Wednesday, November 04, 2020

Election 2020

Everyone probably has their own version of how an unexpected election outcome merely confirms their existing opinions. So with that pinch of salt, our existing opinion: Pundits and pollsters have failed to grapple with the role of polarization in elections, and especially the role of elections in driving polarization

Would Americans have been as polarized by Coronavirus if it was not an election year? 

The fact that pollsters have now major misses two USA presidential elections in a row is a warning sign that should be familiar to social scientists -- you think you are undertaking a statistically scientific estimate of some fixed underlying parameter, but the very circumstance that makes you want to estimate that number is also causing that number to be fluid! 

Another way to say this. Pollsters and polling analysts (one in particular) tend to think about "news" as something that causes a measurable shift in opinion. But there is no plausible "news" that explains what is emerging in the election outcome. It is more that the election result is itself news about the level of polarization.

To be slightly more technical, polarization is endogenous. It can't be measured by polling, because the polling is correlated with a cause of polarization.  

No comments: