Read on Twitter
1 | Election polling failed in 2020, the most striking example being perhaps the failure of any poll to predict Susan Collins' re-election.
Unfortunately, the problem with surveys affects more than our politics. It limits our ability to measure population health.
Unfortunately, the problem with surveys affects more than our politics. It limits our ability to measure population health.
2 | People who do household epidemiological surveys have been telling me for years that rates of refusal have been rising. These surveys are part of the empirical foundation of population health.
3 | When you have < 40% participation rates, which is not uncommon, your disease prevalence estimates become full of selection effects.
4 | There are other ways to estimate the prevalence of diseases or risk factors. But they also have serious biases, and we used to rely on random household surveys to provide a gold standard. We are losing that.
5/fin | People are tired of being surveyed. There may also be losing trust in public health institutions, which is discouraging for many reasons. I don't know how we fix this.
