Non-Parametric Alternatives Flashcards
Assumptions for data to be parametric
- data level of measurement to be continuous
- normally distributed
- equal variances across groups to compare
Which non-parametric test would you use in the place of the Student’s t test?
Mann-Whitney U test
Which non-parametric test would you use in the place of the paired t-test?
Wilcoxon signed-rank test
Which non-parametric test would you use in the place of ANOVA?
Kruskal-Wallis H test
Which non-parametric test would you use in the place of repeated measures ANOVA?
Friedman test
How are the scores ordered in non-parametric rankings?
scores are ranked from smallest to largest with 1 assigned to the smallest score and n to the highest score
T/F: non-parametric tests make no assumption on the distribution of data
True
What are 2 reasons non-parametric tests may not be testing the null hypothesis of interest
null hypotheses that can be studied using non-parametric tests tend to be very restrictive
there is not much choice for non-parametric tests
T/F Non-parametric methods are focused on estimation rather than significance testing
False: more focused on significance
Non-parametric tests can’t calculate CI
Mann-Whitney U test
used to test mean difference between two independent groups
Wilcoxon Signed-Rank Test
used to test mean difference between two matched groups
Kruskal-Wallis H test
used to test mean difference between 3 or more independent groups
Friedman test
used to test the mean difference between 3 or more related groups
List the weaknesses of non-parametric tests
less powerful than parametric
may not be testing null hypothesis of interest
null hypotheses using non parametric tests tend to be restrictive
not much choice
can’t be related to the CI
