RESEARCHING THE REAL WORLD



MAIN MENU

Basics

Orientation Observation In-depth interviews Document analysis and semiology Conversation and discourse analysis Secondary Data Surveys Experiments Ethics Research outcomes
Conclusion

References

Activities

Social Research Glossary

About Researching the Real World

Search

Contact

© Lee Harvey 2012–2024

Page updated 8 January, 2024

Citation reference: Harvey, L., 2012–2024, Researching the Real World, available at qualityresearchinternational.com/methodology
All rights belong to author.


 

A Guide to Methodology

8. Surveys

8.1 Introduction to surveys
8.2 Methodological approaches
8.3 Doing survey research
8.4 Statistical Analysis

8.4.1 Descriptive statistics
8.4.2 Exploring relationships

8.4.3 Analysing samples

8.4.3.1 Generalising from samples
8.4.3.2 Dealing with sampling error
8.4.3.3 Confidence limits
8.4.3.4 Statistical significance
8.4.3.5 Hypothesis testing
8.4.3.6 Significance tests

8.4.3.6.1 Parametric tests of significance
8.4.3.6.2 Non-parametric tests of significance

8.4.3.6.2.1 Chi-square test
8.4.3.6.2.2 Mann Witney U test
8.4.3.6.2.3 Kolmogorov-Smirnov Test
8.4.3.6.2.4 H test
8.4.3.6.2.5 Sign test
8.4.3.6.2.6 Wilcoxon test
8.4.3.6.2.7 Friedman test
8.4.3.6.2.8 Q test

8.4.3.7 Summary of significance testing and association: an example

8.4.4 Report writing

8.5 Summary and conclusion


8.4.3.6.2.4 The Kruskal-Wallis H test

When to use the H test
Hypotheses
What we need to know
Sampling distribution
Assumptions
Testing Statistic
Critical values
Worked examples
Unworked examples

When to use the Kruskal-Wallis H test
The Kruskal-Wallis H test is direct extension of the Mann-Whitney U test for more than two independent samples.

The test can only be used if all the samples can be ranked in numerical size and all samples are in the same units;  thus the data must be at least ordinal.

Top

 

Hypotheses
H0: The samples are from identical populations

HA: The samples are not from identical populations, (i.e. the samples are significantly different). Two-tail test

Top

 

What we need to know
The data in the samples must be at least ordinal so that each item in the combined sample can be ranked in order of size (remembering to retain the identity of each item).

Top

 

Sampling distribution
The sample distribution is denoted as H and the distribution of H approximates the chi-square (χ2) distribution for j-1 degrees of freedom (where j is the number of samples).

Top

 

Assumptions
There are no assumptions underlying the use of this test provided the samples are independent.

Top

 

Testing statistic
For all sizes of sample the testing statistic is:

H = χ2 = (12/(N(N+1)).(∑R2/n)) - 3(N+1)

where N= the sum of the sample sizes, i.e. ∑n

R = sum of ranks for each sample. So R2/n has to be calculated for each separate sample and then summed.

Top

 

Critical values
The critical value of the testing statistic can be found from tables critical χ2 values for the relevant significance level and degrees of freedom (i.e. j-1, where j is the number of samples).

An on-line version of the chi-square critical values table can be found, for example, at the Engineering Statistics Handbook:

https://www.itl.nist.gov/div898/handbook/eda/section3/eda3674.htm (accessed 24 May 2020)

Top

 

Worked example
A group of 25 people in the UK were given a political spectrum test (0 extreme left wing through to 20 extreme right wing). The group consisted of three samples based on their views about the monarchy: 'in favour of dissolution', 'indifferent', 'retain monarchy'. Is there any evidence to suggest right wingers are more in favour of retaining the monarchy?

Political spectrun scores for the three groups:

Dissolve monarchy: 12, 16, 3, 10, 7, 8, 10

Indifferent: 15, 15, 2, 11, 0, 19, 15, 9, 5, 9, 11

Retain monarchy: 1, 4, 20, 17, 6, 13, 18

Hypotheses:

H0: There is no significant difference in the scores between the three attitude about the monarchy groups.

HA: There is a significant difference in the the scores.

This is a two-tail test of significance.

Significance level: 5%

Testing statistic: H

H = χ2 = (12/(N(N+1)).(∑R2/n)) - 3(N+1)

where N= the sum of the sample sizes, i.e. ∑n and R = sum of ranks for each sample.

Critical value: From tables of critical χ2 for two degrees of freedom (j-1=2),

critical χ2 = 5.99.

Decision rule: reject H0 if calculated χ2 > 5.99

Computation:

Rank each item of the combined sample scores retaining identity (where scores are the same the ranks are averaged over the equal scores):

Dissolve Indifferent Retain
Score Rank Score Rank Score Rank
3 4 0 1 1 2
7 8 2 3 4 5
8 9 5 6 6 7
10 12.5 9 10.5 13 17
10 12.5 9 10.5 17 22
12 16 11 14.5 18 23
16 21 11 14.5 20 25
    15 19    
    15 19    
    15 19    
    19 24    
  R1 = 83   R2 = 141   R3 = 101
  n1 = 7   n2 = 11   n3 = 7

N = n1 + n2 + n3 = 7 + 11 + 7 =25

∑R2/n = 832/7 + 1412/11 + 1012/7 = 984.14 + 1807.36 + 1457.28 = 4248.78

H = χ2 = (12/(N(N+1)).(∑R2/n)) - 3(N+1)

H = χ2 = (12/(25(26)).((4228.78)) - 3(25+1) = 0.01846(4248.78) - 78 = 78.432 - 78 = 0.43

 

Decision: Cannot reject the null hypotheses (H0) as calculated chi-square (0.43) is less than the critical value of 5.99; there is no signifcant difference between samples.

Top

 

Unworked examples

1. The following scores on a scale of conservativism (0 to 100, with 100 being ultraconservative) were achived by three seminar groups who together constituted a first-year politics degree course. Is there any significant difference in the results of the three groups. What does the result suggest?

Seminar 1: 76, 14, 23, 29, 41, 56, 83 42

Seminar 2: 76, 63, 24, 88, 37, 69, 49, 48

Seminar 3: 60, 67, 63, 29, 28, 51, 35, 80, 40

 

2. A survey of parents to see how important it was for them that their senior-school children achieved sufficient grades to apply for university. A score of 1 in the table below suggests an indifference to going to university while a score of 9 indicates it is seen as essential by parents. Is there any significant difference between the three occupation groups?

Manual occupation: 2,2,2,3,4,4,4,6,6,6,6,8,9,9,9,9,9

Clerical occupation: 4,4,6,6,7,7,7,7,7,7,9,9,9,9,9,9,9

Professional occupation: 1,1,3,5,5,5,8,8,8,8,8,9,9,9,9,9,9

 

Next 8.4.3.6.2.5 Sign test

Top