non-parametric tests Flashcards
(15 cards)
What is a parametric test?
A parametric test assumes data are normally distributed and uses parameters like mean and standard deviation.
Why use a non-parametric test?
Non-parametric tests are used when data are ordinal or skewed, meaning assumptions of normality and equal intervals are not met.
When should you use the Mann-Whitney U test?
Use the Mann-Whitney U test when comparing two independent groups with ordinal or non-normal data.
What is the rationale behind the Mann-Whitney U test?
It ranks all scores and checks if group ranks are randomly mixed (H₀ true) or clustered at ends (H₀ false).
How is the Mann-Whitney U statistic calculated?
U = n1 × n2 + (n1(n1+1)/2) - T, where T is the rank total of one group. SPSS typically handles this calculation.
When should you use the Wilcoxon Signed-Rank Test?
Use it for comparing two related (paired) samples with ordinal or skewed data.
How does the Wilcoxon test work?
It ranks the absolute differences between paired scores and compares the sum of positive and negative ranks.
When is the Kruskal-Wallis H test used?
It’s used to compare 3 or more independent groups with ordinal or non-normal data.
What does the Kruskal-Wallis test tell you?
It indicates if there is any significant difference between groups but not which specific groups differ.
How can you follow up on a significant Kruskal-Wallis test?
Use Mann-Whitney tests for pairwise comparisons and apply Bonferroni correction to control for Type I error.
When should you use the Friedman Test?
Use the Friedman test for comparing 3 or more related conditions (within-subjects) with ordinal or non-normal data.
What does the Friedman test show?
It shows whether there is a significant difference across conditions but not between which pairs specifically.
What follow-up tests are used after a Friedman test?
Follow up with Wilcoxon signed-rank tests for pairwise comparisons and apply Bonferroni correction.
What is a key limitation of non-parametric tests?
They discard information about score magnitudes by only using ranks, which can reduce statistical power.
What is a common feature of non-parametric tests?
They are all based on ranked data rather than raw scores, making them suitable for skewed or ordinal data.
