NON-PARAMETRIC TEST OF

DIFFERENCE

SALVADOR A. MIRALLES
Non-parametric statistics can be applied in a
wide variety of research situations.
It is sometimes called Distribution-free Methods
because the inferences are
based on a test statistics whose sampling
distribution does not depend on
the specific distribution of the population from
which the sample is drawn.
 Meaning of non-parametric Statistics

Non-parametric statistics is a method of hypothesis

testing and estimation that has certain desirable
properties that hold under relatively mild assumptions
regarding the underlying populations from which the
data are obtained.
This nonparametric statistics usually labeled as a test
statistics is more appropriate for a small sample size. It
is also used for the data when there are extreme values
in the given data. The strength of nonparametric
statistics can compare favorable with parametric
statistics and in the case where all the assumptions are
not met for the application of the latter approach, non
parametric can provide a support result.
The data may categorized into two types
1. categorical data
2. continuous data

1. Categorical data are nominal

and ordinal (non-parametric)

2. Continuous data are interval

and ratio (parametric)
 Levels of Measurement
A. Nominal Level

1. first level of measurement

2. characterized by data that consist of names,
labels or categories only
3. cannot be arranged in ordering scheme
4. have no numerical value
5. sometimes called categorical scales or
categorical data
6. classifies persons or objects into two or more
categories
Examples of Data in Nominal Level

[Link] (Male, Female)

[Link] Color (Blue, Green, Brown, Black)
[Link] of House (Bungalow, duplex, Ranch)
4. Employment Status (full-time, part-time,
unemployed)
5. Educational attainment (elementary
graduate, high school graduate, bachelor’s
degree holder. master’s degree holder,
doctorate degree holder)
6. Type of School (public, private)
Examples of Survey Questions that use nominal data

1. What is the teacher’s gender? Encircle one.

Male 1
Female 2

2. Describe your current employment status? Encircle

one.
Full-time 1
Part-time 2
Not currently employed 3
Retired 4

3. Indicate your current marital status. Encircle one.

Single (never married) 1
Married 2
Single (widowed, separated) 3
 B. Ordinal Level
Examples of Data in Ordinal Level
1. High School Class Ranking (1st, 2nd, 3rd, etc.)
2. Social Economic Class (First, Middle, Lower)
3. Pain level ( none, mild, moderate,, severe
4. Comparison ( greater than, less than or equal )
5. Attitude toward s subject
6. Social status
7. Rubrics
Examples of Survey Questions that use ordinal scale

1. How much education have you completed? Encircle one.

College graduate 1
With some units in college 2
High school graduate 3
Never finished high school 4
[Link] is your latest student evaluation? Encircle
one.
Outstanding 1
Very satisfactory 2
Satisfactory 3
Fair 4
Poor 5
3. How often during the last month did you find
yourself under stress? Encircle one.
Always 5
Very Often 4
Fairly Often 3
Sometimes 2
Almost never 1
 The main reason we should distinguish
between them is that the method used is
different depending on the type of data.
Categorical data use NON-PARAMETRIC
STATISTICS while Continuous data use
PARAMETRIC STATISTICS.
 Mann-Whitney Test
The Mann-Whitney test is non-parametric
alternative to the t-test of two independent
samples which require only the assumption of
any continuous and at least the data measured
on an ordinal level of measurement. The test
does not require that the samples, which are
to be compared, be of the same size. The
interpretation of the test is essentially
identical to the interpretation of the
t- statistic for independent samples.
 Wilcoxon Signed-Ranks Test
Wilcoxon signed-ranks test is the non-
parametric alternative to the t-
statistics of dependent samples which
require only the assumption of any
continuous and at least an ordinal level
of measurement scales. The
interpretation of the test is similar to
that of the t-test for dependent
samples. The test statistic used when
there are no or few ties and n<50
 Kruskal-Wallis Test
Kruskal-Wallis test is an extension of a
Mann-Whitney test. It is a nonparametric
alternative to the one-way analysis of
variance. Thus, sometimes it is called as
Kruskal-Wallis One-Way Analysis of Variance
by Rank. This test requires that the data be
measured on an ordinal scale of
measurement and assumes only continuous
distributions.
 Friedman Test
Friedman test is an extension of
the Wilcoxon signed ranks test. It is a non-
parametric alternative to the two-way
analysis of variance. Hence, it is sometimes
called as Friedman two-way analysis of
variance by ranks. The test requires data
measured on the at least an ordinal level of
measurement and have been drawn from
any continuous populations, not necessarily
normal distributions.
 Non-PARAMETRIC STATISTICS
1. Mann-Whitney Test- difference
( alternative for t-test) - difference
2. Wilcoxon Signed-Ranks Test
( alternative for t-test) - difference
3. Kruskal-Wallis Test
( alternative for one-way ANOVA ) - difference
4. Friedman Test
( alternative for two way ANOVA )- difference
 PARAMETRIC NON-PARAMETRIC
DIFFERENCE
STATISTICS STATISTICS

   

   

t- test Mann-Whitney Test

t-test Wilcoxon Signed-Rank Test

One-Way ANOVA Kruskall-Wallis Test

Two-Way ANOVA Friedman Test

 INDEPENDENT
   

NOMINAL ORDINAL CONTINUOUS

NOMINAL CHI SQUARE CHI SQUARE TWO NOMINAL

TEST TEST GROUPS : T- TEST
DEPENDENT

CHI SQUARE CHI SQUARE TEST OR SPEARMAN'S

ORDINAL
TEST SPEARMAN'S rHO
rHO

T-TEST (2 SPEARMAN'S
CONTINUOUS PEARSON'S r
GROUPS) + rHO
Cohen'S D
 PARAMETRIC NON-PARAMETRIC
DIFFERENCE
STATISTICS STATISTICS

   

   

t- test Mann-Whitney Test

t-test Wilcoxon Signed-Rank Test

One-Way ANOVA Kruskall-Wallis Test

Two-Way ANOVA Friedman Test

 PARAMETRIC
ORDINAL DATA NOMINAL DATA
STATISTICS

   

   

MEAN MEDIAN MODE

 TRY TO ANSWER
1. Pre – test and post test scores were
obtained from a group of 10 high
school senior students. The pre-test
scores were collected before the
experiment was conducted and the
post-test scores were obtained at the
conclusion of the experiment. Is there
a difference between the pre-test
and post-test
scores?
Answer
2. A 50-item test is to be
conducted to new applicants for
BS Statistics coming from three
different schools at a certain
university. Is there a different in
the median scores among the
students from three different
schools ( A, B, and C )?
answer
3. Eight judges were selecting a Mutya ng
Mindoro. One of the judges, being a statistician,
suggested that they rank each of the five
contestants ( from one to five with a rank of
one being their top choice ). Is there a
difference in the judgement of each judge?
answer
4. The overall teaching efficiency
rating of both Male and Female
faculty members of the college of
Arts and Sciences at a certain
University during the first semester
of 2021-2022. Determine if there
exist a difference in their median
teaching efficiency rating.
MANN-WHITNEY
 REFERENCE

STATISTICS MADE SIMPLE FOR RESEARCHERS

By: ABUBAKAR S. ASAAD
*THANK YOU