TimelyScholarFree Tools › Which Statistical Test Should I Use?
Answer a few questions about your goal, outcome and study design, and get the right statistical test with its non-parametric alternative.
FreeNo account neededYour data stays in your browser
Compares means between two independent groups.
Alternative: Mann-Whitney U test
Always check your test's assumptions and your field's reporting guidelines. Unsure? Our statisticians can plan the analysis with you.
Compare groups, test a relationship or predict an outcome.
Continuous, ordinal, binary, categorical or time to event.
Number of groups and whether they are independent or paired.
The right test depends on three things: what you want to find out, the type of outcome variable and the study design. Continuous, roughly normal data suit parametric tests such as the t-test and ANOVA; skewed or ordinal data suit rank-based tests such as Mann-Whitney and Kruskal-Wallis.
Paired designs, where the same people are measured twice, need paired tests. Categorical outcomes use chi-square or exact tests, and time-to-event outcomes need survival methods.
Use the non-parametric alternative shown, or consider a transformation. With large samples, t-tests are fairly robust to non-normality.
One-way ANOVA for roughly normal data, or the Kruskal-Wallis test otherwise, followed by post-hoc comparisons.