Hypothesis test selector

This is a decision guide, not a calculator: it names a test, says why, and lists the assumptions to check. It does not compute statistics or p values, and it cannot see your data, so every "yes" answer about normality or spread is your claim, not a finding. "Not sure" answers deliberately route to versions that keep their meaning when the assumption fails, which costs a little power in exchange for honesty. Independence of observations is assumed in every path; no answer here can fix a design where the same case is counted several times.