Emily Oster on Pregnancy, Causation, and Expecting Better
What this covers
Emily Oster, an economist at the University of Chicago, sits down with host Russ Roberts to examine how medical evidence should guide pregnancy decisions. Oster's core claim is that standard pregnancy prohibitions often rest on weak causal ground—frequently mere correlations tangled up in confounding—and that pregnant women deserve actual numbers and magnitudes rather than blanket rules. The conversation builds from specific food and substance restrictions (alcohol, caffeine, deli meat, raw eggs) through broader questions about medical decision-making: How do we establish causation when observational studies abound but randomized trials are rare? What does "safe" really mean when everyday risks like car travel dwarf the microscopic added risks from moderate food choices? Oster argues the right framework is neither paternalistic absolutism nor recklessness, but Marginal Risk Analysis—understanding what actual difference an action makes, in magnitude, and letting individuals weigh that against their own preferences.
The episode ranges across pregnancy's medical landscape: the Spurious Correlation / The Dreaded Third Thing that makes caffeine-miscarriage links misleading (nausea, not caffeine, may drive the pattern); moderate alcohol consumption, where large European studies show no harm to children; listeria risk from restricted foods, which come mostly from unpredictable sources like cantaloupe anyway; and toxoplasmosis from cats, which is rarer than the advice suggests. The conversation also covers C-section overuse, the evidence for doulas as genuine interventions, and why the power imbalance during labor makes informed consent difficult. Roberts notes that Oster's public concession of error in her earlier hepatitis-B research—updating her position when better data arrived—marks her as unusual among academics. Throughout, the discussion treats pregnancy as a case study in how to interpret uncertainty and make decisions with incomplete information.
Oster argues that many standard pregnancy prohibitions are not supported by the medical evidence once you account for confounding and causation, and that women should be given the actual numbers to make their own risk-benefit decisions rather than being handed blanket safe/unsafe rules.
- Observational pregnancy studies are riddled with confounders (the 'dreaded third thing'), so raw correlations like caffeine-miscarriage are misleading
- Moderate alcohol and many restricted foods show little or no causal harm in better-designed studies
- Decision-making under uncertainty requires data, magnitudes, and individual preferences, not paternalistic absolutism
Doctors are poorly trained in data analysis and statistics, causing them to think in binary safe-or-unsafe terms rather than recognizing risk as a continuum.
- Doctors are particularly ill-suited to risk-tradeoff questions because they typically receive little training in data analysis, statistics, or decision-making, leading them to think in binary safe/unsafe terms rather than recognizing a continuum of risk that economists naturally consider.
“They don't typically get trained in data analysis. They are certainly not trained in statistics or decision-making. They don't have a very good appreciation of uncertainty.”
Economists rarely concede error publicly; Oster's published acknowledgment that her original conclusion was wrong places her in an exceptionally small class.
- It is extremely rare in economics for a researcher whose paper is countered to concede they were actually wrong; the norm is to defend the result by re-running regressions with new controls, so Oster's public concession via a paper showing her original conclusion was wrong places her in a very small class of academics.
“how rare it is that when a paper comes out that counters what the original paper found, someone then concedes that they were actually wrong.”