Science Fictions Episode 10: Cash transfers
21 snips
Sep 12, 2023 The podcast explores cash transfers as a means to improve people's lives. It discusses the evidence supporting cash transfers, concerns about misuse of funds, and the lack of evidence on long-term effects. The hosts also touch on subgroup analysis, pre-registration, technical language in neuroscience research, AI consciousness, and a Canadian study on homelessness inspired by cash transfers in developing countries.
AI Snips
Chapters
Transcript
Episode notes
Mental-Health Effects Are Modest
- Meta-analyses show small but significant mental-health benefits in adults and adolescents from cash transfers.
- Larger or surprising effects would be more suspect; moderate effects are plausible and believable.
Big Journals Don't Guarantee Rigor
- Prestigious journals can publish blunt observational analyses linking government cash programs to mortality changes.
- Such studies risk noise, confounding, and subgroup fishing if not pre-registered or carefully corrected.
Beware Subgroup Fishing
- Post-hoc subgroup analyses (e.g., effects in women but not men) raise the risk of false positives from multiple testing.
- Pre-registration and multiple-comparison correction would clarify which findings are confirmatory versus exploratory.