Study Schedule

WEEK TOPIC LEARNING OUTCOMES ACTIVITIES

CHAPTER 5 NON-PARAMETRIC TEST
(GROUP COMPARISON)
Lesson 5 Mann – Whitney U-Test
September 14 1. Use Mann-Whitney test to Engage: Query
- 16, 2020 test the assumptions of Explore: Overview of Mann-
the t-test. Whitney Test
Explain: Describe
Elaborate: Challenge Problem
Evaluate: Activity 5.1
Enrichment: Assignment

September 17, Completion of Let’s Try This and File your activity in your red
2020 Gauge Your Learning Activities long clear book.

CHAPTER 5: NON-PARAMETRIC TEST (GROUP COMPARISON)

Lesson 5: Mann – Whitney Test

Learning Objectives: At the end of the lesson, the students are expected to:
1. Use Mann-Whitney test to test the assumptions of the t-test.

Introduction:
According to Glen (2015), that Mann-Whitney U test is the nonparametric
equivalent of the two-sample t-test. Furthermore, t-test makes an assumption about the
distribution of a population (i.e. that the sample came from a t-distributed population),
the Mann Whitney U Test makes no such assumption.

Lesson

Query
“Your ability to solve problems and make
good decisions is the true measure of your
skill as a leader.” – Brian Tracy
The independent sample t-test is only applicable when the data is normally
distributed and the number of samples is at least 30, on determining significant differences.
What appropriate statistical tool will be used if the assumption is not meet? Can we still use
the independent sample t-test?
Answer is NO.

46
 Since comparing two independent samples when the outcome is not normally
distributed and the samples are small, a nonparametric test is appropriate.
Mann-Whitney U or sometimes called
t U
Wilcoxon Rank Test is used to compare
differences between two independent groups
when the dependent variable is either ordinal or
continuous, but not normally distributed and the
number of samples are small.
This test is run by ranking all observations and then finding the sum of the ranks in
each group. If the values in one group are generally higher than in the other, the rank sums
will differ. The test statistic, U is a measure of the difference in rank sums.

Overview

Example1:
The effectiveness of advertising for two rival products (Brand X and Brand Y) was
compared. Market research at a local mall was carried out, with the participants being
shown adverts for two rival brands of coffee, which they then rated on the overall likelihood
of them buying the product (out of 10, with 10 being "definitely going to buy the product").
Half of the participants gave ratings for one of the products, the other half gave ratings for
the other product. (At 0.05 level of significance)

Ho: The two populations are equal

H1: The two populations are not equal

We have two conditions, with each participant taking part in only one of the
conditions. The data are ratings (ordinal data), and hence a nonparametric test is

47
appropriate - the Mann-Whitney U test (the non-parametric counterpart of an
independent measures t-test).

Rank all scores together, ignoring which group they belong to.

Brand X Brand Y
Rating Rank Rating Rank
2 1.5 5 5.5
2 1.5 6 7.5
3 4 7 9
4 5 8 8
5 5.5 9 9
6 7.5 10 10

Brand X Brand Y
Rating Rank Rating Rank
2 1.5 5 5.5
2 1.5 6 7.5
3 4 7 9
4 3 8 10
5 5.5 9 11
6 7.5 10 12
Sum of Brand X 23 Sum of Brand Y 55

The test statistic for the Mann Whitney U test is denoted U and is the smaller of U 1
and U2,
𝑛1 (𝑛1 + 1)
𝑈1 = 𝑛1 𝑛2 + − 𝑅1
2

𝑛2 (𝑛2 + 1)
𝑈2 = 𝑛1 𝑛2 + − 𝑅2
2

Where R1 = sum of the ranks for group 1 and R2 = sum of the ranks for group 2. Then,
R1 = 23
and R2 = 55.

6 (6 + 1)
𝑈1 = 6 (6) + − 23
2

48
 6(7)
𝑈1 = 36 + − 23
2

𝑈1 = 36 + 21 − 23

𝑈1 = 34

6(6 + 1)
𝑈2 = 6 (6) + − 55
2

6(7)
𝑈2 = 36 + − 55
2

𝑈2 = 36 + 21 − 55

𝑈2 = 2

Thus, smaller values of U will be taken as the computed U value. By referring to table
of Mann-Whitney U table 0.05, Critical value n1 * n2; 6 * 6 = 5.

U = 2 has the smaller values of U. Since U < CV or 2 < 5, then the two population are
equal, the buyers are definitely buying both brands.

Describe
Give one problem that requires a Mann-Whitney test to use. Explain your reason
why Mann-Whitney should be used in the problem. Determine the difference between
independent t-test and Mann-Whitney test. When to use an independent t-test? When to
use Mann-Whitney test?

Challenge Problem

The Protein–IQ Experiment Revisited

You believe a high-protein diet during infancy will increase intellectual functioning in
children in your geographical area. You decide to investigate this hypothesis using the
independent groups design. Alpha is set at the beginning of the experiment to 0.05. Six
children are randomly sampled from all children 1 year old living in your geographical area
and then further randomly divided into two groups of three children each. The control
group is fed a low-protein diet for 3 years, and the experimental
group gets a high-protein diet for the same time period. At the end of the 3 years, each child
is given an IQ test. The results are in the table below.

49
 Data from protein–IQ experiment repeated
Control Group 1 Experimental Group 2
84 194
88 101
98 105

a) What is the directional alternative hypothesis?

b) What is the null hypothesis?
c) What can we conclude? Use 0.05.
d) To what population does this conclusion apply?

Evaluation

Activity 5.1: Solve the following problems using the 5 steps of hypothesis testing. (10 points
each)

1. Consider a Phase II clinical trial designed to investigate the effectiveness of a new

drug to reduce symptoms of COVID 19. A total of n=10 participants are randomized
to receive either the new drug or a placebo. Participants are asked to record the
number of episodes of shortness of breath over a 1week period following receipt of
the assigned treatment. The data are shown below.
Placebo 7 5 6 4 12
New Drug 3 6 4 2 1

Use 0.01 level of significance.

2. Suppose you wished to determine if there was a difference in the biomass of male
and female Juniper trees. The researcher randomly selected 6 individuals of each
gender from the field, dry them to constant moisture, chip them, and then weigh
them to the nearest kg.
Raw Male 74 77 78 75 72 71
data Female 80 83 73 84 82 79
Use 0.05 level of significance

Enrichment
Assignment

Someone has told you that men are better in abstract reasoning than women. You
are skeptical, so you decide to test this idea using a nondirectional hypothesis. You
randomly select eight adult men and eight adult women living in your hometown and

50
administer an abstract reasoning test. A higher score reflects better abstract reasoning
abilities. You obtain the following scores:

a) What is the alternative hypothesis? Assume a nondirectional hypothesis is

appropriate.
b) What is the null hypothesis?
c) Using 𝛼 = 0.05, what do you conclude?
d) To what population do your results apply?

51
 WEEK TOPIC LEARNING OUTCOMES ACTIVITIES
CHAPTER 5 NON-PARAMETRIC TEST
(GROUP COMPARISON)
Lesson 5.2 Kruskal-Wallis H-Test
September 17 - 1. Perform Kruskal-Wallis H- Engage: Query
22, 2020 test. Explore: Overview of Kruskal-
Wallis H-test
Explain: Describe
Elaborate: Challenge Problem
Evaluate: Activity 5.2
Enrichment: Assignment

September 23, Completion of Let’s Try This and File your activity in your red
2020 Gauge Your Learning Activities long clear book.

Lesson 5.2: KRUSKAL-WALLIS (H-TEST)

Learning Objectives: At the end of the lesson, the students are expected to:
1. Perform Kruskal-Wallis H-test.

Introduction
According to MacFarland and Yates (2019), Kruskal–Wallis H-Test for one-way
Analysis of Variance (ANOVA) by Ranks is often viewed as the nonparametric equivalent of
the parametric one-way Analysis of Variance (ANOVA). Furthermore, Kruskal–Wallis Test, as
a nonparametric test, is used with ranked data, particularly for when:
(1) the data are ordinal and do not meet the precision of interval data,
(2) there are serious concerns about extreme deviation from normal distribution, and
(3) there is considerable difference in the number of subjects for each comparative group.

The test statistic for the Kruskal Wallis test (mostly denoted as H) is defined as follows:
12 𝑇𝑖2
𝐻= ∑ − 3 (𝑁 + 1)
𝑁(𝑁 + 1) 𝑁𝑖

Where 𝑇𝑖 = rank sum for the ith sample i = 1, 2,…,k, degree of freedom k -1

In Kruskal-Wallis test, the H value will not have any impact for any two groups in which
the data values have same ranks. Either increasing the largest value or decreasing the
smallest value will have zero effect on H. Hence, the extreme outliers (higher and lower
side) will not impact this test.

52
Lesson

Query

Your social science teacher decides to

investigate how attitudes towards the control of
law and order vary between supporters of three
political groups (Nacionalista, Liberal or PDP-
Laban) in the Philippines.
The attitudes are examined based on responses to Likert-style questionnaires that
elicit answers to a series of questions measured on a scale of 1 (in favor of greater control) to
7 (in favor of less control). To test the significant difference between the three political
groups, is analysis of variance an appropriate statistical tool for this problem? Why or why
not?
Answer is NO. Since the data is ranking, one-way ANOVA is not applicable to the
given data. The H test is used when the assumptions for one-way ANOVA aren’t met (like
the assumption of normality). The Kruskal-Wallis H test is a rank-based nonparametric test
that can be utilized on determining significant differences between two or more groups of
an independent variable on a continuous or ordinal dependent variable.

Overview
The test statistic for the Kruskal Wallis test (mostly denoted as H) is defined as
follows:
12 𝑇𝑖2
𝐻= ∑ − 3 (𝑁 + 1)
𝑁(𝑁 + 1) 𝑁𝑖

Where 𝑇𝑖 = rank sum for the ith sample i = 1, 2,…,k, degree of freedom k -1

In Kruskal-Wallis test, the H value will not have any

impact for any two groups in which the data values have
same ranks. Either increasing the largest value or
decreasing the smallest value will have zero effect on H.
Hence, the extreme outliers (higher and lower side) will
not impact this test.

53
Example 1.
In a manufacturing unit, four teams of operators were randomly selected and sent to
four different facilities for machining techniques training. After the training, the supervisor
conducted the exam and recorded the test scores. At 95% (0.05) confidence level does the
scores are same in all four facilities?

Facility 1 Facility 2 Facility 3 Facility 4

64 77 71 52
81 76 86 65
56 84 88 68
87 59 51 82

Solution:
Step 1:
Null Hypothesis H0: The distribution of operator scores are same
Alternative Hypothesis H1: The scores may vary in four facilities

Step 2:
0.05 level of significance.

Step 3:
Kruskal Wallis H test
Degree of freedom: k – 1 = 4 – 1 = 3
Critical Region: 7.815 (based in the chi square table, refer appendix D)

Step 4:
Rank all of the scores, ignoring which group they belong to. The procedure for
ranking is as follows: the lowest score gets the lowest rank.
Facility 1 R Facility 2 R Facility 3 R Facility 4 R
64 5 77 10 71 8 52 2
81 11 76 9 86 14 65 6
56 3 84 13 88 16 68 7
87 15 59 4 51 1 82 12
T1 34 T2 36 T3 39 T4 27

When N = 4 (Since there are 4 sample from each facilities)

12 𝑇12 𝑇22 𝑇32 𝑇42
𝐻= (∑ + + + ) − 3(16 + 1)
16(16 + 1) 4 4 4 4

54
 12 342 362 392 272
𝐻= (∑ + + + ) − 3(17)
16(17) 4 4 4 4

12 1156 1296 1521 729

𝐻= (∑ + + + ) − 51
272 4 4 4 4

𝐻 = 0.0441 (289 + 324 + 380.25 + 182.25) − 51

𝐻 = 0.0441 (1175.50) − 51

𝐻 = 0.8396
Step 4:
Since H < CR or 0.8396 < 7.815. Therefore, Ho is accepted. The scores are same in all
four facilities.

Describe

Give one problem that requires a Kruskal Wallis H test to use. Explain your reason
why Kruskal Wallis H test should be used in the problem. Determine the difference between
Analysis of variance and Kruskal Wallis H test. When to use an Analysis of variance? When to
use Kruskal Wallis H test?

Challenge
The following data represent the operating times in hours for three types of scientific
pocket calculators before a recharge is required:

Use the Kruskal-Wallis test, at the 0.01 level of significance, to test the hypothesis that
the operating times for all three calculators are equal.

Evaluation
Activity 5.2

Carry out a complete test of hypothesis for the following problems.

1. A TESDA evaluator is interested in comparing the evaluation results of the following

group of students undergone TESDA training from dress making, baking and table setting. Is

55
a difference in evaluation results among the different groups of students from different field
of specialization? (Use 0.01 level of significance)
Dress making Baking Table Setting
185 190 201
179 182 195
192 178 180
165 181 187
174 209 215

2. Suppose that the department head of mathematics is interested in the differences among
final exam scores from Mr. Red, Ms. White, and Mrs. Blue’s algebra classes. The Department
Head takes random samples of students from each of these classes and records their final
exam scores as shown:

Mr. Red Ms. White Mrs. Blue

52 66 63
46 49 65
62 64 58
48 53 70
57 68 71
54 73

Determine if there is a difference between the final exam scores of the three
teachers at 0.05 level of significance.

Enrichment Activities

Assignment
In an experiment to determine which of three different missile systems is preferable,
the propellant burning rate is measured. The data, after coding, are given in the table below.
Use the Kruskal-Wallis test and a significance level of α = 0.05 to test the hypothesis that the
propellant burning rates are the same for the three missile systems.

56