📊 Data Science & Analytics · Critical Consumption
Read other people's A/B tests critically
Validity checklist catching underpowered tests, peeking, and metric tricks in published results.
intermediate~25 minData ScientistsAnalystsResearch Scientists
Steps
- 1Check sample size math against claimed effect size first
- 2Look for stopping rules: when did they decide it was over?
- 3Verify primary metric matches headline claim exactly
- 4Inspect guardrails: what degraded while the winner won?
- 5Segment sanity: does the effect survive obvious subgroups?
- 6Estimate business impact yourself from raw numbers provided
Common Pitfalls
- ▲Headline metrics switched post-hoc from preregistered ones
- ▲Relative percentages hiding absolute effects near zero
Commands
Install with skills CLI
$ npx skills add aniruddhaadak80/skills --skill critical-consumption-ab-test-reading-checklistInstall globally
$ npx skills add aniruddhaadak80/skills --skill critical-consumption-ab-test-reading-checklist -gTags
#experimentation#critical-thinking#statistics#data-science#critical-consumption