Let the research question lead
Do not select a test because it is familiar or appears advanced. First decide whether the study aims to compare groups, examine association, estimate a relationship or predict an outcome. The analytical method must answer that specific question.
Identify the design and measurement level
Ask whether observations are independent or paired, how many groups are compared, and whether the outcome is continuous, ordinal or categorical. These choices narrow the suitable tests considerably.
- Two independent groups with a continuous outcome: independent-samples t-test.
- The same participants measured twice: paired-samples t-test.
- Three or more independent groups with a continuous outcome: one-way ANOVA.
- Two categorical variables: chi-square test of independence.
- A continuous outcome with several predictors: multiple linear regression.
- A binary outcome with predictors: binary logistic regression.
Check assumptions before interpreting results
A test recommendation is only a starting point. Examine missing data, outliers, independence, distribution, expected cell counts and model diagnostics. When assumptions are not reasonable, a transformation, robust method, non-parametric alternative or redesigned model may be needed.
Report effect sizes and confidence intervals alongside p-values. Statistical significance alone does not establish practical importance, good measurement or a credible causal claim.