Which test?
Table 1. Some common tests for simple comparisons and when they should be used.
|
Characteristics of outcome variables |
||
Purpose |
Continuous, normal |
Continuous, non-normal OR discrete |
Categorical |
Compare one mean with a population value/ Describe one mean |
One sample z-test/ t-test |
One sample z-test if the sample size is large |
One sample z-test/ exact binomial test |
Compare 2 independent groups |
Independent samples z-test/ t-test |
Mann-Whitney U/ Wilcoxon Sum of ranks |
Chi-squared test or Normal approximation or Fisher’s exact test |
Compare 2 paired/ matched groups |
Paired t-test |
Wilcoxon signed ranks test/ sign test |
McNemar’s test |
Compare 3 or more independent groups |
One way Analysis of Variance |
Kruskal Wallis test |
Chi-squared test or ordinal chi-squared test |
Compare 3 or more matched groups |
Repeated measures Analysis of variance |
Friedman test |
Cochrane Q |