Polling Accuracy

A lot of political polling looks like playing darts.  If the small center bullseye is truth, most polls are kind of in the outer ring or even hit the wall outside the dartboard.  And the bias of error in most polls in recent elections has been to understate support for the Republican candidate, often by several percentage points.  In close elections, that is the difference between apparently winning or losing.  And because the bias tends to run in one direction, you can’t help but think it is intentional, or that at a minimum the pollsters recognize the error but don’t want to figure out why and fix it.

There is no question that polling has gotten harder.  Finding a truly representative sample, particularly when many people simply won’t talk to a pollster or will lie, is very difficult.  You don’t know how those who respond and those who won’t differ from the actual total population.  But when your polls are consistently wrong you have to try a little harder than most pollsters are to get it right.  Despite every evidence of momentum toward Republican candidates, for example, the New York Times recently published a poll from Siena supposedly showing the Dem candidate ahead in critical Senate races.  I think we know what party controls the NYT.

In any event, here is a chart, borrowed from my friends at Powerline, that shows average pollster error in the 2020 election.  Everyone overstated the Dem vote, some by more than 5 percentage points.  One reason I semi-trust Trafalgar is that they seem to have figured out how to be more accurate.  Funny how most of the media-sponsored polls tend to be the most inaccurate.  In any event, don’t ever let polling influence whether or not you vote, get out there and make it count and makes sure all your relatives and friends and co-workers and acquaintenances vote too.