NONPARAMETRIC TEST

Dr.dr. SISWANTO, MSc

FACULTY OF MEDICINE
UNIVERSITY OF BRAWIJAYA

1
 INTRODUCTION

Critical reading of the literature requires the

capability to determine whether the
conclusions are supported by the data
presented.

Part of this determination involves deciding

whether the results are statistically valid.

2
 A variable is a
characteristic of the
thing you are studying
or observing
Length of …
Weight of …
Variables are characteristics
that can be measured or Number of … Amount of …
categorized 3
What is the variable
if you are studying
the birth weight of
dolphins?
birthweight

Dolphins are
the subjects

20 4
 Ratio
Numeric
Interval
Variable
Scale
Nominal
Categorical
Ordinal
Nominal variable

Variables that are “named”, i.e. classified into one

or more qualitative categories that describe the
characteristic of interest

• no ordering of the different categories

• no measure of distance between values

6
Nominal variable

In medicine, nominal variables are often used to

describe the patient. Examples of nominal
variables might include:
• Gender (male, female)
• Eye color (blue, brown, green, hazel)
• Surgical outcome (dead, alive)
• Blood type (A, B, AB, O)

7
Ordinal variable

Variables that have an inherent order to the

relationship among the different categories

• an implied ordering of the categories (levels)

• quantitative distance between levels is unknown
• distances between the levels may not be the same
• meaning of different levels may not be the same for
different individuals

8
Ordinal variable

In medicine, ordinal variables often describe the

patient’s characteristics, attitude, behavior, or
status. Examples of ordinal variables might
include:

• Stage of cancer (stage I, II, III, IV)

• Education level (elementary, secondary, college)
• Pain level (mild, moderate, severe)
• Satisfaction level (very dissatisfied, dissatisfied,
neutral, satisfied, very satisfied)
• Agreement level (strongly disagree, disagree,
neutral, agree, strongly agree) 9
Numeric variable

Variables have equal intervals between values,

the zero point is meaningful, and the numerical
relationships between numbers is meaningful.
Examples of ratio variables:

• Weight (50 kilos, 100 kilos, 150 kilos, etc.)

• Pulse rate
• Respiratory rate

10
Level of measurement
• Higher level variables can always be expressed at a
lower level, but the reverse is not true.
• For example, Body Mass Index (BMI) is typically
measured at an interval-level such as 23.4.
– BMI can be collapsed into lower-level Ordinal
categories such as: (ordinal)
• >30: Obese
• 25-29.9: Overweight
• <25: Underweight

or Nominal categories such as: (nominal)

• Overweight
• Not overweight
11
Nonparametric tests, or distribution free tests are
so-called because the assumptions underlying their
use are “fewer and weaker than those associated
with parametric tests” (Siegel& Castellan, 1988)

Parametric tests are more robust and for the most

part require less data to make a stronger conclusion
than nonparametric tests.

12
 PARAMETRIC TEST

Three parameters of the data must be true or are

assumed
1. The data need to be normally distributed (The
observations in each group are normally
distributed)
2. The variance of each group is equal to that of
any group (homogeneity of variances)
3. The data need to be continuous (interval/ratio)
13
If the data do not meet the criteria for a
parametric test (normally distributed, equal
variance, and continuous), it must be
analyzed with a nonparametric test.

14
 CHI SQUARE
The chi-squared test is usually used to compare
multiple groups where the independent variable
and the dependent variable are binary
(nominal/qualitative/categorical data)
This test is very well illustrated with an example of
a 2X2 table of data comparing 2 groups, one
receiving a treatment and the other not, and 2
outcomes of cure versus continued disease (not
cured)

Test of significance of the difference between two /more15

proportions
 Table . Sputum conversion after one month of treatment in
the vitamin D and placebo groups

Variables Sputum conversion Total p value

Yes No

Placebo 8 (36%) 14 (64%) 22

Vitamin D 14 (67%) 7 (33%) 21 0,04

Total 22 21 43

16
 FISHER EXACT TEST

Along the same lines as the chi-squared test is

Fisher exact
test.
This test is complicated to calculate but is used if
the table of expected outcomes has any single value
of less than 2 or if more than 20% of the values are
less than 5.

18
 LUNG CA + LUNG CA -

SMOKER a+b
a b

c d
NONSMOKER
c+d

a+c b+d

19
 How to calculate of expected frequency/outcomes

 a = (a+c) X (a+b) / N

 b = (b+d) X (b+a) / N

 c = (c+a) X (c+d) / N

 d = (d+c) X (d+b) / N

 N = a+b+c+d

 Degree of fredom (r-1) (c-1) 20

 MANN-WHITNEY U TEST

The counterpart nonparametric test to independent

t test (parametric test)

It is analogous to the independent t-test for

continuous variable but can be used for ordinal
data. This test compares 2 independent populations
to determine whether they are different.

21
To carry out these two test, we proceed as follows:

1. Combine the observations from both samples

and arrange them in an array from the smallest to
the largest
2. Assign ranks to each of the observations
3. List the ranks from one sample separately from
those of the other
4. Separately sum the ranks for the first and second
sample

22
Mann-Whitney Test for two independent samples: Number
of prenatal care visits for mothers bearing babies of low and
of normal birth weight

23
 WILCOXON SIGNED-RANK TEST

The counterpart nonparametric test to paired t test

(parametric test)

It is analogous to the paired t-test for continuous

variable but can be used for ordinal data.

If the null hypothesis is true, the sum of the

positive ranks and the sum of the negative ranks
are expected to be roughly equal.
24
Wilcoxon Signed-Rank Test: Number of cigarettes usually
per day. Before and after pregnancy

25
 KRUSKAL-WALLIS TEST
The Kruskal-Wallis Test is the nonparametric equivalent
of the one-way ANOVA.
This technique is an alternative to the one-way ANOVA
• three or more group
• the groups are independent
• the samples are selected are not normally distributed
• the sample do not have equal variances
• data scale is ordinal

26
To use the Kruskal-Wallis Test
• Combine the observation of the various groups
and arrange them in array from lowest to highest
• Assign ranks to each of the observations and
replace them in each of the groups ( if you start
with ordinal data, this conversion would not be
necessary
• The ranks are summed in each of the groups, and
the test statistic H is computed

27
Scores of three types of Ranks
teachers

28
 SPEARMAN RANK CORRELATION
COEFFICIENT
The sample correlation coefficient, r, is a measure of the
strength of the linier association between two quantitative
variables ((Pearson’s product-moment r)
Spearman Correlation is a measure of the association
between two qualitative data (two rank variable/ordinal)
Like the Pearson product correlation coefficient, the
Spearman rank coefficient is calculated to determine how
well 2 variables for individual data points can predict each
other.

29
 SPEARMAN RANK CORRELATION
COEFFICIENT

Similar to the Pearson correlation coefficient, the

test statistic is from -1 to 1, with -1 being a perfect
negative correlation and 1 a perfect positive
correlation.

30
31
32
 FRIEDMAN TEST

This test is sometimes called the Friedman two-

way analysis of variance by ranks.
The Friedman Test is the nonparametric
equivalent of the Repeated Measure ANOVA.
It is for use with k repeated (or correlated)
measures where measurement is at least ordinal.
The null hypothesis states that all k samples are
drawn from the same population, or from
populations with equal medians.
33
Example
The table on the left (below) shows reaction time
data from 5 subjects, each of whom was tested in
3 conditions (A, B, and C).
The Friedman ANOVA uses ranks, and so the first
thing we must do is rank order the k scores for
each subject.
The results of this ranking are shown in the table
on the right, and the sum of the ranks (ΣRi ) for
each treatment is shown at the bottom.

34
35
 Multiple Logistic Regression
 Logistic Regression is used to analyze the relationship
between two variables when the dependent variable is
binary . This differs from normal simple linear regression
where the dependent variable is a continuous numeric
variable. A simple logistic regression model is by
 logit(pi) = log(pi / (1 pi)) = b0 + b1* xi
 where
 pi is the response to be modeled
b0 is the intercept parameter
b1 is the slope parameter
xi is an array of independent variables
36
 Multiple Logistic Regression
 Use multiple logistic regression when you have one
nominal and two or more measurement /independent
variables. The nominal variable is the dependent (Y)
variable; you are studying the effect that the independent
(X) variables have on the probability of obtaining a
particular value of the dependent variable. For example,
you might want to know the effect that blood pressure, age,
and weight have on the probability that a person will have
a heart attack in the next year.

37
 Multiple Logistic Regression

Multiple Logis
ln[Y/(1−Y)]=a+b1X1+b2X2+b3X3...

38
 CONTOH
 Penelitian untuk mengetahui faktor
prediktor terjadinya syok pada penderita
demam berdarah. Adapun faktor prediktor
yang akan diteliti meliputi : jenis kelamin,
perdarahan, trombosit, hematokrit dan
hepatomegali

39
 Example
 Penelitian untuk mengetahui faktor
prediktor terjadinya syok pada penderita
demam berdarah. Adapun faktor prediktor
yang akan diteliti meliputi : jenis kelamin,
perdarahan, trombosit, hematokrit dan
hepatomegali

40
 Research to determine the predictor factor
of shock in dengue fever patients. The
predictor factors to be studied include: sex,
bleeding, thrombocyte, hematocrit and
hepatomegaly

41
42
logit(pi) = log(pi / (1 pi)) = b0 + b1* xi
Y= Constanta + a1x1+a2x2+a3x3……+aix1

Y=1,538+1,189(perdarahan)+
1,233(hepatomegali)+1,137(hematokrit)

P= 1/(1+e-Y )
P=probabilitas
e= bilangan natural = 2,7

43
GOOD LUCK !

44