Choosing the right statistical test is mostly a matter of three questions: what kind of outcome you have, how many groups you are comparing, and whether those groups are independent or paired. Answer them below and we will point you to the correct test — and, where we have one, the calculator for it.
The quick version
| Your outcome | Two independent groups | Three or more groups | A relationship |
|---|---|---|---|
| A number / scale | Independent t-test | One-way ANOVA | Pearson correlation / regression |
| A rank / ordinal | Mann-Whitney U | Kruskal-Wallis | Spearman correlation |
| A category | Chi-square test | Chi-square test | Chi-square of association |
Paired or repeated measures on the same people change the choice: a paired t-test, a Wilcoxon signed-rank test, or a repeated-measures ANOVA respectively.
Not sure your choice is right?
Send us your research question and variables and a statistician will confirm the correct test — and run it if you like. Fixed price, by email.
Frequently asked questions
How do I know if my outcome is a number, a rank, or a category?
A number (or scale) is something measured, like height or a test score. A rank is ordered but not evenly spaced, like a satisfaction rating from 1 to 5. A category is a label with no order, like pass/fail or course type.
What is the difference between independent and paired groups?
Independent groups contain different people (for example, a treatment group and a separate control group). Paired or repeated measures come from the same people measured twice, such as before and after an intervention.
What if none of these fit my design?
Complex designs — multiple predictors, nested data, or mixed models — need a tailored approach. Send us the details and we will recommend and run the right analysis.