Tools

Which Test Should I Use?

The eternal question. Answer a few plain-English questions about your data and design, and get a recommendation — plus the interactive lesson where you can see that test in action. No jargon required.

A recommendation, not a verdict. Real analyses depend on details no flowchart can see — check the test's assumptions (Assumptions & When They Break), and remember that a significant p-value still needs an effect size next to it.

Pick one test, not many. Running several of these on the same data and keeping whichever turns up p < .05 is the fast route to a false positive — the garden of forking paths. Decide your analysis before you look at the data, ideally by preregistering it.

Planning a whole study? The Plan My Analysis wizard takes the test you land on here and builds a full, printable plan — assumptions to check, the sample size you need, and an APA write-up skeleton. Plan your analysis →

The whole map on one screen

Prefer to scan? The table below is the cheat-sheet version, and every test opens its interactive lesson. Want it on paper? Print the one-page which-test cheat sheet. Prefer the full story in prose, with honest advice on messy designs? Read Choosing Statistics for Your Dissertation.

Your situationTest → lessonSection
One group vs. a known valueOne-sample t-test1.11
Same people, measured twicePaired t-test1.11
Two separate groupsIndependent t-test (Welch)1.12
3+ separate groups, one factorOne-way ANOVA2.1
3+ groups, two crossed factorsFactorial ANOVA2.3
Same people, 3+ conditionsRepeated-measures ANOVA2.4
Group comparison, adjusting for a covariateANCOVA3.4
Groups compared on several outcomes at onceMANOVA3.11
Any of the above, assumptions brokenRank-based alternative2.6
Two continuous variables, associationPearson correlation2.8
Two categorical variables, associationChi-square test2.7
Predict a number, one predictorSimple linear regression2.9
Predict a number, several predictorsMultiple regression3.1
Predict yes/noLogistic regression3.7
Predict a countPoisson / negative binomial GLM4.4
Nested / clustered / repeated dataMixed (multilevel) model4.5
Time until an event, with dropoutsKaplan–Meier / survival analysis4.8
Combining many studiesMeta-analysis4.10
Predict as accurately as possible (not test a hypothesis)Machine-learning modelsML 4.1