observed minus expected frequencies.
4. Square each of the differences.
5. Divide each squared difference by the expected frequency for its
category.
6. Add up the results of step 5 for all the categories.

The numerator of the chi-square statistics is obtained by measuring the


difference between the data (the fovalues) and the hypothesis (represented
Self-instructional Material 114
Non- parametric tests
by the fevalues) and squaring them. The expected frequencies for the
goodness-of-fit test are determined by expected frequency = fe=pnwhere p
is the hypothesized proportion (according to H0) of observations falling
into a category and n is the size of the sample. At the end, we add the NOTES
values to obtain the total discrepancy between the data and the hypothesis.
Thus, a large value for chi-square indicates that the data do not fit the
hypothesis, and leads us to reject the null hypothesis.
In order to determine the significance of Chi-square degrees of
freedom are computed. For Chi-square test of goodness of fit ,df is
obtained by the formula C-1 where C is the number of categories in the
variable. Degrees of freedom for the test for independence are computed
by df= (R – 1)(C – 1) where R is the number of row categories and C is the
number of column categories.

10.2.4. Uses of chi-square.

1. The chi-square may be used as a test of equal probability


hypothesis. Suppose for example, 100 students answer an item in
an attitude scale the item has five response options. According to
equal probability hypothesis, the expected frequency of responses
given by 100 students would be 20 in each. The chi-square test
would test whether or not the equal probability hypothesis is
tenable.

2. The second use of the chi-square test is in testing the significance


of the independence hypothesis. The chi-square is a measure of the
degree of relationship in such a situation.

3. The third important use of chi-square is in testing a hypothesis


regarding the normal shape of a frequency distribution. When chi-
square is used in this connection, it is commonly referred to as a
test of goodness of fit.

4. The fourth use of chi-square is in testing the significance of several


statistics. In testing the significance of phi-co-efficient, co-efficient
of concordance and co-efficient of contingency we convert the
values obtained from these test into chi-square values. If the chi-
square value appears to be a significant one, we also take their
original values as significant.

Check your Progress =2


3. For a chi-square test, the observed frequencies are always whole
numbers. (True or false?)

4. For a chi-square test, the expected frequencies are always whole


numbers. (True or false?)

115 Self-instructional Material


Non- parametric tests 5. A researcher has developed three different designs for a computer
keyboard. A sample of n = 60 participants is obtained, and each individual
tests all three keyboards and identifies his or her favourite. The frequency
NOTES distribution of preferences is as follows:

Design A Design B Design C


23 12 25

a. What is the dfvalue for the chi-square statistic?


b. Assuming that the null hypothesis states that there are no preferences
among the three designs, find the expected frequencies for the chi-square
test.

10.3 Contingency Co-efficient:


The contingency coefficient is a coefficient of association that tells
whether two variables or data sets are independent or dependent of each
other. It is also known as Pearson’s Coefficient. It is based on the chi-
square statistic, and is defined by:

in this formula:

 χ2 is the chi-square statistic,


 N is the total number of cases or observations in our analysis/study,
 C is the contingency coefficient.

The contingency coefficient helps us decide if variable b is ‗contingent‘ on


variable a. However, it is a rough measure and doesn‘t quantify the
dependence exactly; It can be used as a rough guide:

 If C is near zero (or equal to zero) you can conclude that your
variables are independent of each other; there is no association
between them.
 If C is away from zero there is some relationship; C can only take on
positive values.

10.4 Median test:


The median test provides a nonparametric alternative to the independent-
measures ttest (or ANOVA) to determine whether there are significant
differences among two or more independent samples. The null hypothesis
for the median test states that the different samples come from populations
that share a common median (no differences). The alternative hypothesis

Self-instructional Material 116


Non- parametric tests
states that the samples come from populations that are different and do not
share a common median.

10.4.1Steps in the computation of the median test NOTES


1. The first step in computation of a median test is to compute a
common median for both distribution taken together.

2. Subsequently, a 2 x 2 contingency table is set.one for individuals


with scores above the median and one for individuals with scores
below the median. Finally, for each sample, count how many
individuals scored above the combined median and how many
scored below. These values are the observed frequencies that are
entered in the matrix. The frequency-distribution matrix is
evaluated using a chi-square test for independence.
3. The values are substituted in the following formula

N  IAD  BCI 
2

x 
2

 A  B  C  D  A  C  B  D 
4. When frequencies are less than 5 Yate‘s correction need to be
applied. Formula for Yates correction

N  IAD  BCI /  N / 2
2

x 
2

 A  B  C  D  A  C  B  D 
5. Testing the significance of the obtained X2 value with the help of
the table

Illustration

Group A 16, 17, 18, 12, 14, 9, 7, 5, 20, 22, 4, 26, 27, 5, 10, 19

Group B 28, 30, 33, 40, 45, 47, 40, 38, 42, 50, 20, 18, 18, 19

For computing median, both distribution all put together

49 - 53 1
44 - 48 2
39 - 43 3
( N / 2  F )i
34 - 38 1 Median  f 
fm
29 - 33 2
24 - 28 3

117 Self-instructional Material


 15  13 
Non- parametric tests
19 - 23 5  18.5   5
 5 
NOTES 14 - 18 5
9 - 13 3 = 20.5
4-8 5

The 2 x 2 contingency table for the above data

Above Mdn Not above Mdn Total

A B

Group A 3 13 16

C D

Group B 10 4 14

Total 13 17 30

30  / 3(4)  (13)(10) / 
2

R2 
(16) (14) (13) (17)

= 8.44

df = (r-1) (c-1) = 1

The obtained chi-square value of 8.44 exceeds the value of 6.65 in


the probability table for chi-square. Hence we reject the Null Hypothesis
and conclude that the two samples have not been drawn from the same
population on from population having equal medians.
10.5 Sign test:
The sign test may be used for testing the significance of differences
between two correlated samples in which the data is available either in
ordinal measurement or simply expressed in terms of positive and negative
signs, showing the direction of differences existing between the observed
scores of matched pairs. The null hypothesis test here is that the median
change is zero, that is there are equal numbers of positive and negative
signs. For example, a clinician may observe patients before therapy and
after therapy and simply note whether each patient got better or worse.
Note that there is no measurement of how much change occurred; the
clinician is simply recording the direction of change. Also note that the
direction of change is a binomial variable; that is, there are only two
values. In this situation it is possible to use a binomial test to evaluate the
data. Traditionally, the two possible directions of change are coded by
signs, with a positive sign indicating an increase and a negative sign
Self-instructional Material 118
Non- parametric tests
indicating a decrease. When the binomial test is applied to signed data, it is
called a sign test.

In many cases, data from a repeated-measures experiment can be NOTES


evaluated using either a sign test or a repeated-measures t test. In general,
you should use the t test whenever possible. However, there are some
cases in which a t test cannot or should not be used, and in these situations,
the sign test can be valuable. Four specific cases in which a t test is
inappropriate or inconvenient are described below.
1. When you have infinite or undetermined scores, a t test is impossible,
and the sign test is appropriate.
2. Often it is possible to describe the difference between two treatment
conditions without precisely measuring a score in either condition.
3. Often a sign test is done as a preliminary check on an experiment before
serious statistical analysis begins.
4. Occasionally, the difference between treatments is not consistent across
participants. This can create a very large variance for the difference scores.
As we have noted in the past, large variance decreases the likelihood that a
t test will produce a significant result. However, the sign test only
considers the direction of each difference score and is not influenced by the
variance of the scores

10.6 Friedman test

The Friedman test evaluates differences among three or more


groups from a repeated-measures design. The scores are the ranks obtained
by rank ordering the scores for each participant. With three conditions, for
example, each participant is measured three times and would receive ranks
of 1, 2, and 3.

Check your Progress- 3

6. What is the logic behind median test?

7. A researcher used a chi-square test for goodness of fit to determine


whether people had any preferences among three leading brands of potato
chips. Could the researcher have used a binomial test instead of the chi-
square test? Explain why or why not.

8. Why do you think t test is more powerful than the sign test?

10.7 Let’s sum up:


Chi-square tests are nonparametric techniques that test hypotheses about
the form of the entire frequency distribution. Two types of chi-square tests
are the test for goodness of fit and the test for independence. The data for
these tests consist of the frequency or number of individuals who are
119 Self-instructional Material
Non- parametric tests located in each category. The test for independence is used to assess the
relationship between two variables. The null hypothesis states that the two
variables in question are independent of each other. That is, the frequency
NOTES distribution for one variable does not depend on the categories of the
second variable. On the other hand, if a relationship does exist, then the
form of the distribution for one variable depends on the categories of the
other variable. The median test is a non-parametric alternative for
independent measures t test or ANOVA . The sign test evaluates the
difference between two treatments using the data from a repeated measures
design. The difference scores are coded as being either increases (+) or
decreases (-). The friedman test is a non-parametric alternative for Two
way ANOVA

10.8 Unit End Exercises


1. What are non – parametric tests? List its advantages and disadvantages.

2. A developmental psychologist would like to determinewhether infants


display any colour preferences. Astimulus consisting of four colour patches
(red, green,blue, and yellow) is projected onto the ceiling above acrib.
Infants are placed in the crib, one at a time, andthe psychologist records
how much time each infantspends looking at each of the four colours. The
colour thatreceives the most attention during a 100-second testperiod is
identified as the preferred colour for that infant. The preferred colours for a
sample of 60 infants areshown in the following table:
Red Green Blue Yellow
20 12 18 10

Do the data indicate any significant preferences among the four colors?
Test at the .05 level of significance.

3. What is contingency co-efficient? When is it used?

4. When should a sign test be preferred over t test?

10.9 Answers to check your progress


1. False. Non parametric tests make very few assumptions about the
population distribution and for this reason they are called distribution free
statistics

2. False. Non parametric statistics require measurement based upon a


nominal scale and ordinal scale.

3.True. Observed frequencies are obtained by counting people in the


sample.

Self-instructional Material 120


Non- parametric tests
4. False. Expected frequencies are computed and may be fractions or
decimal values.

5. a. df= 2 NOTES
b. According to the null hypothesis one-third of the population would
prefer each design. The expected frequencies should show one-third of the
sample preferring each design. The expected frequencies are all 20.

6. The logic behind the median test is that whenever several different
samples are selected from the same population distribution, roughly half of
the scores in each sample should be above the population median and
roughly half should be below.

7. No, the binomial test cannot be used when there are three categories.

8.t test is preferred over sign test as it uses the actual difference scores not
just the signs.

10.10 Suggested Readings

1. Tattao, L. (2007). Basic Concepts in Statistics. Retrieved from


https://books.google.com/books?id=og4a_700L-4C on January 10, 2018

2.Mangal S.K. (1987). Statistics in Psychology and Education. New Delhi:


PHI Learning Private Ltd.
3.Garrett.H.E.(1981). Statistics in Psychology and Education.Vakils, Feffer
and Simons Ltd.
4. Aron, Aron., E. N., Coups , E.(2012). Statistics for Psychology.(3rd Ed).
Prentice Hall.

121 Self-instructional Material


Preparation of Data for Computer

BLOCK IV: ANALYSIS AND


NOTES INTERPRETATION OF DATA

UNIT 11: PREPARATION OF DATA


FOR COMPUTER
11.1 Introduction:
11.2 Software packages that are used for statistical analysis
11.2.1. Programs for Quantitative Research.
11.2.2 Programs for Qualitative Research
11.3 Let’s sum up
11.4 Unit End Exercises
11.5 Answers to check your progress
11.6 Suggested Readings

11.1 1ntroduction:
For statistical analysis we think of data as a collection of different
pieces of information or facts. These pieces of information are called
variables. A variable is an identifiable piece of data containing one or
more values. Those values can take the form of a number or text
(which could be converted into number). For data analysis your data
should have variables as columns and observations as rows. The first
row should have the column headings. Make sure your dataset has at
least one identifier (for example, individual id, family id, etc.). That is,
it should have a serial number for each observation
Example:

id Var1 Var2 Var3 Var 4


1 7 3 5 2
2 6 9 10 3
3 5 6 4 5
4 4 7 1 8

In the table above ―var‖ stands for variable. Instead of this, we can
actually have the actual variable that is used for the study. If we have
more than one group, all data of group 1 has to be entered followed by
group 2, group 3 and so on.

Example id Gender Self - Self - Aspiration


esteem efficacy
Group 1 1 1 13 15 12
2 2 19 10 13
Group 2 3 2 16 14 15
4 1 17 21 18

Self-instructional Material 122


Preparation of Data for Computer

Note that a nominal variable like gender has been converted into
number. This is done by assigning values to each gender. For example, NOTES
Male has been assigned a value of 1 and female has been assigned a
value of 2. It is also possible to enter the data in terms of category as
either ―male‖ or ―female‖ itself.

Once you have your data in the proper format, before you perform any
analysis you need to explore and prepare it first. Here are a few things to
keep in mind before you start

1. Make sure variables are in columns and observations in rows.

2. Make sure you have all variables you need.

3. Make sure there is at least one id.

4. If times series make sure you have the years you want to include in your
study.

5. Make sure missing data has either a blank space or a dot (‗.‘)

6. Make sure to make a back-up copy of your original dataset.

7. Have the codebook handy.

8. If you are using datasets with categorical variables you need to clean
them by getting rid of the non-response categories like ‗do not know‘, ‗no
answer‘, ‗no applicable‘, ‗not sure‘, ‗refused‘, etc. No response categories
not only affect the statistics of the variable, it may also affect the
interpretation and coefficients of the variable if we do not remove them.

Check your Progress 1:

 While entering data into a software, the variables are entered in rows:
True/False
 How do you enter missing data into your data sheet?

11.2 Software packages that are used for statistical analysis:


Statistical software are programs which are used for the statistical analysis
of the collection, organization, analysis, interpretation and presentation of
data. Statistical packages are collections of software designed to aid
in statistical analysis and data exploration. In the last two decades more
and more software packages have been designed to help with data analysis.
The software is designed for questionnaire-based research,
called quantitative research, and for other types of research, such as
interviews and focus groups, which is called qualitative research. Let's
have a look at the programs and packages for these two types of research.

123
Self-instructional Material
Preparation of Data for Computer

11.2.1. Programs for Quantitative Research

NOTES There are numerous software packages for quantitative research and listed
below are few common ones:
1. (IBM) SPSS, (Statistical Package for the Social Sciences): SPSS is
the most widely used statistics software package within human behaviour
research. SPSS has built-in data manipulation tools such as recoding,
transforming variables. Social scientists generally learn SPSS as their main
package. It gives many options for statistical analysis and offers output in
the form of tables, graphs, pie-charts and other diagrams. The software is
easy to use, but that does not mean it is problem-free. The data put in must
be correct and the suitable form of analysis chosen, otherwise the results
will be misleading or simply wrong. Thus, you still need a knowledge of
statistics to choose the appropriate method and necessary parameters of the
analysis.
2. R (R Foundation for Statistical Computing): R is an integrated suite
of software facilities for data manipulation, calculation and graphical
display. It includes an effective data handling and storage facility, a suite
of operators for calculations on arrays, in particular matrices, a large,
coherent, integrated collection of intermediate tools for data analysis and
graphical facilities for data analysis and display either on-screen or on
hardcopy.
3. SAS (Statistical Analysis Software): SAS is perfect for traditional
analysis of variance and linear regression and meets both specialized and
enterprise-wide statistical needs. SAS uses the newest techniques for
statistical analysis and large data tasks for any size of data sets. The
software is organized in such a way that it helps you to access and manage
data, build and deploy statistical models.

4. STATA: Stata is an unified software which provides you with the


complete package required for data analysis, management and graphics. It
is one of the most important softwares available online for statistical
analysis

5. Minitab: Minitab is one of the best statistical softwares available


online. The software allows you to easily transfer Microsoft Excel XlSX
files directly into Minitab Express. You can evaluate confidence intervals
for a parameter of interest such as median or proportion using available
resampling techniques. Minitab helps you to find the best regression
equations with the help of a model reduction technique.

6. Microsoft Excel: Microsoft Excel is spread sheet software. In Excel,


you can perform some Statistical analysis. Working in excel is a little
tedious compared to other software packages as excel does not have built-
in data manipulation tools such as recoding, and transforming variables.
Self-instructional Material 124
Preparation of Data for Computer

Sharing an Excel spread sheet is a highly tedious affair. Spread sheets


create ample opportunities for accidental data loss, which makes it NOTES
impossible to share crucial data and information.

7. Max Stat: This is very easy-to-use and affordable statistical software


available online. Since it provides step by step analysis it is handy for
students and young scholars.

11.2.2 Programs for Qualitative Research

Very often, qualitative research involves dealing with long texts, and so
the software involves such functions as text retrieval, coding, and building
conceptual networks. For example, this software works if you want to
analyze the content of a document for key words or themes and then rank
the words relating to a topic by frequency of use. For example, the
Computer Assisted Qualitative Data Analysis (CAQDAS) Networking
Project has been set up to help researchers find out which is the best
software for their needs.
Advantages of using qualitative data analysis software include being
freed from manual and clerical tasks, saving time, being able to deal with
large amounts of qualitative data, having increased flexibility, and having
improved validity and auditability of qualitative research. Concerns
include increasingly deterministic and rigid processes, privileging of
coding, and retrieval methods; reification of data, increased pressure on
researchers to focus on volume and breadth rather than on depth and
meaning, time and energy spent learning to use computer packages,
increased commercialism, and distraction from the real work of analysis.
Hence it is recommended researchers consider the capabilities of the
package, their own computer literacy and knowledge of the package, or the
time required to gain these skills, and the suitability of the package for
their research. (John & Johnson, 2000)
Check your Progress: 2
 What do you understand by the term statistical software?
 What do you understand by the term qualitative research?
 What advantage does SPSS have over excel?
11.3 Let’s sum up:

For many students, the thought of having to undertake statistical analyses


is uncomfortable as they find mathematics and statistics difficult.
However, there is lots of support available to make a student comfortable
with undertaking statistical analyses such as online courses , you tube
videos and even MOOC courses . In addition, there are a multitude of
statistical software packages available that can aid in analysis of data
125
Self-instructional Material
Preparation of Data for Computer

obtained from both quantitative and qualitative research. It is for most part
very easy to learn to use them. To the question ―How do I get my data into
NOTES a statistical package?‖. The answer is that most statistical software can read
data directly from an Excel spread sheet, so using Excel is often the easiest
solution. Secondly, you can always enter data directly into a statistical
package, since they nearly all have some form of inbuilt spread sheet.
Another solution is to use software like Survey Monkey to collect the data.
Survey Monkey has the facility to convert the data into an Excel spread
sheet or SPSS format. Though software packages exist to aid the
researcher in data analysis, researcher‘s intelligence and skill is required in
choosing the appropriate statistical tool for data analysis and interpretation
11.4 Unit End Exercises

1. Why is it important to clean the non –response categories?


2. What are the advantages and disadvantages of using data analysis
packages for qualitative research?
3. List statistical software packages commonly employed with
quantitative research.

11.5 Answers to check your progress

1. False

2. The missing data should either a blank space or a dot (‗.‘).

3. Statistical software are programs which are used for


the statistical analysis of the collection, organization, analysis,
interpretation and presentation of data.

4. Qualitative research methods are defined as a process that focuses on


obtaining data through open-ended and conversational communication
such as focus groups and interviews.

5. Excel does not have built-in data manipulation tools such as recoding,
and transforming variables like SPSS.

11.6 Suggested Readings:


6. https://www.princeton.edu/~otorres/DataPrep101.pdf
7. https://woofresh.com/statistical-analysis-software/
8. https://www.researchgate.net/publication/12189428_The_pr
os_and_cons_of_data_analysis_software_for_qualitative_re
search.

Self-instructional Material 126


Inferential Statistics

UNIT 12: INFERENTIAL STATISTICS NOTES

12.1Introduction to Inferential Statistics


12.2 Types of Hypothesis
12.3 The Meaning of Statistical Inference
12.3.1 Relationship between probability and Inferential
Statistics
12.4 Methods of statistical Inference
12.4.1 Statistical Estimation
12.4.2 Statistical hypothesis testing

12.5 Let us sum up


12.6 Unit end Exercises
12.7 Answers to check your progress
12.8 Suggested Readings

12.1 Introduction to Inferential Statistics:

Inferential statistics are methods that use sample data as the basis for
drawing general conclusions about populations. However, a sample is not
expected to give a perfectly accurate reflection of its population. In
particular, there will be some error or discrepancy between a sample
statistic and the corresponding population parameter. There are two
sources of error that may result in samples being different from the
population from which it is drawn that include sampling error and
sampling bias.

Sampling error is the natural discrepancy, or amount of error, between a


sample statistic and its corresponding population parameter. Sampling bias
is a bias in which a sample is collected in such a way that some members
of the intended population have a lower sampling probability than others.A
common cause of sampling bias lies in the design of the study or in the
data collection procedure, both of which may favour or disfavour
collecting data from certain classes or individuals or in certain conditions

There are two methods of inferential statistics. They are estimation and
hypothesis testing. There are few key concepts that form the foundations of
inferential statistics. They include knowledge of a) Probability (covered in
Chapter 6) b) z score (covered in chapter 7) c) Estimation and Hypothesis
testing procedure d) The distribution of sample means. The first two topics
has been covered and the next two will be presented in the current and the
next chapter respectively
12.2 Types of Hypotheses:

a) Null hypothesis (Ho): The null hypothesis states that in the general
population, there is no change, no difference, or no relationship. In the

127 Self-instructional Material


Inferential Statistics

context of an experiment Ho predicts that independent variable (treatment)


will have no effect on the dependent variable for the population.
NOTES b) The alternate hypothesis (H1) states that there is a change, there is a
difference, or there is a relationship. In the context of an experiment, H1
predicts that the independent variable (treatment) will have an effect on the
dependent variable.

Check your Progress - 1


1. What do you mean by the term statistics and parameters?
2. What are the sources of sampling bias?
3. What is the goal of inferential statistics?
4. Null hypothesis predicts that independent variable have an effect on the
dependent variable: T/F

12.3 The meaning of statistical inference:


Statistical inference is defined as the process inferring the properties of the
given distribution based on the data. In other words, it deduces the
properties of the population by conducting hypothesis testing and obtaining
estimates from the data obtained from samples.

12.3.1.Relationship between probability and statistical inference:


Statistics are, in one sense, all about probabilities. Inferential statistics deal
with establishing whether differences or associations exist between sets of
data. The data comes from the sample we use, and the sample is taken from
a population. So we need to think about whether the sample represents the
population from which it has been taken. The larger the sample we take the
greater the probability that it is representative of the population. If we took
the whole population for our study the probability would = 1, since the
sample = the population. A sample smaller means that we cannot guarantee
that it is similar to the population. In order for generalizability from sample
to population to be meaningful, we aim to reduce sampling error.
Probability forms a direct link between samples and the populations from
which they come. The following example provides a brief overview of
probability can be used in the context of inferential statistics. The research
begins with a population that forms a normal distribution with a mean of μ
= 400 and a standard deviation of = 20. A sample is selected from the
population and a treatment is administered to the sample. The goal for the
study is to evaluate the effect of the treatment. To determine whether the
treatment has an effect, the researcher simply compares the treated sample
with the original population. If the individuals in the sample have scores
around 400 (the original population mean), then we must conclude that the
treatment appears to have no effect. On the other hand, if the treated
individuals have scores that are noticeably different from 400, then the
researcher has evidence that the treatment does have an effect. Notice that

128
Self-instructional Material
Inferential Statistics

the study is using a sample to help answer a question about a population;


this is the essence of inferential statistics. NOTES

The problem for the researcher is determining exactly what is meant by


―noticeably different‖ from 400. If a treated individual has a score of X =
415, is that enough to say that the treatment has an effect? What about X =
420 or X =450? On way of answering it is to use z scores. We know, from
the normal probability distribution that a z-score value beyond z +2.00 (or
–2.00) was an extreme value and, therefore, noticeably different. But the
choice of +2 or -2 was arbitrary. The other way to answer this question is
to use probability, to help us decide exactly where to set the boundaries.

Figure shows the setting of boundaries for alpha level of 0.05. The middle
95% of the distribution consist of high probability values if the null
hypothesis were true and the extreme 5%, which when divided in half
produces proportions of 0.0250 in the right-hand tail and 0.0250 in the left-
hand tail. Using the unit normal table, we find that the z-score boundaries
for the right and left tails are z - +1.96 and z = –1.96, respectively. The
boundaries set at z + or - 1.96 provide objective criteria for deciding
whether our sample provides evidence that the treatment has an effect.
Specifically, we use the sample data to help decide between the following
two alternatives:

1. The treatment has no effect. After treatment, the scores still average μ =
400.
2. The treatment does have an effect. The treatment changes the scores so
that, after treatment, they no longer average μ = 400.

As a starting point, we assume that the first alternative is true and


the treatment has no effect. In this case, treated individuals should be no
different from the individuals in the original population, which is shown in
the above figure. If our assumption is correct, it is extremely unlikely
129 Self-instructional Material
Inferential Statistics

(probability less than 5%) for a treated individual to be outside the +/- 1.96
boundaries. Therefore, if we obtain a treated individual who is outside the
boundaries, we must conclude that the assumption is probably not correct.
NOTES
In this case, we are left with the second alternative (the treatment does
have an effect) as the more likely explanation. Notice that we are
comparing the treated sample with the original population to see if the
sample is noticeably different. If it is different, we can conclude that the
treatment seems to have an effect. Now we are defining ―noticeably
different‖ as meaning ―very unlikely.‖ We are using the sample data and
the + or - 1.96 boundaries, which were determined by probabilities, to
make a general decision about the treatment. If the sample falls outside the
boundaries we make the following logical conclusion:
a. This kind of sample is very unlikely to occur if the treatment has no
effect.
b. Therefore, the treatment must have an effect that changed the sample.

On the other hand, if the sample falls between the + or - 1.96 boundaries,
we conclude:
a. This is the kind of sample that is likely to occur if the treatment has no
effect.
b. Therefore, the treatment does not appear to have had an effect.

12.4 Methods of statistical Inference

12.4.1 Statistical Estimation: An important aspect of statistical inference


is using estimates to approximate the value of an unknown population
parameter. For example, sample means are used to estimate population
means; sample proportions, to estimate population proportions.

An estimate of a population parameter may be expressed in two ways:

a) Point estimate: A point estimate of a population parameter is a single


value of a statistic. For example, the sample mean x is a point estimate of
the population mean μ. Similarly, the sample proportion p is a point
estimate of the population proportion P. Point estimates are rarely stated.

b) Interval estimate: An interval estimate is defined by two numbers,


between which a population parameter is said to lie. For
example, a < x < b is an interval estimate of the population mean μ. It
indicates that the population mean is greater than a but less than b. Of
course we may be wrong in supposing that the stated limits contain the
population value. Other things being equal, if we set wide limits, the
likelihood that the limits will include the population value is high and if we
set narrow limits , there is a greater risk of being wrong. Because the
130
Self-instructional Material
Inferential Statistics

option of setting wider or narrower limit exists, any statement of limits NOTES
must be accompanied by indication of the probability that the limits
contain the population parameters. The limits themselves are usually
referred to as confidence interval. To express a confidence interval, you
need three pieces of information.

 Confidence level
 Statistic
 Margin of error

Given these inputs, the range of the confidence interval is defined by


the sample statistic + margin of error. And the uncertainty associated with
the confidence interval is specified by the confidence level.

Check your progress – 2

5. What are confidence intervals?

12.4.2 Statistical hypothesis testing: Hypothesis testing is a statistical


procedure that allows researchers to use sample data to draw inferences
about the population of interest. The general goal of hypothesis treating is
to determine whether or not a particular treatment has an effect on a
population. The test is performed by selecting a sample, administering the
treatment to the sample, and then comparing the result with the original
population. If the treated sample is noticeably different from the original
population, then we conclude that the treatment has an effect, and we reject
null hypothesis (H0). On the other hand if the treated sample is still similar
to the original population, then we conclude that there is no evidence for
treatment effect and we fail to reject H0.
The critical factor in this decision is the size of the difference between the
treated sample and the original population. A large difference is evidence
that the treatment worked, a small difference is not sufficient to say that the
treatment has any effect. The hypothesis testing can either be in the one
tailed or two tailed test format. In a directional hypothesis test or a one
tailed treat, the statistical hypothesis (null hypothesis (H 0) and alternate
hypothesis (H1) specify either an increase or a decrease in the population
mean score that is, they make a statement about the directions of the effect.

In very simple terms, the hypothesis-testing procedure is as follows:

For the purpose of illustration, let‘s take the following example. Let‘s say
our researcher selects a test for which adults older than 65 have an average
score of µ= 80 with a standard deviation of σ = 20. The distribution of test
scores is approximately normal. The researcher‘s plan is to obtain a sample
of n =25 adults who are older than 65, and give each participant a daily
dose of a blueberry supplement that is very high in antioxidants. After
taking the supplement for 6 months, the participants are given the
131 Self-instructional Material
Inferential Statistics

neuropsychological test to measure their level of cognitive function. If the


mean score for the sample is noticeably different from the mean for the
general population of elderly adults, then the researcher can conclude that
NOTES
the supplement does appear to have an effect on cognitive function. On the
other hand, if the sample mean is around 80 (the same as the general
population mean), the researcher must conclude that the supplement does
not appear to have any effect.

Step 1: State the hypothesis:


First, we state a hypothesis about a population. Usually the hypothesis
concerns the value of a population parameter. For the study in example ,
the null hypothesis states that the blueberry supplement has no effect on
cognitive functioning for the population of adults who are more than 65
years old.
In symbols, this hypothesis is H0: µ (with supplement) = 80
For this example, the alternative hypothesis states that the supplement does
have an effect on cognitive functioning for the population and will cause a
change in the mean score. In symbols, the alternative hypothesis is
represented as H1: µ with supplement ≠ 80

Step 2: Set the criteria for decision:


Eventually the researcher uses the data from the sample to evaluate the
credibility of the null hypothesis. The data either provide support for the
null hypothesis or tend to refute the null hypothesis. In particular, if there is
a big discrepancy between the data and the null hypothesis, then we
conclude that the null hypothesis is wrong.To formalize the decision
process, we use the null hypothesis to predict the kind of sample mean that
ought to be obtained. Specifically, we determine exactly which sample
means are consistent with the null hypothesis and which sample means are
at odds with the null hypothesis.

For our example, the null hypothesis states that the supplement has no
effect and the population mean is still µ = 80. If this is true, then the
sample mean should have a value around 80. Therefore, a sample mean
near 80 is consistent with the null hypothesis. On the other hand, a sample
mean that is very different from 80 is not consistent with the null
hypothesis. To determine exactly which values are ―near‖ 80 and which
values are ―very different from‖ 80, we examine all of the possible sample
means that could be obtained if the null hypothesis is true. For our
example, this is the distribution of sample means for n = 25. According to
the null hypothesis, this distribution is centered at µ = 80. The distribution
of sample means is then divided into two sections:

1. Sample means that are likely to be obtained if H0 is true; that is, sample
means that are close to the null hypothesis

132
Self-instructional Material
Inferential Statistics

2. Sample means that are very unlikely to be obtained if H0 is true; that is,
sample means that are very different from the null hypothesis NOTES

The alpha level To find the boundaries that separate the high-probability
samples from the low-probability samples, we must define exactly what is
meant by ―low‖ probability and ―high‖ probability. This is accomplished
by selecting a specific probability value, which is known as the level of
significance, or the alpha level, for the hypothesis test. The alpha (α) value
is a small probability that is used to identify the low probability samples.
By convention, commonly used alpha levels are α = .05 (5%), α =.01 (1%),
and α = .001 (0.1%). For example, with α = .05, we separate the most
unlikely 5% of the sample means (the extreme values) from the most likely
95% of the sample means (the central values).

The extremely unlikely values, as defined by the alpha level, make up what
is called the critical region. These extreme values in the tails of the
distribution define outcomes that are not consistent with the null
hypothesis; that is, they are very unlikely to occur if the null hypothesis is
true. Whenever the data from a research study produce a sample mean that
is located in the critical region, we conclude that the data are not consistent
with the null hypothesis, and we reject the null hypothesis.
.
Technically, the critical region is defined by sample outcomes that are very
unlikely to occur if the treatment has no effect (that is, if the null
hypothesis is true). Reversing the point of view, we can also define the
critical region as sample values that provide convincing evidence that the
treatment really does have an effect. For our example, the regular
population of elderly adults has a mean test score of µ=80. We selected a
sample from this population and administered a treatment (the blueberry
supplement) to the individuals in the sample. What kind of sample mean
would convince you that the treatment has an effect? It should be obvious
that the most convincing evidence would be a sample mean that is really
different from µ=80. In a hypothesis test, the critical region is determined
by sample values that are ―really different‖ from the original population.

Step 3: Collect data and compute sample statistics:

At this time, we select a sample of adults who are more than 65


years old and give each one a daily dose of the blueberry supplement. After
6 months, the neuropsychological test is used to measure cognitive
function for the sample of participants. Notice that the data are collected
after the researcher has stated the hypotheses and established the criteria
for a decision. This sequence of events helps to ensure that a researcher
makes an honest, objective evaluation of the data and does not tamper with
the decision criteria after the experimental outcome is known. Next, the

133 Self-instructional Material


Inferential Statistics

raw data from the sample are summarized with the appropriate statistic like
z or t.
NOTES
X   obtained difference between data & hypothesis
Z 
X Standard distance expected by chance

Step 4: Make a decision

In the final step, the researcher uses the z-score value or t score obtained in
step 3 to make a decision about the null hypothesis according to the criteria
established in step 2. There are two possible outcomes:

1. The sample data are located in the critical region. By definition, a


sample value in the critical region is very unlikely to occur if the null
hypothesis is true. Therefore, we conclude that the sample is not consistent
with H0 and our decision is to reject the null hypothesis. Remember, the
null hypothesis states that there is no treatment effect, so rejecting H0
means that we are concluding that the treatment did have an effect.

2. The second possibility is that the sample data are not in the critical
region. In this case, the sample mean is reasonably close to the population
mean specified in the null hypothesis (in the center of the distribution).
Because the data do not provide strong evidence that the null hypothesis is
wrong, our conclusion is to fail to reject the null hypothesis. This
conclusion means that the treatment does not appear to have an effect.

Check your progress -3


6. A researcher selects a sample of n=16 individuals from a normal
population with the mean of µ=40 and σ=8. A treatment is administered to
the sample and, after treatment , the sample mean is M=43.If the researcher
uses a hypothesis test to evaluate the treatment effect , What z score would
be obtained for this sample?

7. A z score value in the critical region means you should accept the null
hypothesis – T/F

8. A small value near zero for the z score statistic is evidence that a sample
data are consistent with null hypothesis – T/F

12.5 Let us sum up:

In thisChapter, we introduced inferential statistics as a basis for drawing


statistical inferences about the population from the sample data. Two main
methods of statistical inference namely estimation and hypothesis testing
has been covered. The two types of hypothesis and the procedure
underlying hypothesis testing has been structured into a four step process
for recall.

134
Self-instructional Material
Inferential Statistics

12.6 Unit end Exercises:


NOTES
1. Explain why probability is the basis for statistical inference.

2. Discuss the logic underlying the hypothesis testing procedure.

3. An automobile manufacturer claims that a new model will average μ -


45 miles per gallon with σ = 4. A sample of n= 16 cars is tested and
averages only M = 43 miles per gallon. Is this sample mean likely to occur
if the manufacturer‘s claim is true? Specifically, is the sample mean within
the range of values that would be expected 95% of the time? (Assume that
the distribution of mileage scores is normal.)

12.7 Answers to check your progress:

1. Measures such as mean, median, variance etc descriptive of the sample


are called statistics and measures descriptive of the population are called
parameters.

2. An improper study design as well as a faulty data collection procedure


can contribute to sampling bias.

3. The goal of inferential statistics is to infer or make conclusions about the


population parameters based on sample statistics

4. False

5. A confidence interval, in statistics, refers to the probability that a


population parameter will fall between two set values for a certain
proportion of times. Confidence intervals measure the degree of
uncertainty or certainty in a sampling method.

6. The standard error is 2 points and z= 3/2= 1.50

7. False
8. True

12.8 Suggested Readings

1. Garrett.H.E.(1981). Statistics in Psychology and Education. Vakils,


Feffer and Simons Ltd.
2. Gravetter,J., &Wallnau,B.(1995). Essentials of statistics (2nd Ed.).West
Publishing Company.
3. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural sciences
(9th Ed.). Cengage Learning India Private Ltd.
4. Mangal S.K. (1987). Statistics in Psychology and Education. New Delhi:
PHI Learning Private Ltd.

135 Self-instructional Material


The Random Sampling Distribution

UNIT 13: THE RANDOM SAMPLING


NOTES DISTRIBUTION

13.1Population and Samples


13.2The Distribution of sample means
13.3Properties of the distribution of sample means
13.3.1 The shape of the distribution of sample means
13.3.2 The mean of the distribution of sample means: The
expected value of M
13.3.3 The standard error of M
13.4 Sampling distribution of differences between means
13.5 Introduction to t statistics
13.5.1: Illustration of hypothesis testing with t statistics
for Independent measures research design
13.5.2 : Illustration of hypothesis testing with t statistics
Repeated measures research design
13.6 Choice of alterative hypothesis: One tailed and two
tailed tests of significance
13.7 Let us sum up
13.8Unit end exercises
13.9 Answers to check your progress
13.10 Suggested Readings

13.1 Population and Samples


Before you begin, reading this unit, make sure you have refreshed your
knowledge on random sampling, z scores, probability and normal
distribution. Whenever a score is selected from a population, you should be
able to compute a z-score that describe exactly where the score is located in
the distribution. If the population is normal, you also should be able to
determine the probability value for obtaining any individual score. In a
normal distribution, for example, any score located in the tail of the
distribution beyond z (+) or (-) 2.00 is an extreme value, and a score this
large has a probability of only p = 0.0228. However, the z-scores and
probabilities that we have considered so far are limited to situations in
which the sample consists of a single score

Most research studies involve much larger samples, such as n = 25


preschool children or n = 100 American Idol contestants. In these
situations, the sample mean, rather than a single score, is used to answer
questions about the population. In this chapter we extend the concepts of z-
scores and probability to cover situations with larger samples. In particular,

136
Self-instructional Material
we introduce a procedure for transforming a sample mean into a z-score. The Random Sampling Distribution

Thus, a researcher is able to compute a z-score that describes an entire


sample. As always, a z-score near zero indicates a central, representative
sample; a z-score beyond 2.00 or –2.00 indicates an extreme sample. Thus,
it is possible to describe how any specific sample is related to all the other
possible samples. In addition, we can use the z-scores to look up NOTES
probabilities for obtaining certain samples, no matter how many scores the
sample contains.

Suppose, for example, a researcher randomly selects a sample of n = 25


students from the state college. Although the sample should be
representative of the entire student population, there are almost certainly
some segments of the population that are not included in the sample. In
addition, any statistics that are computed for the sample are not identical to
the corresponding parameters for the entire population. For example, the
average IQ for the sample of 25 students is not the same as the overall
mean IQ for the entire population. This difference, or error, between
sample statistics and the corresponding population parameters is called
sampling error .Furthermore, samples are variable; they are not all the
same. If you take two separate samples from the same population, the
samples are different. They contain different individuals, they have
different scores, and they have different sample means. How can you tell
which sample gives the best description of the population? Can you even
predict how well a sample describes its population? What is the probability
of selecting a sample with specific characteristics? These questions can be
answered once we establish the rules that relate samples and populations

13.2 Distribution of sample means:

Two separate samples probably are different even though they are taken
from the same population. The samples have different individuals, different
scores, different means, and so on. In most cases, it is possible to obtain
thousands of different samples from one population. With all these
different samples coming from the same population, it may seem hopeless
to try to establish some simple rules for the relationships between samples
and populations. Fortunately, however, the huge set of possible samples
forms a relatively simple and orderly pattern that makes it possible to
predict the characteristics of a sample with some accuracy. The ability to
predict sample characteristics is based on the distribution of sample means.

The distribution of sample means is the collection of sample means for all
of the possible random samples of a particular size (n) that can be
obtained from a population. Since a sampling distribution is a distribution
of statistics obtained by selecting all of the possible samples of a specific
size from a population, the distribution of sample means is an example of a
sampling distribution. In fact, it often is called the sampling distribution of
M.
137 Self-instructional Material
The Random Sampling Distribution

If you actually wanted to construct the distribution of sample means, you


would first select a random sample of a specific size (n) from a population,
NOTES calculate the sample mean, and place the sample mean in a frequency
distribution. Then you would select another random sample with the same
number of scores. Again, you would calculate the sample mean and add it
to your distribution. You would continue selecting samples and calculating
means, over and over, until you had the complete set of all the possible
random samples. At this point, your frequency distribution would show the
distribution of sample means.

What do you expect the distribution to look like? Since samples tend to be
representative of the population, it is reasonable to expect the sample
means should pile up around the population mean. The pile of sample
means should tend to form a normal-shaped distribution as most of the
samples should have means close to µ, and it should be relatively rare to
find sample means that are substantially different from µ. The distribution
of sample means is finally used to answer probability questions about
sample means.

13.3 Properties of the distribution of sample means

13.3.1 The shape of the distribution of sample means

It has been observed that the distribution of sample means tends to be a


normal distribution. In fact, this distribution is almost perfectly normal, if
either of the following two conditions is satisfied:

1. The population from which the samples are selected is a normal


distribution.

2. The number of scores (n) in each sample is relatively large, around 30 or


more. (As n gets larger, the distribution of sample means more closely
approximates a normal distribution. When n > 30, the distribution is almost
normal, regardless of the shape of the original population. However when n
< 30, it tends to take the form of ―t‖ distribution)

The fact that the distribution of sample means tends to be normal is not
surprising. Whenever you take a sample from a population, you expect the
sample mean to be near to the population mean. When you take lots of
different samples, you expect the sample means to ―pile up‖ around µ,
resulting in a normal-shaped distribution.

13.3.2 The mean of the distribution of sample means: The expected


value of M

The distribution of sample means is centered around the mean of the


population from which the samples were obtained. In fact, the average
value of all the sample means is exactly equal to the value of the
population mean. This fact should be intuitively reasonable; the sample
138
Self-instructional Material
means are expected to be close to the population mean, and they do tend to The Random Sampling Distribution

pile up around µ. The formal statement of this phenomenon is that the


mean of the distribution of sample means always is identical to the
population mean. This mean value is called the expected value of M.

13.3.3 The standard error of M NOTES


To completely describe this distribution, we need one more characteristic,
variability. The standard deviation of the distribution of sample means, M,
is called the standard error of M. The standard error, like standard
deviation serves the two purposes for the distribution of sample means.

1. The standard error describes the distribution of sample means. It


provides a measure of how much difference is expected from one sample
to another. When the standard error is small, then all of the sample means
are close together and have similar values. If the standard error is large,
then the sample means are scattered over a wide range and there are big
differences from one sample to another. The magnitude of the standard
error is determined by two factors:

(a) The size of the sample: In general, as the sample size increases, the
error between the sample mean and the population mean should decrease.
This rule is also known as the law of large numbers

(b) The standard deviation of the population from which the sample is
selected: When the sample consists of a single score (n=1), the standard
error is the same as the standard deviation (M).

2. Standard error measures how well an individual sample mean represents


the entire distribution. Specifically, it provides a measure of how much
distance is reasonable to expect between a sample mean and the overall
mean for the distribution of sample means.

However, because the overall mean is equal to µ, the standard error also
provides a measure of how much distance to expect between a sample
mean (M) and the population mean (µ). Remember that a sample is not
expected to provide a perfectly accurate reflection of its population.
Although a sample mean should be representative of the population mean,
there typically is some error between the sample and the population. The
standard error measures exactly how much difference is expected on
average between a sample mean, M, and the population mean, µ.

Check your Progress - 1

1. How is sampling error different from standard error?


2. When does the distribution of sample means take the shape of a normal
curve?

139 Self-instructional Material


The Random Sampling Distribution

Sampling distribution of differences between means:

NOTES Until this point, all the inferential statistics we have considered involve
using one sample as the basis for drawing conclusions about one
population. Although these single sample techniques are used occasionally
in real research, most research studies require the comparison of two (or
more) sets of data. For example, a social psychologist may want to
compare men and women in terms of their political attitudes, an
educational psychologist may want to compare two methods for teaching
mathematics, or a clinical psychologist may want to evaluate a therapy
technique by comparing depression scores for patients before therapy with
their scores after therapy. In each case, the research question concerns a
mean difference between two sets of data. There are two general research
designs that can be used to obtain the two sets of data to be compared:

1. The two sets of data could come from two completely separate groups of
participants. For example, the study could involve a sample of men
compared with a sample of women. Or the study could compare grades for
one group of freshmen who are given laptop computers with grades for a
second group who are not given computers.

2. The two sets of data could come from the same group of participants.
For example, the researcher could obtain one set of scores by measuring
depression for a sample of patients before they begin therapy and then
obtain a second set of data by measuring the same individuals after 6
weeks of therapy.

The first research strategy, using completely separate groups, is called an


independent measures research design or a between-subjects design. These
terms emphasize the fact that the design involves separate and independent
samples and makes a comparison between two groups of individuals. The
second research strategy using same participants is called repeated
measures research design

Whether it is a single sample mean whose population mean we are


hypothesizing or the difference two sample means, the procedure however
remains the same. That is instead of having a single mean, if we have data
related to differences between means of a number of samples , the
sampling distribution of differences between these means will also look
like a normal curve in case of large samples and like a t distribution in
case of small samples. The standard deviation of this distribution is called
standard error of the differences between means and may be taken as a
yardstick against which we may safely test the obtained difference between
sample means.

140
Self-instructional Material
13.4 Introduction to t statistics : The Random Sampling Distribution

It is usually impossible or impractical for a researcher to observe every


individual in a population. Therefore, researchers usually collect data
from a sample. There exist statistical procedures that permit researchers
to use a sample mean to test hypothesis about a population. These
statistical procedures are based on a few basic notions, which can be NOTES
summarized as follows.
1. A sample mean ( x ) is expected more or less to approximate its
population mean (). This permits us to use the sample mean to test a
hypothesis about the population mean.
2. The standard error provides a measure of how well a sample mean
approximates the population mean


 x (standard error of the mean) = (Standard deviation of
n
the population)

3. To quantify our inference about the population, we compare the


obtained sample mean ( x ) with the hypothesized population mean ()
by computing a z-score statistic.

X   obtained difference between data & hypothesis


Z 
X Standard distance expected by chance

The short coming of z score as an inferential statistic is that the z score


formula requires more information than is usually available. Z scores
require that we know the value of the population standard deviation
which is needed to compute standard error. Without standard error, we
have no way of quantifying the expected amount of distance (or error)
between sample mean and population mean. We have no way of
making precise, quantitative inferences about the population based on z
- score.
The ‗t‘ statistic is relatively simple solution to the problem of not
knowing the population standard deviation (). When the value of  is
not known, we use the sample standard deviation in its place.
The sample standard deviation was developed specifically to be an
unbiased estimate of the population standard deviation.
SS
s
n 1
(Sample standard deviation)
SS = sum of squares
using this sample statistic we can estimate the standard error.
s
sX  (sample standard deviation)
n

141 Self-instructional Material


The Random Sampling Distribution

s X = Estimated standard error


The estimated standard error (s x ) is used as an estimate x when the
NOTES value of (population standard deviation) is unknown. It is computed
from the sample standard deviation and provides an estimate of the
standard distance (expected by chance) between the sample mean ( x )
and the population mean ().

The ‗t‘ statistic is computed now by substituting the estimated standard


error in the denominator of the z score formula.
X 
t
sX
The only difference between the‗t‘ formula and z formula is that the z
score formula uses the actual population standard deviation ( ) and t
statistic uses the sample standard deviation as an estimate when  is
unknown.

Degrees of freedom and ‘t’ statistic

We must know the sample mean before we can compute the sample
standard deviation. This places a restriction on sample variability such
that only n-1, scores in a sample are free to vary. The value n-1 is
called the degrees of freedom (or df) for the sample standard deviation.
The greater the value of df is for a sample, the better s represents , and
the better the ‗t‘ statistic approximates the z - score. This should make
sense because the larger the sample (n) is, the better the sample
represents its population. Thus, the degrees of freedom associated with
a sample mean also describe how well ―t‖ represents ―z‖.

The t distribution

The t distribution is not a normal distribution. How well the‗t‘


distribution approximates a normal distribution is determined by
degrees of freedom. In general, the greater the sample size (n) is, the
larger the degrees of freedom (n-1) are, better the t distribution
approximates the normal distribution.
There is a different sampling distribution of ‗t‘ (a distribution of all
possible sample t values) for each possible number of degrees of
freedom. The ‗t‘ distribution has more variability than the normal
distribution, especially when of values are small. The ‗t‘ distribution
tends to be flatter and more spread out, whereas the normal z
distribution has more of a central peak. This is because of the fact that
the standard error in the ‗t‘ formula is not a constant because it is
estimated. That is estimated standard error is based on the sample
standard deviation, which will vary in value from sample to sample.
However when the value of df increases, the variably in the ‗t‘
distribution decreases and it more clearly resemble the normal

142
Self-instructional Material
distribution because with greater df, S  will more clearly estimate The Random Sampling Distribution
X

  , and when df is very large, they are nearly the same.


X

The‘t’ distribution table


The numbers in the‗t‘ distributions table are the values of t that
separate the tail from the main body of the distribution. Proportions for NOTES
one or two tails are listed on the op of the table, and df values for ―t‖
will be listed in the first column.

13.5.1‘t’ STATISTICS FOR AN INDEPENDENT MEASURES


RESEARCH DESIGN

Hypothesis tests with the independent measures ‘t’ statistics.

Step – 1: State the hypothesis. H0 & H1 and select an alpha level for the
independent measures t test. The hypotheses concern the difference
between the two population means.

H0 = M1 - M2 = 0

H1 = M1 - M2 ≠ 0

The hypothesis testing procedure will determine whether or not the


data provide evidence for a mean difference between the two populations.
At the conclusion of the hypothesis test, we well decide either to.

a) Reject H0 (we conclude that the data indicate a significant


difference between the two population mean) or to

b) Fail to reject H0 (The data do not provide sufficient evidence to


conclude that the difference exist.

Step 2: Locate the critical region. The critical region is defined as sample
data that would be extremely unlikely if the null hypothesis were
true. In this case, we will be locating extremely unlikely ―t‖ values.

Step 3: Get the data and compute the test statistic.

Sample statistic  hypothized population parameter


t
Estimated s tan dard error

Step 4: Make a decision: If the‗t‘ statistic we compute is in the critical


region, we reject H0. Otherwise we conclude that the data do not provide
sufficient evidence that the two populations are different.

Computation of difference between two means for small but


independent samples

Example: Two groups of 10 students each got the following scores on an


attitude scales.
143 Self-instructional Material
The Random Sampling Distribution

Group I: 10, 9, 8, 7, 7, 8, 6, 5, 6, 4

Group II: 9, 8, 6, 7, 8, 8, 11, 12, 6, 3


NOTES
Compute the means for both groups and test the significance of the
difference between these two means.

Sample 1 Sample 2
2
X1 x1 x1 X2 x2 x22
10 3 9 7 1 1
9 2 4 8 0 0
8 1 1 6 -2 4
7 0 0 7 -1 1
7 0 0 8 0 0
8 1 1 8 0 0
6 -1 1 11 3 9
5 -2 4 12 4 16
6 -1 1 6 -2 4
4 -3 9 5 -3 9
Total = 70 x12=30 Total = 80 x12=44

M1 = 70/10 = 7 M2 = 80/10 = 8

Pooled S.D or  =  x12   x2 2


( N1  1)  ( N 2  1)

30  44
 2.03
99

1 1
SED or D =  
N1 N 2

1 1
 2.03 
10 10

2.03
  0.908
2.2361

M1  M 2 7 8 1
t    1.1
D 0.908 0.908

df = (N1 + N2 ) - 2

= 10 + 10 - 2

= 18

144
Self-instructional Material
From the ‗t‘ distribution, we find that the critical value of ‗t‘ with The Random Sampling Distribution

18 degrees of freedom at 5%level of significance is 2.10. One computed


value of i.e. 1.1 is quite smaller than the critical table value 2.10 and hence
is not significant. Therefore, the null hypothesis cannot be rejected and as a
result, the given difference in sample means being insignificant, can only
be attributed to some chance factors or sampling fluctuations. NOTES
13.5.2.‘t’ STATISTIC FOR REPEATED MEASURES RESEARCH
DESIGN

A repeated measures study is one in which a single sample of individuals is


tested more than once on the dependent variable. The same subject are
used for every treatment condition.

Occasionally researches will try to approximate the advantages of a


repeated measures study by using a technique known as matched subjects.
In matched subjects study, each individual in one sample is matched with a
subject in the other sample. The matching is done so that the two
individuals are equivalent with respect to a specific variable that the
researcher would like to control.

The single sample ‗t‘ statistic is defined by the formula

X M
t
s
x

X= Sample mean

M = Population Mean

S s = Estimated standard error



x

The sample mean comes from the data. The standard error (also
computed from the sample data), gives a measure of the error between the
sample mean and the population mean, M.

For the repeated measures design, the sample data are difference
scores and one identified by the letter D, rather than X. Therefore, we will
substitute D‘s in the formula in place of X‘s to emphasize that we are
dealing with difference scores instead of X values. Also, the population
mean that is of interest to us is the population mean difference (the mean
amount of change for the entire population). With these simple changes,
the formula for the repeated measures design becomes,

D  D
t
s
D

145 Self-instructional Material


The Random Sampling Distribution

 D = Hypothesized population difference Estimated standard error of


difference
NOTES
s   Estimated standard error of difference
D

Illustration:

A researcher in behavioural medicine believes that stress often


makes asthma symptom worse for people who suffer from this respiratory
disorder. Because of the suspected role of stress the investigator decides to
examine the effect of relaxation training on the severity of asthma
symptoms. A sample of five patients is selected for the study. During the
week before treatment, the investigator records the severity of their
symptoms by measuring how many doses of medicine are needed for
asthma attacks. Then the patients receive relaxation training. For the week
following training; the researcher once again records the number of doses
required by each patient. Do these data indicate that relaxation training
alters the severity of symptoms.

Patient Week before training Week after training D D2


A 9 4 -5 25
B 4 1 -3 9
C 5 5 0 0
D 4 0 -4 16
E 5 1 -4 16
 D  16 D 2   66

 D 16
D   3.2
n 5

(  D) 2 (16)2
SS   D2   66 
n 5
= 66 - 51.2 = 14.8

SS 14.8
s   1.92
n 1 4

Next, use sample standard deviation to compute the estimated


standard error.

146
Self-instructional Material
s 192 The Random Sampling Distribution
SD    0.86
n 5

Finally, use the sample mean ( D ) and the hypothesized population


mean
along with the estimated standard error to compute the value for the ‗t‘ NOTES
statistic.

D   D 3.2  0
t   3.72
s .86
D

Locate the critical region from the ‗t‘ distribution table, you should
find that the critical values are +2.776. The t value we obtained falls in the
critical region the investigators rejects the null hypothesis and concludes
that relaxation training does affect the amount of medication needed to
control the asthma symptoms.

Check your progress – 2


3. When do we employ independent measures research design?
4. When is a t statistic preferred over z?

13.6 Choice of Alternate Hypothesis: One tailed and Two tailed test of
significance
The general goal of hypothesis testing is to determine whether a particular
treatment has any effect on a population. The test is performed by selecting
a sample, administering the treatment to the sample, and then comparing
the result with the original population. Generally sample should represent
the population and the extent to which sample statistics represent
population parameters is an indication of the significance of the computed
sample mean. If the treated sample is noticeably different from the original
population, we reject H0 and conclude that computed sample mean is
statistically significant then and that the treatment has an effect. On the
other hand, if the treated sample is still similar to the original population,
then we fail to reject H0 conclude that there is no convincing evidence for a
treatment effect, and. The critical factor in this decision is the size of the
difference between the treated sample and the original population. A large
difference is evidence that the treatment worked; a small difference is not
sufficient to say that the treatment had any effect.

One tailed tests of significance: In a directional hypothesis test, or a one-


tailed test, the statistical hypotheses (H0 and H1) specify either an increase
or a decrease in the population mean. That is, they make a statement about
the direction of the effect. For example, a special training program is
expected to increase student performance, or alcohol consumption is
expected to slow reaction times. In these situations, it is possible to state
the statistical hypotheses in a manner that incorporates the directional
147 Self-instructional Material
The Random Sampling Distribution

prediction into the statement of H0 and H1. The result is a directional test,
or what commonly is called a one-tailed test. The following example
NOTES demonstrates the elements of a one-tailed hypothesis test.

Two tailed tests of significance: The term two-tailed comes from the fact
that the critical region is divided between the two tails of the distribution.
This format is by far the most widely accepted procedure for hypothesis
testing. In case of two tailed test, we are interested if the difference sample
means obtained from two populations is likely or unlikely. We are
interested only in the magnitude of difference and not the direction of
difference. Consequently when an experimenter wishes to test the null
hypothesis, Ho : M1-M2 =0, against its possible rejection and finds that it is
rejected, he concludes that a difference really exist between the means. No
assertion is made about the direction of difference. Hence two tailed tests
are named non-directional tests of significance

The major distinction between one-tailed and two-tailed tests is the


criteria that they use for rejecting H0. A one-tailed test allows you to reject
the null hypothesis when the difference between the sample and the
population is relatively small, provided that the difference is in the
specified direction. A two-tailed test, on the other hand, requires a
relatively large difference independent of direction.

Check your progress – 3

5. What is directional test of significance?

6. What is the difference between one tailed and two tailed tests of
significance?

7. If a researcher predicts that a treatment will increase scores, then the


critical region for a one-tailed test would be located in the right-hand
tail of the distribution. (True or false?)

8. If the sample data are sufficient to reject the null hypothesis for a one-
tailed test, then the same data would also reject H0 for a two-tailed test.
(True or false?)

13.7 Let us sum up:

The distribution of sample means is defined as the set of Ms for all the
possible random samples for a specific sample size (n) that can be obtained
from a given population. The distribution of sample means is normal when
the population from which the samples are selected is normal and when the
size of the samples is relatively large (n = 30 or more).The mean of the
distribution of sample means is identical to the mean of the population
from which the samples are selected. The standard deviation of the
148
Self-instructional Material
distribution of sample means is called the standard error of M. The The Random Sampling Distribution

sampling distribution of means is then used to answer probability questions


about sample means. Proceeding on the same lines, if we had two different
samples taken from two different population, we would need to estimate
standard error of difference between the means and use the sampling
distribution of differences between means to answer probability questions NOTES
about the differences in sample means.

The ―t‖ statistic was introduced as an alternate to z score for


employing it in situation where the population standard deviation is
unknown to the researcher, which most often is the case. Testing the
significance of difference between means from two independent samples as
well as related samples have been highlighted. In addition, directional and
non –directional tests of significance have been presented. This shows that
appropriate statistics need to be employed keeping in mind the research
design of the study, for proper conclusions to be drawn about the
population taken for study.

13.8 Unit end exercises


1. What is ―t‖ distribution? How is it different from z distribution?

2. What do you understand by the term degrees of freedom?

3. In a study examining overweight and obese college football players,


Mathews and Wagner (2008) found that on average both offensive and
defensive linemen exceeded the at-risk criterion for body mass index
(BMI). BMI is a ratio of body weight to height squared and is commonly
used to classify people as overweight or obese. Any value greater than 30
kg/m2 is considered to be at risk. In the study, a sample of n =17 offensive
linemen averaged M =34.4 with a standard deviation of s = 4.0. A sample
of n =19 defensive linemen averaged M = 31.9 with s = 3.5. Use an
independent-measures t test to determine whether there is a significant
difference between the offensive linemen and the defensive linemen. Use a
two-tailed test with α= .01.

4. A major oil company would like to improve its tarnished image


following a large oil spill. Its marketing department develops a short
television commercial and tests it on a sample of
n = 7 participants. People‘s attitudes about the company are measured with
a short questionnaire, both before and after viewing the commercial. The
data are as follows:

Person Before (X1) After (X2)


A 15 15
B 11 13
C 10 18
D 11 12
149 Self-instructional Material
The Random Sampling Distribution

E 14 16
F 10 10
NOTES G 11 19

Was there a significant change? Note that participants are being tested
twice—once before and once after viewing the commercial. Therefore, we
have a repeated-measures design.

13.9 Answers to check your progress:

1. A sampling error is a statistical error that occurs when an analyst does


not select a sample that represents the entire population of data and
the results found in the sample do not represent the results that would
be obtained from the entire population. The standard error of
the sample mean is an estimate of how far the sample mean is likely
to be from the population mean. It is the standard deviation of the
distribution of sample means.

2. The distribution of sample means will take the shape of a normal


curve when n is large and when the population from which the sample
is drawn is normal.

3. The term independent refers to separate samples which are not related
to one another. Independent measures research design is used to make
a comparison between two groups of individuals. Example Boys Vs
Girls on numerical ability.

4. t statistic is preferred over z, when the value of population standard


deviation is unknown.

5. In a directional hypothesis test, the statistical hypotheses (H 0 and H1)


specify either an increase or a decrease in the population mean. That
is, they make a statement about the direction of the effect.

6. The major distinction between one-tailed and two-tailed tests is the


criteria that they use for rejecting H0.

7. True. A large sample mean, in the right-hand tail, would indicate that
the treatment worked as predicted.

8. False. Because a two-tailed test requires a larger mean difference, it is


possible for a sample to be significant for a one-tailed test but not for
a two-tailed test.

150
Self-instructional Material
The Random Sampling Distribution

13.10 Suggested Readings

1. Haslam,S.M.,&McGarty.C.(2003).Research Methods and statistics


in Psychology..Sage Publications Ltd.
NOTES
2. King,B.M., & Bear.G.(2001). Statistical Reasoning in Psychology
and Education(3rd Ed.). John Wiley& Sons Inc.

4. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural


sciences (9th Ed.). Cengage Learning India Private Ltd.

151 Self-instructional Material


Interpreting the results of UNIT 14: INTERPRETING THE
hypothesis testing

RESULTS OF HYPOTHESIS TESTING


NOTES
NOTES
14.1 A statistically significant difference versus practically important
difference.

A statistical significance does not necessarily mean that the results are
practically significant in a real-world sense of importance. The hypothesis
testing procedure determines whether the sample results that you obtain are
likely if we assume the null hypothesis is correct for the population. If the
results are sufficiently improbable under that assumption, then we can
reject the null hypothesis and conclude that an effect exists. In other words,
the strength of the evidence in our sample has passed our defined threshold
of the significance level (alpha) and hence the results are statistically
significant. We use p-values to determine statistical significance
in hypothesis tests , Consequently, it might seem logical that p-values and
statistical significance relate to importance. However, that is false because
conditions other than large effect size such as a large sample size and low
sample variability can produce tiny p-values.

As the sample size increases, the hypothesis test gains greater statistical
power to detect small effects. With a large enough sample size, the
hypothesis test can detect an effect that is so miniscule that it is
meaningless in a practical sense. When our sample data have low
variability, hypothesis tests can produce more precise estimates of the
population‘s effect. This precision allows the test to detect tiny effects.
Statistical significance indicates only that you have sufficient evidence to
conclude that an effect exists. It is a mathematical definition that does not
know anything about the subject area and what constitutes an important
effect. While statistical significance relates to whether an effect exists,
practical significance refers to the magnitude of the effect. However, no
statistical test can tell us whether the effect is large enough to be important
in your field of study. Instead, we need to apply our subject area
knowledge and expertise to determine whether the effect is big enough to
be meaningful in the real world.

For example, suppose you are evaluating a training program by comparing


the test scores of program participants to those who study on their own.
Further, we decide that the difference between these two groups must be at
least five points to represent a practically meaningful effect size. An effect
of 4 points or less is too small to care about. After performing the study,
the analysis finds a statistically significant difference between the two
Self-instructional Material 152
Interpreting the results of
hypothesis testing
groups. Participants in the study program score an average of 3 points
higher on a 100-point test. While these results are statistically significant,
the 3-point difference is less than our 5-point threshold. Consequently, our
study provides evidence that this effect exists, but it is too small to be
meaningful in the real world. The time and money that participants spend NOTES
on the training program are not worth an average improvement of only 3
points. We can use confidence intervals to determine practical significance.

Check your progress - 1


1. What is statistical significance based on?

14.2 Errors in hypothesis testing:


It is possible that the data will lead you to reject the null hypothesis when
in fact the treatment has no effect. Remember: Samples are not expected to
be identical to their populations, and some extreme samples can be very
different from the populations that they are supposed to represent. If a
researcher selects one of these extreme samples by chance, then the data
from the sample may give the appearance of a strong treatment effect, even
though there is no real effect.

Let‘s say a research study examines how a food supplement that is high in
antioxidants affects the cognitive functioning of elderly adults. Suppose
that the researcher selects a sample of n = 25 people who already have
cognitive functioning that is well above average. Even if the blueberry
supplement (the treatment) has no effect at all, these people will still score
higher than average on the neuropsychological test when they are tested
after 6 months of taking the supplement. In this case, the researcher is
likely to conclude that the treatment does have an effect, when in fact it
really does not. This is an example of what is called a Type I error.

Type I error occurs when a researcher rejects a null hypothesis that is


actually true. In a typical research situation, a Type I error means that the
researcher concludes that a treatment does have an effect when, in fact, it
has no effect. The problem is that the information from the sample is
misleading. In most research situations, the consequences of a Type I error
can be very serious. Because the researcher has rejected the null hypothesis
and believes that the treatment has a real effect, it is likely that the
researcher will report or even publish the research results. A Type I error,
however, means that this is a false report. Thus, Type I errors lead to false
reports in the scientific literature. Other researchers may try to build
theories or develop other experiments based on the false results. A lot of
precious time and resources may be wasted. The Probability of a Type I
Error occurs when a researcher unknowingly obtains an extreme, non-
representative sample. Specifically, the probability of a Type I error is
equal to the α (alpha) level. Occasionally, however, sample data are in the
153 Self-instructional Material
critical region just by chance, without any treatment effect. When this
Interpreting the results of
hypothesis testing
occurs, the researcher makes a Type I error; that is, the researcher
concludes that a treatment effect exists when in fact it does not.
NOTES Fortunately, the hypothesis test is structured to minimize the risk that this
NOTES will occur.

Type II error occurs when a researcher fails to reject a null hypothesis


that is really false. In a typical research situation, a Type II error means
that the hypothesis test has failed to detect a real treatment effect. A Type
II error occurs when the sample mean is not in the critical region even
though the treatment has had an effect on the sample. Often this happens
when the effect of the treatment is relatively small. In this case, the
treatment does influence the sample, but the magnitude of the effect is not
big enough to move the sample mean into the critical region. Because the
sample is not substantially different from the original population (it is not
in the critical region), the statistical decision is to fail to reject the null
hypothesis and to conclude that there is not enough evidence to say that
there is a treatment effect. The consequences of a Type II error are usually
not as serious as those of a Type I error. In general terms, a Type II error
means that the research data do not show the results that the researcher had
hoped to obtain. Unlike a Type I error, it is impossible to determine a
single, exact probability for a Type II error. Instead, the probability of a
Type II error depends on a variety of factors and therefore is a function,
rather than a specific number.

Check your progress – 2


2. What is the relationship between alpha level and Type I error?
3. Under what circumstances is a Type II error likely?

14.3 Statistical power

An approach to determine the size or strength of a treatment effect is to


measure the power of the statistical test. The power of a test is defined as
the probability that the test will reject the null hypothesis if the treatment
really has an effect. The power of a statistical test is the probability that the
test will correctly reject a false null hypothesis. That is, power is the
probability that the test will identify a treatment effect if one really exists.
However, the second outcome, rejecting H0 when there is a real effect, is
the power of the test. Researchers typically calculate power as a means of
determining whether a research study is likely to be successful. Thus,
researchers usually calculate the power of a hypothesis test before they
actually conduct the research study. In this way, they can determine the
probability that the results will be significant (reject H 0) before investing
time and effort in the actual research. To calculate power, however, it is
first necessary to make assumptions about a variety of factors that

Self-instructional Material 154


Interpreting the results of
hypothesis testing
influence the outcome of a hypothesis test. Factors such as the sample size,
the size of the treatment effect, and the value chosen for the alpha level can
all influence a hypothesis test.

14.4 Statistical decision making: Levels of Significance versus p –


values
NOTES

Levels of significance: To find the boundaries that separate the high-


probability samples from the low-probability samples, we must define
exactly what is meant by ―low‖ probability and ―high‖ probability. This is
accomplished by selecting a specific probability value, which is known as
the level of significance, or the alpha level, for the hypothesis test. The
alpha (α) value is a small probability that is used to identify the low
probability samples. By convention, commonly used alpha levels are α =
.05 (5%), α = .01 (1%), and α=.001 (0.1%). For example, with α = .05, we
separate the most unlikely 5% of the sample means (the extreme values)
from the most likely 95% of the sample means (the central values). The
extremely unlikely values, as defined by the alpha level, make up what is
called the critical region. These extreme values in the tails of the
distribution define outcomes that are not consistent with the null
hypothesis; that is, they are very unlikely to occur if the null hypothesis is
true. Whenever the data from a research study produce a sample mean that
is located in the critical region, we conclude that the data are not consistent
with the null hypothesis, and we reject the null hypothesis.

The alpha level, or the level of significance, is a probability value that is


used to define the concept of ―very unlikely‖ in a hypothesis test. The
critical region is composed of the extreme sample values that are very
unlikely (as defined by the alpha level) to be obtained if the null hypothesis
is true. The boundaries for the critical region are determined by the alpha
level. If sample data fall in the critical region, the null hypothesis is
rejected. Technically, the critical region is defined by sample outcomes
that are very unlikely to occur if the treatment has no effect (that is, if the
null hypothesis is true). Reversing the point of view, we can also define the
critical region as sample values that provide convincing evidence that the
treatment really does have an effect.

The boundaries for the critical region: To determine the exact location
for the boundaries that define the critical region, we use the alpha-level
probability and the unit normal table. In most cases, the distribution of
sample means is normal, and the unit normal table provides the precise z-
score location for the critical region boundaries. With α =.05, for example,
the boundaries separate the extreme 5% from the middle 95%. Because the
extreme 5% is split between two tails of the distribution, there is exactly
2.5% (or 0.0250) in each tail. Thus, for any normal distribution, the
extreme 5% is in the tails of the distribution beyond z = +1.96 and z = -
155 Self-instructional Material
1.96. These values define the boundaries of the critical region for a
Interpreting the results of
hypothesis testing
hypothesis test using α=.05. Similarly, an alpha level of α=.01 means that
1%, or .0100, is split between the two tails. In this case, the proportion in
NOTES each tail is .0050, and the corresponding z-score boundaries are z = + 2.58
NOTES and z= - 2.58. Below is a figure showing the critical regions

p-value : A p-value is a probability. It is usually expressed as a proportion


which can also be easily interpreted as a percentage:
 P = 0.50 represents a 50% probability or a half chance.
 P = 0.10 represents a 10% probability or a one in ten chance.
 P = 0.05 represents a 5% probability or a one in twenty chance.
 P = 0.01 represents a 1% probability or a one in a hundred chance.
P-values become important when we are looking to ascertain how
confident we can be in accepting or rejecting our hypotheses. Because we
only have data from a sample of individual cases and not the
entire population we can never be absolutely (100%) sure that
the alternative hypothesis is true. However, by using the properties of
the normal distribution we can compute the probability that the result we
observed in our sample could have occurred by chance. To clarify, we can
calculate the probability that the effect or relationship we observe in our
sample could have occurred through sampling variation and in fact does
not exist in the population as a whole. The strength of the effect (the size of
the difference between the mean scores), the amount of variation in scores
(indicated by the standard deviation) and the sample size are all important
in making the decision.

Conventionally, where there is less than a 5% probability that the results


from our sample are due to chance the outcome is considered statistically
significant. Another way of saying this is that we are 95% confident there
is a ‗real‘ difference in our population. This is our confidence level. We are
therefore looking for a p-value that is less than .05, commonly written as p
<.05. Results significant at the 1% level (p<.01), or even the 0.1% level

Self-instructional Material 156


Interpreting the results of
hypothesis testing
(p<.001), are often called "highly" significant and if we want to be more
sure of our conclusions you can set your confidence level at these lower
values. It is important to remember these are somewhat arbitrary
conventions - the most appropriate confidence level will depend on the
context of the study. NOTES

The way that the p-value is calculated varies subtlety between different
statistical tests, which each generate a test statistic (called, for example, t,
F or X2 depending on the particular test). This test statistic is derived from
your data and compared against a known distribution (commonly a normal
distribution) to see how likely it is to have arisen by chance. If the
probability of attaining the value of the test statistic by chance is less than
5% (p<.05) we typically conclude that the result is statistically significant.

Check your progress - 4

4. When do we reject a null hypothesis?


5. What is the need for a power test?
6. If a sample mean is in the critical region with α = .05, it would still
(always) be in the critical region if alpha were changed to α = .01. (True
or false?)

14.5 Let us sum up:

In this chapter we introduced chief ideas relating to interpretation of a


hypothesis test. It is worthy to note that a statistically significant difference
may not have any practical significance. Whatever decision is reached in a
hypothesis test, there is always a risk of making the incorrect decision.
There are two types of errors that can be committed A Type I error is
defined as rejecting a true H0. This is a serious error because it results in
falsely reporting a treatment effect. The risk of a Type I error is determined
by the alpha level and, therefore, is under the experimenter‘s control. A
Type II error is defined as the failure to reject a false H0. In this case, the
experiment fails to detect an effect that actually occurred. The probability
of a Type II error cannot be specified as a single value and depends in part
on the size of the treatment effect. The power of a hypothesis test is
defined as the probability that the test will correctly reject the null
hypothesis.

14.6 Unit End exercises:

1. Write short notes on errors in hypothesis testing with the


suitable example.
2. What are probability values and how do you interpret them?

157 Self-instructional Material


14.7 Answers to check your progress
Interpreting the results of
hypothesis testing
1. Statistical significance is the likelihood that a relationship between
two or more variables is caused by something other than chance
NOTES and is based on probability.
NOTES
2. The alpha levelfor a hypothesis test is the probability that the test
will lead to a Type I error. That is, the alpha level determines the
probability of obtaining sample data in the critical region even
though the null hypothesis is true

3. Type II error is likely when the effect of the treatment is relatively


small. That is, the magnitude of the effect is not big enough to
move the sample to the critical region.

4. We reject the null hypothesis when we find that our sample mean is
in the critical region.

5. Power analysis is normally conducted before the data


collection. The main purpose underlying power analysis is to help
the researcher to determine the smallest sample size that is suitable
to detect the effect of a given test at the desired level of
significance. The reason for applying power analysis is that,
ideally, the investigator desires a smaller sample because larger
samples are often costlier than smaller samples.

6. False. With α= .01, the boundaries for the critical region move
farther out into the tails of the distribution. It is possible that a
sample mean could be beyond the .05 boundary but not beyond the
.01 boundary.
14.8 Suggested Readings

1. Gravetter,F.J., &Wallnau,L.B.(2013). Statistics for behavioural sciences


(9th Ed.). Cengage Learning India Private Ltd.
2. Aron, Aron., E. N., Coups , E.(2012). Statistics for Psychology.(3rd Ed).
Prentice Hall.
3. Garrett.H.E.(1981). Statistics in Psychology and Education. Vakils,
Feffer and Simons Ltd.

Self-instructional Material 158


Model Question Paper
Distance Education – CBCS – (2018-19 Academic Year
Onwards)
MODEL QUESTION PAPER
B.Sc(Psychology).,
11933 - PSYCHOLOGICAL STATISTICS
Time : 3 Hours
Marks :75
PART – A (10X 2 = 20 Marks)

I. Answer all questions.

1 What is a continuous variable?


2. What is a histogram?
3. Give the z score formula.
4. When do you use the mean?
5. What do you mean by the term Statistical inference?
6. Define Standard Error
7. Define Probability
8. Define variance
9. What is Contingency co-efficient?
10. Define ANOVA.
PART – B (5X 5 = 25 Marks)

II . Answer all questions choosing either (a) or (b).

11. a) Explain the importance of statistics.


or
b) Construct a frequency distribution table from the scores
given below:
18 29 0 1 4 6 2 0
11 10 8 21 17 8 5 6
15 21 19 17 16 13 14 1
24 25 22 21 22 28 27 26

12. a) Compute Pearson‘s product moment correlation for the


following data.
X- 10, 12, 14, 14, 16
Y- 18, 20. 22. 17, 18 Compute mean and median from the
following data.

Scores Frequency
14-15 3
12-13 8
10-11 15
8-9 20
6-7 10
4-5 4
159 Self-instructional Material
13. a) Explain the process of hypothesis testing.
Model Question Paper
or
b) . Compute Range, Average deviation, Standard deviation
and Variance for the following data.

17, 15, 13, 15, 16, 14, 12, 18.

14. a) Explain scales of measurement with suitable examples.


or

b) List statistical software packages that are useful in


analysing quantitative and qualitative research data

15. a) Explain one-tailed and two-tailed tests of significance.


Or
b) Explain errors in hypothesis testing

PART – B (3X10 = 30 Marks)

III. Answer any 3 out of 5 questions.

16. Explain Chi- square as a goodness of fit.

17. Give the null hypothesis for simple regression. How would you
interpret the results of the simple regression?

18. A developmental psychologist is examining problem solving


ability of grade school children. Each child is given a
standardized problem solving task, and the psychologist records
the number of errors. Use the data given to test whether there is
a significant differences among the three age groups.

5 year olds 6 year olds 7 year olds


5 6 0
4 4 1
6 2 2

19. Explain measures of asymmetry.

20. a) At Smart university , all new freshmen are given an English


proficiency test during the week of registration. Scores are
normally distributed with the mean of 70 and SD of 10. The
university has decided to place the top 25% into honors English and
the bottom 20 % into remedial English. What scores separate the
upper 25% and lower 20% of the students from the rest?

b) In a normal distribution of 1000 aptitude test scores with the


mean of 60 and standard deviation of 8, How many scores fall i)
Above 76 ii) Below 80 iii) Between 48 and 52.
160

You might also like