September 9, 2020

Another look - why the polls don't scare me

Intuitively everyone on the right understands that Clinton was crushing president Trump in the polls in 2016 and still lost.  I've been looking at the polls in 2016 by month. After filtering for likely voter polls in the RealClearPolitics polls for September 2016, and comparing those results to the final outcome of the presidential election there is a consistent bias in the results towards Clinton.  Looking state by state you can see the bias in the polling leaning heavily towards Hillary Clinton:

As you can see there were 13 states in September that exhibited a polling bias towards Clinton, two that exhibited no bias and two that exhibited a bias, or error, towards president Trump. Some of the bias was pretty dramatic at this point in the race.

Some of these states had the bias tighten later in the race.  Some did not.  The point is the polls now are either biased or inaccurate compared to what they will like like the day before the election.  One counter-argument could be that the voters' choices changed during that time as well.  Fair enough - there was all of that James Comey garbage of the email investigation back and forth going on.

But if you look back at this point in the race there was a media exuberance that Clinton was going to win 350+ electoral college votes.  Remember this?

Isn't that fun to watch in retrospect?

Even if the pollsters have made adjustments and accounted for 75% of those biases (which is a stretch since there is little evidence that is the case), there's still a swing of a couple of points towards president Trump and that impacts his chances of re-election pretty significantly.

FiveThirtyEight has rated various pollsters as having a bias towards Democrats or Republicans or neither, but that is a more generic bias and understates the bias I have found.  It's possible both biases apply.  Most of the FiveThirtyEight polls seem to have benefited Democrats.  While there is potentially some overlap between their identified biases and my 2016 pollster biases, the overall impact is to Trump's benefit and potentially more than my cautious view of granting pollsters a 75% correction this time around.  The polls don't scare me.

No comments:

Post a Comment

Disagreement is always welcome. Please remain civil. Vulgar or disrespectful comments towards anyone will be removed.

Related Posts Plugin for WordPress, Blogger...

Share This