Thread by @grudkev

The correlation between Trump's share of the vote in 2016 and the percentage of people fully vaccinated in that state is an absurdly high -0.85. As someone who runs all kind of data analysis in the social sciences, getting this kind of a correlation is nearly impossible.

In states where Trump received <40% of the vote, 67% are fully vaccinated.

In states where Trump received >60%, 48% are fully vaccinated.

Trump's vote share explains 73% of the variation in vaccination rates. Increasing it by 1% decreases the vaccination rate by 0.68%.

I suspect there are 3 separate causal mechanisms.

1.Trump voters are less likely to get vaccinated.

2. Trump voters in a position of authority are disproportionately likely to discourage vaccination.

3. GOP-ruled states take fewer Covid-related precautions (e.g., lockdowns).

The risk of dying from a Covid infection is 14 times higher for the unvaccinated and the risk of infection itself is 6 times higher. We're talking about tens (if not hundreds) of thousands of unnecessary deaths based on political loyalties. covid.cdc.gov/covid-data-tra…

Here's a county-level graph of Trump's 2020 vote share and the current vaccination rate in Georgia. Correlation isn't nearly as strong (-0.41), but still clearly in the same direction. #gapol

I realize this site is composed of several echo chambers and I'm getting a visit from one that doesn't quite share my beliefs. If I may be so bold: nothing is gained by someone making a generic statement about how stats can be wrong if they don't show how THESE stats are wrong.

And these are the kind of invaluable insights one obtains from conversing with the other echo chamber. But hey, I'm sure high confidence in one mistaken believe isn't correlated with high confidence in other mistaken beliefs.

I believe the social science equivalent to the Hippocratic Oath is the Hypo-Deductive Oath: I swear by Bayes the Mathematician to pre-register my studies and to not engage in p-hacking.