Psychological Statistics
Psychological Statistics
Psychological Statistics
(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)
B.Sc. (PSYCHOLOGY)
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
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
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
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
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
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
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 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:
(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:
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.
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
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.
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
5, 7, 4, 5, 6, 5, 4
10
Self-instructional Material
Frequency Table Statistics
d. Place the interval containing the highest score value at the top
g. When possible make the lower score limits multiples of the interval
width
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.
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.
----
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.
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.
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
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.
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
17 Self-instructional Material
Diagrammatic and
Graphical
NOTES
18
Self-instructional Material
Diagrammatic and
Graphical
NOTES
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.
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
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.
Shows the distribution of sales of the laptop industry between five companies
22
Self-instructional Material
Diagrammatic and
Graphical
NOTES
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.
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.
∑ 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
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
N = Total frequency
X A
x' =
i
--------- ---------
---
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.
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.
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.
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.
The median is 11
This again shows that the position of each score in the series needs to be
ascertained first and then median should be determined.
( N / 2) F
Mdn L i
f
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
---------
50 / 2 15
Mdn 395 5
11
10
395 5 44.05
11
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
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.
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.
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.
F 1 2 7 5 3 2
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
Harmonic means are often used in averaging things like rates (e.g.,
the average travel speed given aduration of several trips)
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.
Self-instructional Material 34
UNIT 4: MEASURES OF DISPERSION Measures Of Dispersion
4.5 Introduction
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.
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.
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
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.
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.
( 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
x2
N
Self-instructional Material 38
52, 50, 56, 68, 65, 62, 57, 70 Measures Of Dispersion
382
8
47.75 6.91
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
We will calculate the variance and standard deviation for the following
population of N = 5 scores: 1, 9, 5, 8, 7
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)
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.
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.
Self-instructional Material 42
4.6 Concept of Skewness and Kurtosis: Measures Of Dispersion
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.
45 Self-instructional Material
Measures Of Dispersion 4.10 Suggested readings:
6. Gupta, S.P. (2014). Statistical Methods. (44the.d.) New Delhi
:Roopak Printers.
Self-instructional Material 46
Correlation
UNIT 5: CORRELATION
5.1: Introduction to Correlation NOTES
5.1.1: Characteristics of relationship
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.
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.
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
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?)
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
Mean of series x, x = 25
Mean of series y, y = 50
xy 250
r =
x 2 y 2 250 1000
= 0.5
51 Self-instructional Material
Correlation
5.3.2 Spearman’s Rank-Order correlation
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
6d 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
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 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.
Self-instructional Material 54
Correlation
NOTES
Figure: The X and Y data points and the regression line for the n = 8
pairs of scores
distance = Y- 𝑌
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.
𝑆𝑃
b=r
𝑆𝑆𝑥
𝑆𝑌
b=r
𝑆𝑥
a = My – bMx
𝑌 = bX + a
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
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
𝑌= X + 4
\
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
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.
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
X 2 1 3 2 4
Y 5 6 3 7 4
59 Self-instructional Material
Correlation
2.True. The numerical value indicates the strength of the relationship. The
sign only indicate the direction of the relationship.
4. The Pearson correlation measures the degree and direction of the linear
relationship between two variables.
6.Y= 2X + 25
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,
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) .
.
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.
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.
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
a)-0.00001
b)0.5
c)0
e) 1
6.4.1: Union/Addition:
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
A = Spade Ace
B = an Ace
A ∩ B = Spade Ace
P(A) = 1/52
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
66
Self-instructional Material
Concept of Probability
Example 1: Drawing a Spade Ace
A = an Ace NOTES
B = a Spade
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?
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
2. of different colors
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.
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
69
Self-instructional Material
Normal distribution
UNIT 7: NORMAL DISTRIBUTION
1. For this curve, mean, median and mode are the same.
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.
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.
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.
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.
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.
Quantile deviation
Kurtosis
90th percentile 10th percentile
74
Self-instructional Material
Normal distribution
1. Used as a Model
(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.
3. Ability grouping
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
75
Self-instructional Material
Normal distribution
Some illustrations of the applications of the normal curve
X1 M1 80 70
Z score in Maths = 0.5
1 20
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.
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.
77
Self-instructional Material
Normal distribution
NOTES
M - 0.525 or
80 - 0.525 16 = 71.6 or 72
78
Self-instructional Material
Normal distribution
NOTES
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.
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.
Solution:
80
Self-instructional Material
Normal distribution
of the mean as covered by the groups B and A on the right side of the
mean.
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.
81
Self-instructional Material
Normal distribution
7.6 Unit – End exercises
Mathematics Physics
NOTES M 60 33
S.D 8 9
Achievement scores 70 67
of students
4. Bring out the utility of the normal curve in the field of psychology.
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
83
Self-instructional Material
Normal distribution
8.2 The Binomial Distribution
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
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
86
Self-instructional Material
Normal distribution
for a hypothesis test: Do the sample data provide sufficient evidence
to make a conclusion about the population?
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
1
H0: p = p(heads) = (The coin is balanced.)
2
1
H0 – P = p (guessing correctly) = (The outcome is
4
1
H0 : For 16-year-olds, p = p(accident) = (Not different
12
H0: For this year, p = p(fail) = 30% (Not different from last year's
class)
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:
= 𝑛𝑝𝑞
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.
1 1
H0 :
p = p(deep side) = 𝑎𝑛𝑑 𝑞 =
2 2
1
H1 : p ≠ (There is a preference.)
2
We use α = .05
91
Self-instructional Material
Normal distribution
𝑋−𝑝𝑛 3−13.5 − 10.5
z = = = = - 4.04
𝑛𝑝𝑞 27
1 1 2.60
2 2
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.
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.
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.
5.False.
8.7Suggested Readings
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
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.
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.
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
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.
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.
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.
or
Hence the procedure for the analysis of variance involves the following
main tasks:
X = Grand Sum
X
= Grand or general mean
N
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
X 1 X 2 X
, ,...
(ii) Group means, N N , and the grand mean N
(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:
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 :
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 :
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.
1 1
D
n1 n2
Where
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.
Solutions
Organization of Data
1 28
4
Group means = n1 7
NOTES
2 35
5
n2 7
3 33
4.71
n3 7
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 ) :
28 28 35 35 33 33
438.85
7 7 7
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
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
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
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
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).
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).
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
10.19
8.56
1.19
1.86
1.56
1.19
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.
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
( 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.
Prior to using the chi square test there are certain requirements that must be
met. These are :-
Null hypothesis for Chi square goodness of fit :Generally H0 falls into
NOTES one of the following categories:
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.
in this formula:
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.
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
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
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
8. Why do you think t test is more powerful than the sign test?
Do the data indicate any significant preferences among the four colors?
Test at the .05 level of significance.
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.
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:
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.
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
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 (‗.‘)
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.
While entering data into a software, the variables are entered in rows:
True/False
How do you enter missing data into your data sheet?
123
Self-instructional Material
Preparation of Data for Computer
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.
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:
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. False
5. Excel does not have built-in data manipulation tools such as recoding,
and transforming variables like SPSS.
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.
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
128
Self-instructional Material
Inferential Statistics
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.
(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.
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
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
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.
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
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:
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.
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
134
Self-instructional Material
Inferential Statistics
4. False
7. False
8. True
136
Self-instructional Material
we introduce a procedure for transforming a sample mean into a z-score. The Random Sampling Distribution
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
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.
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.
(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).
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.
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.
140
Self-instructional Material
13.4 Introduction to t statistics : The Random Sampling Distribution
x (standard error of the mean) = (Standard deviation of
n
the population)
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
142
Self-instructional Material
distribution because with greater df, S will more clearly estimate The Random Sampling Distribution
X
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
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.
Group I: 10, 9, 8, 7, 7, 8, 6, 5, 6, 4
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
30 44
2.03
99
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
X M
t
s
x
X= Sample mean
M = Population Mean
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
Illustration:
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
146
Self-instructional Material
s 192 The Random Sampling Distribution
SD 0.86
n 5
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.
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.
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
6. What is the difference between one tailed and two tailed tests of
significance?
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?)
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
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.
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.
7. True. A large sample mean, in the right-hand tail, would indicate that
the treatment worked as predicted.
150
Self-instructional Material
The Random Sampling Distribution
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.
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.
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
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.
4. We reject the null hypothesis when we find that our sample mean is
in the critical region.
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
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. Give the null hypothesis for simple regression. How would you
interpret the results of the simple regression?