EconTalk

John Ioannidis on Statistical Significance, Economics, and Replication

Jan 22, 2018
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

Economics Research Footprint

  • A meta-analysis of economics research revealed small, underpowered studies similar to neuroscience.
  • Exaggerated effects are common, often two-fold or even four-fold.
ADVICE

Meta-Analysis Validity

  • Consider meta-analysis validity on a case-by-case basis, as it's not always superior to single studies.
  • A flawed meta-analysis can result from flawed individual studies, but it can still reveal patterns.
INSIGHT

Low Power Problems

  • Low-powered studies have a low probability of detecting true effects (e.g., 8.5% chance in minimum wage studies).
  • This leads to false negatives and inflated effect sizes when something is detected.
Get the Snipd Podcast app to discover more snips from this episode
Get the app