Quantitative analysis: choose the right test and interpret honestly

Statistics is a tool for honesty, not persuasion. The right test and full reporting beat impressive p-values every time.

Test by design, not by result

The choice of test follows from data type, number of groups and the question: comparison, association or change over time. Decide the analysis before you see the data. Trying tests until something becomes 'significant' (p-hacking) is one of the gravest errors in quantitative research, and reviewers spot it fast.

Report more than the p-value

A p-value alone says little. Report effect size, confidence interval and sample size, so the reader can judge both statistical and practical significance. A small p-value with a trivial effect is not a discovery.

Pre-registering hypotheses and analysis strengthens credibility and counters p-hacking — increasingly expected in serious research.

Check the assumptions

Every test rests on assumptions — normality, homogeneity of variance, independence. Checking and reporting these, and choosing robust alternatives when they're violated, is what separates reliable analysis from number-magic. Be open when the assumptions don't hold.

Common pitfalls

• Trying tests until something becomes significant (p-hacking).
• Reporting a p-value with no effect size.
• Ignoring the test's assumptions.

Ready to move on when…

• The test is chosen by design, not by result.
• Effect size and confidence interval are reported.
• The assumptions are checked and addressed.