6. Kruskal-Wallis and Freidman tests Flashcards
(17 cards)
What kind of tests are these? (Kruskal-Wallis and Friedman)
Omnibus tests
What does an omnibus test mean?
It investigates the overall difference between several conditions but doesn’t tell us where the difference lies
What is the theory behind these test?
(Mann- Whitney U & Wilcoxon as well)
It is based on ranked data
We order scores from highest to lowest, we collect to find an average of rank score for each group and subtract this from the mean rank score
What are the two types of error we can make?
Type one- false positive
Type two- false negative
What is the Bonferroni correction?
It limits the chances that we will commit a type 1 error by changing the alpha level
What are some benefits of an omnibus test?
One test
One p-value
What are some disadvantages of an omnibus test?
We don’t know where the difference is
Two groups could significantly differ form each other but the whole test may not show this
What test is conducted after an omnibus test is significant?
A Post-hoc test
What is a planned comparison?
Refers to tests you had pre-planned
What are some assumptions of the Friedman test?
3+ conditions
Within-subjects
Ordinal, Interval or ratio data
What are the parameters for Kendall’s W?
0.1= small
0.5= medium
0.8= large
What is a pairwise comparison?
Method of judging or ranking items by comparing them in pairs- two at a time
What is the Conover test?
It takes the ranks of the Friedman test and analyses them taking into account all conditions in the post hoc test
What are some assumptions of the Kruskal-Wallis test?
3+ conditions
Between subjects
Ordinal or scale data (non-parametric)
What is the measure of effect size for Kruskal- Wallis?
Eta squared
What does eta-squared tell us?
How much varience in our results is accounted for by our IV
What are some suggested parameters for eta squared?
Small- 0.1
Medium- 0.6
Large- .14
