Overview of People I Mostly Admire — Episode 17: Emily Oster
This episode is a long, wide-ranging conversation between Steve Levitt and economist Emily Oster. It covers Oster’s evolution from academic economist to popular public intellectual and data-driven advisor on pregnancy, parenting, and — more recently — COVID-era schooling. Key themes: how to interpret imperfect evidence, decision-making under uncertainty, the costs of remote schooling, her COVID school data work, and several of her substantive research projects (including a celebrated correction of a prior finding).
Major themes and takeaways
- Evidence-first, pragmatic approach: Oster emphasizes finding the best research, explaining uncertainty and trade-offs, and letting readers/families make choices for themselves.
- Decision framing matters: specify realistic alternatives (send child to daycare vs. hire nanny vs. quit job), then choose and move on — avoid perpetual rumination.
- Imperfect data still demands decisions: policymakers and families must weigh harms of inaction (e.g., long school closures) against uncertain risks.
- Selection bias is powerful: recommendations can change who adopts behaviors, creating self-fulfilling observational patterns (e.g., vitamin supplements).
- Science + humility: good scholarship means revising conclusions when better data arrive — Oster highlighted her own example of overturning an earlier hypothesis.
Highlights from her popular work
- Expecting Better (pregnancy): reads and synthesizes pregnancy research for non‑experts; models how to weigh risks and benefits rather than passing judgment. Example: the apparent link between caffeine and miscarriage largely disappears after controlling for confounders (age, nausea).
- Crib Sheet (parenting): data are thinner and effects smaller/less lasting than public conversation implies; many early-childhood choices have limited long-term impact — choose what fits your family.
- Breastfeeding: benefits are real but narrower than public mythology — short-term digestive benefits for infants and reduced maternal breast-cancer risk; no robust evidence for long-term IQ or broad lifelong advantages.
COVID, schools, and the data project
- Newsletter → data collection: Oster began with a newsletter and a small Google form tracking kids/daycares; this grew into a larger COVID School Dashboard that aggregated cases in in-person settings.
- Main empirical findings:
- Student infection rates in schools were low and tracked community transmission.
- Staff/teacher rates were higher than student rates (likely reflecting age and exposure differences).
- Schools that were open often operated as relatively controlled environments (masking, protocols), which limited spread.
- Costs of remote schooling: evidence shows large educational losses, lower attendance/engagement, and harms to emotional and physical well-being for many students — these harms factored strongly into Oster’s advocacy for reopening.
- Advocacy vs. neutrality: Oster admits advocacy complicates neutral analysis but argues we still must act on best-available evidence to minimize overall harm (she supports vaccinating teachers, and considered compensation for higher-risk essential workers).
Key research projects discussed
- Health-fad / self-fulfilling prophecy paper:
- Thesis: media-driven endorsements can change who adopts a behavior; adopters are often healthier, which can inflate observational associations between the behavior and outcomes (vitamin E example).
- Lesson: accounts that ignore adoption-selection can produce misleading causal claims.
- Huntington’s disease and genetic testing:
- Question: why do relatively few at‑risk people take a perfectly predictive test? Oscillates between rational-choice expectations and the emotional value of optimism ("I don’t want to know").
- Insight: anticipation/forecasts about the future can carry utility; people sometimes prefer uncertainty because it preserves hope.
- Hepatitis B and sex-ratio paper (retraction/replication story):
- Initial claim: higher hepatitis B rates might explain skewed sex ratios (part of the “missing women” literature) in some Asian countries.
- Later development: better-quality data contradicted the effect. Oster collected new data, found the effect did not hold, and published the correction — an example she and Levitt frame as rigorous science (and rare in economics work).
- Levitt defends her handling of the reversal as exemplary scientific integrity, although the episode drew criticism in the profession.
Notable quotes & framing lines
- “I am a woman who is prominently discussing vaginas.” — on why reactions to her work sometimes intersect with gendered backlash.
- “There’s no safe, great option here.” — on school reopening trade-offs during COVID.
- “I don’t want to know.” — concise description of why many at-risk individuals decline predictive genetic tests.
Practical advice and recommendations
- For parents making hard choices:
- Frame the decision with clear, realistic alternatives.
- Quantify (when possible) risks and benefits; control for confounders in studies you consult.
- Make a decision, commit for a period, and move on to reduce decision fatigue.
- For public communicators/researchers:
- Be transparent about uncertainty and study limitations.
- Beware of selection effects when interpreting observational findings.
- When new, higher-quality data contradict earlier results, correct and publicize the revision.
- COVID/schools policy:
- Consider prioritizing teachers for vaccination (practical access + supporting reopening).
- Balance low observed in-school spread against substantial harms of prolonged remote schooling.
Why this episode matters
- It’s a primer on applying economic reasoning and empirical skepticism to everyday life and public policy.
- It showcases an intellectual commitment to changing one’s mind in light of better data — a model for evidence-based public engagement.
- The episode distills how to think under uncertainty (both individually for parents and collectively for policymakers) and provides concrete examples from pregnancy, parenting, public health, and academia.
Who should listen / read this summary
- Parents wanting an evidence-based lens on pregnancy and early-childhood choices.
- Educators and policymakers weighing school reopening.
- Researchers and communicators interested in how observational bias and public recommendations interact.
- Anyone curious about a high-profile example of scientific self-correction and the public trade-offs of advocacy vs. neutrality.