checking-test-assumptions

Statistical assumption checking requires running test-specific diagnostics like Shapiro-Wilk for normality and Levene for equal variance. This Skill maps your planned test to the correct assumptions, returns pass-fail verdicts with evidence, and recommends fallbacks like Welch's t-test if checks fail.

Regression diagnostics that fail normality require switching to robust regression or non-parametric alternatives. This Skill evaluates residual checks and Cook's distance, provides explicit pass-fail verdicts with relevant statistics, and directs you to the appropriate fallback test.

Welch's test is needed when the equal variance assumption fails during t-test assumption checking. The Skill applies Levene's test to check variance homogeneity and explicitly recommends switching to Welch's t-test or Mann-Whitney when the pooled t-test assumptions are invalid.

Chi-squared tests require adequate expected cell counts, and when assumption checking reveals insufficient counts, Fisher's exact test is the recommended alternative. The Skill calculates expected counts, returns a fail verdict with evidence, and directs you to the correct fallback.

Logistic regression assumption checking involves evaluating residual patterns andCook's distance for influential observations. The Skill runs these regression diagnostics, provides explicit pass-fail verdicts with supporting statistics, and identifies whether your planned logistic regression is valid.