Comments per file is skewed and clustered — which test?

Test selectionMedium

Problem. Your metric is comments per file. Most files have zero comments, a few have hundreds. Files sit inside teams, and you randomized by team. A colleague ran a two-sample t-test on the raw file-level numbers and got p = 0.03. What do you make of it?

Before you reveal: say your answer out loud, as if you were in the real interview — get your reasoning across clearly first. There is no single correct answer: reading what the interviewer is really after and defending your own thinking is what makes an answer strong.

← How many teams to detect a 2% lift when you randomize by team?Comments +45%, p=0.002, but the split looks off — ship? →