From “I think” to “the data proves”. This guide walks you from your data type to the right statistical test, and tells you how to read the p-valueThe probability used to decide between the two hypotheses; a value below 0.05 means the evidence against the null hypothesis is strong enough to reject it. Click term for full definition. Watch the short video, then work with the toolkit PDF: it explains the method, shows a worked example and gives you a blank template to fill in.
Tool video
The short tool video will appear here.
Hypothesis Test Selector in short
- State H0 and H1. H0: the cause has no effect. H1: it does.
- Check the data type. Continuous (time, weight) or discrete (pass/fail, category)?
- Name the comparison. Against a target, two groups, before/after, three or more groups? Means, variation or proportions?
- Check normality. Before a t-Test or ANOVA. Not normal? Use a non-parametric testA statistical test used instead of a t-Test or ANOVA when your data isn't normally distributed. Click term for full definition.
- Run and decide. p < 0.05: reject H0, the cause is real. p ≥ 0.05: not proven.
Key points
- Start from H0: no effect.
- Data type and comparison pick the test.
- p < 0.05: the cause is real.
- Check normality before t-Test or ANOVA.