Wednesday, November 8, 2017

Virginia Polls Missed by More than National Polls Did Last Year (But Few Seem to Have Noticed)


A year ago, a lot of left-leaning voters expressed their anger at statistician Nate Silver because his FiveThirtyEight election model gave Donald Trump "only" a 30% chance of winning. They assumed that someone with only a 30% chance of winning really had no chance of winning. Such a belief, however, reflects the fact that most of us aren't trained to think probabilistically. However, as I noted back then, a 30% chance is actually pretty high ("Don't Blame Nate Silver"). One way to cultivate just how high a 30% chance is, is to have you imagine a scenario in which bullets are placed in two of a revolver's six chambers. How many of you would be willing to spin the revolver's cylinder, place the gun to your temple, and pull the trigger? My guess is not too many even though there's "only" a 33% chance that you'll put a bullet through your head.

Unlike most of us, however, Silver does think in probabilistic terms. In fact, he repeatedly cautioned readers about overconfidence, noting that Clinton's leads in the battleground states were slim and within the range of sampling error, which meant that if the polls were slightly overestimating Clinton's support, the election could break in Trump's favor. Which, of course, is exactly what happened. But many people still blamed Nate Silver.

Silver wasn't the only one who was criticized. Polling agencies were too. However, the 2016 polls weren't too far off. Their average predicted margin of a Clinton popular vote victory was a little over 3.0%, which was only 1.0% higher than her actual (popular vote) victory of 2.1%. For a presidential election, 1.0% miss is remarkably close. In fact, the 2016 polls outperformed the 2012 polls. However, few noticed (or cared) because in 2012 the polls underestimated the Democrat candidate's popular vote share, while in 2016 they overestimated it.

Interestingly, last night in Virginia's gubernatorial election the polls performed worse than the National polls did in the 2016 presidential election. They had Democrat Ralph Northam winning by approximately 3%, but in the end he won by almost 9%. However, hardly anyone has complained (or noticed), and as far as I can tell, no one has blamed Nate Silver (one of the few who has noticed).

No comments:

Post a Comment