Scribd
Upload
22 views

Lesson 02 - V2

The document discusses two nonparametric statistical tests for comparing paired or related samples: the Wilcoxon signed rank test and the sign test. It provides details on computing the test statistic for the Wilcoxon signed rank test, including using sums of positive and negative ranks. An example is given comparing counseling intervention success rates before and after a new anti-bullying program. The null and research hypotheses are stated and the Wilcoxon signed rank test is selected as appropriate. Test statistics are computed and compared to critical values to determine if the null hypothesis can be rejected.

Uploaded by

Vero Abey
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
22 views

Lesson 02 - V2

The document discusses two nonparametric statistical tests for comparing paired or related samples: the Wilcoxon signed rank test and the sign test. It provides details on computing the test statistic for the Wilcoxon signed rank test, including using sums of positive and negative ranks. An example is given comparing counseling intervention success rates before and after a new anti-bullying program. The null and research hypotheses are stated and the Wilcoxon signed rank test is selected as appropriate. Test statistics are computed and compared to critical values to determine if the null hypothesis can be rejected.

Uploaded by

Vero Abey
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Comparing Two Related

Samples: The Wilcoxon


Signed Rank And The Sign
Test
1. Motivation
 Imagine that you give an attitude test to a small group of people. After you deliver some type
of treatment, say, a daily vitamin C supplement for several weeks, you give that same group of
people another attitude test.
 Finally, you compare the two measures of attitude to see if there is any type of difference
between the two sets of scores. The two sets of test scores in the previous scenario are related
or paired. This is because each person was tested twice.
 In other words, each test score in one group of scores has another test score counterpart.
 The Wilcoxon signed rank test and the sign test are nonparametric statistical procedures for
comparing two samples that are paired or related.
Computing The Wilcoxon Signed Rank
Test Statistic
 The formula for computing the Wilcoxon T for small samples is shown in the following formula.
The signed ranks are the values that are used to compute the positive and negative values in the
formula:

where ΣR+ is the sum of the ranks with positive differences and ΣR- is the sum of the ranks
with negative differences.
 After the T statistic is computed, it must be examined for significance. We may use a table of
critical values. However, if the numbers of pairs n exceeds those available from the table, then a
large sample approximation may be performed.
 For large samples, compute a z-score and use a table with the normal distribution to obtain a
critical region of z-scores.
Computing The Wilcoxon Signed Rank
Test Statistic
 Following formulas are used to find the z-score of a Wilcoxon signed rank test for large samples:

Z* is the Normal Approximation.


At this point, the analysis is limited to identifying the presence or absence of a significant difference
between the groups and does not describe the strength of the treatment. We can consider the effect
size (ES) to determine the degree of association between the groups. We use following Formula to
calculate the ES:

where |z| is the absolute value of the z-score and n is the number of matched pairs included in the
analysis.
Computing The Wilcoxon Signed Rank
Test Statistic
 The ES ranges from 0 to 1. Cohen (1988) defined the conventions for ES as small = 0.10,
medium = 0.30, and large = 0.50.
 Correlation coefficient and ES are both measures of association
Example: Wilcoxon Signed Rank Test
The counseling staff of Clear Creek County School
District has implemented a new program this year
to reduce bullying in their elementary schools.
The school district does not know if the new
program resulted in improvement or
deterioration. In order to evaluate the program’s
effectiveness, the school district has decided to
compare the percentage of successful
interventions last year before the program began
with the percentage of successful interventions
this year with the program in place. In the
following Table, the 12 elementary school
counselors, or participants, reported the
percentage of successful interventions last year
and the percentage this year:
Example: Wilcoxon Signed Rank Test
 State the Null and Research Hypotheses:

 Choose the Appropriate Test Statistic: The data are obtained from 12 counselors, or participants,
who are using a new program designed to reduce bullying among students in the elementary schools.
The participants reported the percentage of successful interventions last year and the percentage this
year. We are comparing last year’s percentages with this year’s percentages. Therefore, the data
samples are related or paired. In addition, sample sizes are relatively small. Since we are comparing
two related samples, we will use the Wilcoxon signed rank test.
Example: Wilcoxon Signed Rank Test
Compute the Test Statistic First,
compute the difference
between each sample pair.
Then, rank the absolute value
of those computed differences.
Using this method, the
differences of zero are ignored
when ranking. We have done
this in Table
Example: Wilcoxon Signed Rank Test
 Compute the sum of ranks with positive differences. Using WS - Table, the ranks with positive
differences are 9, 7, 4.5, 10, 1, 6, 8, and 2. When we add all of the ranks with positive difference
we get ΣR+ = 47.5.
 Compute the sum of ranks with negative differences. The ranks with negative differences are 3
and 4.5. The sum of ranks with negative difference is ΣR− = 7.5.
 The obtained value is the smaller of the two rank sums. Therefore, the Wilcoxon is T = 7.5.
 Determine the Value Needed for Rejection of the Null Hypothesis: Using the Appropriate Table
of Critical Values for the Particular Statistic Since the sample sizes are small, we use WS Table,
which lists the critical values for the Wilcoxon Sign Test. As noted earlier, the two counselors
with score differences of zero were discarded. This reduces our sample size to n = 10. In this
case, we look for the critical value under the two-tailed test for n = 10 and α = 0.05. Table
returns a critical value for the Wilcoxon test of T = 8. An obtained value that is less than or equal
to 8 will lead us to reject our null hypothesis.

You might also like