AI Snips
Chapters
Transcript
Episode notes
Post Hoc Power Analysis Is Useless
- Post hoc power analysis is generally useless because estimating power after the fact depends heavily on observed nonsignificant results.
- The effect size is often underestimated in these cases, making power calculations uninformative.
Complex Models Blur Power Analysis
- Power analysis from Cohen's era was clear for simple tests but is muddled by modern complex models.
- It is virtually impossible to specify power for complex models, making power calculations less meaningful.
Pig Diet Study Lacked Power
- Patrick Curran shared an anecdote about a pig diet study with only six pigs per group.
- The study had no real chance of detecting vision problems, rendering it essentially useless.