When comparing three or more levels of an independent variable for an ordinal dependent variable, one should use a _____ test, specifically _____.
A) nonparametric; the Mann-Whitney U test
B) nonparametric; the Kruskal-Wallis H test
C) parametric; a one-way ANOVA
D) parametric; the Wilcoxon signed-rank test

Question

Quizplus · Accepted Answer

When comparing three or more levels of an independent variable for an ordinal dependent variable, one should use a nonparametric test because the assumptions of parametric tests are not met. The best nonparametric test for this scenario is the Kruskal-Wallis H test. The Mann-Whitney U test is only appropriate for comparing two groups, while the Wilcoxon signed-rank test is used for paired data. Therefore, the correct answer is choice B.

When Comparing Three or More Levels of an Independent Variable