The Power of Nuance: Lessons for Public Health (with Emily Oster)
What this covers
Emily Oster, an economist at Brown University, sits with Russ Roberts to argue that public health communication has systematically failed by treating heterogeneous populations as though they have identical preferences and should obey uniform directives. The conversation ranges across vaccines, fluoridation, raw milk, and the COVID pandemic, grounding a broader claim: when officials hide uncertainty, change advice without explanation, or oversimplify trade-offs, they erode the Trust as Public Health Capital that enables people to make sound decisions. Oster contends that people increasingly do their own research, and officials cannot prevent this by simply issuing blanket commands—they can only choose whether to meet that research with honest complexity or to lose credibility.
The conversation distinguishes decisions by their actual stakes. Childhood vaccines like measles and pertussis rank as clear imperatives: decades of evidence, millions of lives saved, genuine protective power. By contrast, choices like baby-led weaning or co-sleeping practices admit genuine trade-offs where either option works; here, Harm Reduction guidance—teaching safer methods rather than issuing prohibitions—respects that people will choose what they choose anyway. Fluoridation presents a harder case: the evidence for dental benefit is solid, but the conversation fractures when Dose-Response Relationship distinctions get flattened. COVID schooling decisions, Oster argues, ignored elderly preferences and harmed the poorest children the policy claimed to protect. Throughout, both speakers stress that Communicating Uncertainty—naming what officials do not know and how they will learn—preserves trust far better than silent reversals. The underlying tension: whether public health should calculate Trade-offs, Not Solutions for citizens or help citizens navigate those trade-offs themselves.
Oster argues that public health should communicate with nuance and honesty about uncertainty and trade-offs—rather than issuing blanket directives—because in a world where people do their own research, oversimplified or untrue messaging destroys the trust experts need and leads people to make worse choices.
- People have heterogeneous preferences and increasingly do their own research, so single directives often fail and are sometimes wrong on average.
- Blanket bans (e.g., 'never co-sleep') get ignored and forfeit the chance to teach safer behavior.
- Changing advice without explaining why, and hiding stakes or uncertainty, erodes public trust as seen during COVID.
Public health messaging must acknowledge real but small side effects and explain uncertainty to maintain trust.
- Public health messaging on vaccines should not pretend there are no side effects—many kids get a fever about 10 days after the measles vaccine—because acknowledging real but small side effects while emphasizing they are limited maintains trust, whereas denying them causes people to distrust you.
“Many kids do get a fever about 10 days after the measles vaccine. That is a very common side effect. So, I think we don't want to pretend that literally there could never be any... Because, that's not true, and because then people don't trust you.”
- Trust is eroded not only by lying but by failing to explain uncertainty; during COVID, officials changed advice frequently without explaining what new information prompted the change, and changing your mind without explanation makes people distrust you—whereas saying 'we're not sure, here's how we'll learn, and we'll come back' would have kept more people on board.
“changing your mind without explaining is really a way to lose people's trust, because they're, like, 'You told me to do a thing before. That turned out to be wrong. Why is this thing right now?'”
- Communicating about fluoride is hard precisely because there is so much nuance—such as the difference between municipal water fluoride levels and urinary fluoride levels (the latter being roughly half the water level)—that you cannot expect everyone to become an expert in esoteric distinctions, which is why the conversation becomes confusing.
“This is measuring the urinary fluoride level, which turns out to be basically half of the water... we cannot possibly expect everybody to be an expert on the difference between municipal water fluoride levels and urinary fluoridation levels.”
- Translating and explaining research—making findings vivid, conveying uncertainty, and helping people understand data—is a hard skill distinct from producing research, and public health authorities should invest in developing it rather than assuming good communication follows automatically from good science.
“it is something that could be developed and invested in... it is a different skill than producing the research. It's explaining the research.”
Nuanced risk communication and harm-reduction guidance are more effective than blanket prohibitions that people ignore.
- If public health totally dismisses raw milk as incredibly dangerous, it misses the chance to teach people how to reduce risk—such as getting it from a trusted source and a farm that washes things correctly before bottling—lessons that are impossible to convey if communication is not nuanced.
“if we totally dismiss the concept and say, 'Well, this is incredibly dangerous and no one should do it,' we sort of miss the opportunity to explain to people: 'You know, here are some things you want to be thoughtful about.'”
- Because people won't always listen to directives, a blanket prohibition like 'never co-sleep with your baby'—despite co-sleeping being very common—forfeits the chance to teach safer methods, so people co-sleep anyway in less safe ways, making the outcome worse than if nuanced harm-reduction guidance had been offered.
“people aren't listening to it and they're doing something else. And then, you've lost the opportunity to explain to them, say, what's a safer way to do this?”
- If everyone behaved like robots with identical preferences, a single 'here is what we advise' message would be fine, but because people's preferences are heterogeneous, a public health official applying their own preferences to the data and telling everyone what to do can be wrong on average for people whose preferences differ.
“in the world in which people are doing their own research, you need to give them information to help them do that smarter.”
- Public health should respect that a growing share of people are doing their own research and trying to figure out the right thing, so giving them better information to make smarter choices is part of the job of public health rather than simply telling them what to do.
“in the world in which people are doing their own research, you need to give them information to help them do that smarter.”
Public health should distinguish between high-stakes decisions like vaccination and low-stakes choices where either option is acceptable.
- Public health should help people prioritize by distinguishing decisions where the trade-off genuinely matters (e.g., vaccines) from low-stakes choices (e.g., purees versus baby-led weaning) where either option is fine; for the latter, simply tell people any choice works rather than encouraging them to invest heavily in it.
“there are many things where the pediatrician may say, 'Here's what I recommend, but actually it really didn't matter.' So, starting solids... It's really not important.”
- The measles vaccine has been around for decades, has saved millions of lives, and the evidence on both its benefits and risks is clear: childhood vaccinations like measles and pertussis are a very good idea, full stop, because these diseases kill many people and the vaccine is incredibly protective.
“The measles vaccine in particular has been around for many, many decades, has saved literally millions of lives.”
- Raw milk is disproportionately associated with foodborne dairy illness relative to consumption and is not the safest milk to drink, but the actual probability of disease is not spectacularly large—it falls within the range of risks people accept in other parts of life—so choosing raw milk from a trusted local farm is not comparable in evidence quality to choosing not to vaccinate a child.
“it's not that likely that raw milk will cause disease. So, the risk, if you count up the chance and number of cases and how much foodborne illness we see from it, it's not a spectacularly large number.”
School closures were a major policy failure that disproportionately harmed vulnerable children and caused lasting educational damage.
- The October 2020 claim that schools aren't super-spreaders—based on European and early U.S. data—was subsequently confirmed by more careful studies: that first school year, schools that were open did not spread a lot of COVID, even though not literally no one caught it there.
“particularly that first school year when schools were closed, the schools that were open did not spread a lot of COVID.”
- School closures were not a simple equity protection: while poorer kids were more likely to catch COVID, they were also the most likely to experience school closures and lost the most from them, so the policy harmed the very families it was framed as protecting.
“Those kids are also the ones who lost the most. They're the ones who are the most likely to have school closures, and they lost the most from school closures.”
- School closures have proven even more harmful long-term than expected: test scores took an enormous hit everywhere but especially in places that closed schools, and many school years later they have not recovered—only crawling back—meaning there are cohorts of kids who will be affected forever.
“test scores--every place, but especially in places that had closed schools--took an enormous hit during the pandemic. And they basically haven't recovered.”
Fluoridation and pasteurization are effective public health measures that protect vulnerable populations who lack alternative access.
- Municipal water fluoridation should continue because removing it would mostly hurt the most vulnerable—people with good dental access will get fluoride through rinses and dentists anyway, so putting it in the water is what makes fluoride available to groups that otherwise get less access.
“the people who would suffer from taking it out are largely the most vulnerable individuals because people with, like, good access to dentistry... they're going to get these fluoride rinses or other ways that they're getting fluoride.”
- Fluoride protects teeth and reduces cavities, and municipal water fluoridation improves dental health (with strong evidence from Israel where stopping fluoridation increased cavities); neurodevelopmental harms appear only at very high groundwater levels seen in places like China and India—roughly five times U.S. municipal levels—so the dose matters and the real debate is about where the safe threshold lies.
“at low doses, this seems to be good for protecting teeth and doesn't seem to have any negative effects. At high doses, it seems to have negative effects.”
- Pasteurization, developed by Louis Pasteur, kills pathogens and allows milk to be transported and stored safely; in the pre-pasteurization era, transporting milk long distances caused diseases like large tuberculosis outbreaks, so pasteurization saved many millions of lives.
“in the pre-pasteurization era, milk would be transported fairly long distances without pasteurization, and it caused a lot of diseases, like large tuberculosis outbreaks associated with milk.”
- Current COVID vaccine messaging should be more targeted: it is under-emphasizing the value of consistent boosters for older adults while probably overestimating their value to younger people relative to cost, and the result of this untargeted messaging is that nobody is getting them.
“I think we're way under-emphasizing the value for older adults of getting consistent boosters and probably overestimating the value to younger people, at the cost. And the result is nobody's getting them.”
- A major COVID policy failure beyond schools was the failure to protect nursing homes, which were the source of an enormous share of actual deaths, and many terrible things happened there that could have been focused on more.
“the fact that we failed to protect nursing homes, which were the source of such an enormous share of the actual deaths.”
- The idea linking the measles vaccine to autism is not a matter of differing interpretation—the evidence on which it was based was literally made up by someone in the service of money, and large-scale data sets since then show no link.
“the evidence on which that idea is based was literally made up. It's made up. Somebody made it up in the service of money, basically.”
- Oster says she always loves talking to Roberts.
“I always love to talk to you.”