Today’s Entry: @brianmuddradio Is the difference from polling to actual results any different now than in 2010? I feel like many polls are less reliable lately.
Bottom Line: Love them, hate them, maybe both...you can’t escape them if you’re paying attention to the midterm elections at this point. I’m of course talking about political polls. Now, today’s question comes on back of this week’s midweek midterm election update and specifically something I’ve highlighted throughout the course of this cycle. Generic ballot polling. If you’ve followed my coverage of this election cycle, or others previously for that matter, you’ll know I’m not one to get hung up on talking about the latest poll results as though they’re news or newsworthy for that matter. The closest I get is highlighting in passing, Governor DeSantis pacing a double-digit advantage in an average of the polls for example. And there’s a reason for it. We all know many polls have been highly unreliable. If I’m sharing any kind of specific polling with you as though it’s relevant, it’s because I’ve vetted it as being reasonably reliable.
As I mention in my weekly election update, there’s one polling metric which has been accurate in every midterm election cycle since the advent of accredited polling age in the 1930’s. The generic ballot. Whatever party has had the generic ballot polling advantage in an average of available accredited polls, has fared best in midterm elections overall. That’s a 22-0 record, which is even more impressive than the Dolphins perfect season – therefore I think it’s relevant conversation in discussing what’s going on leading up to Election Day. But just because the party favored on the generic ballot has been accurate – doesn't mean the actual percentage represented has been.
Today’s question is in reference to 2010’s generic ballot polling as I’ve compared where we stand in this cycle to where we stood at the same time during that cycle, which resulted in the largest number of seats flipping in one election cycle in American history. As I noted in this week’s update, Republicans are pacing a 3-point advantage while on the same date in 2010 the advantage was 7%. So, in answering your question about how accurate the 2010 generic ballot polling was and whether it’s been less accurate more recently...here goes.
On Election Day 2010 the average generic ballot poll result showed a Republican advantage of 9.4%. The actual result were Republicans obtaining 6.8% more of the vote nationally compared to Democrats. In other words, twelve years ago, the pollsters overestimated the GOP’s performance by 2.6%. That’s notable, because conservatives generally think the polls are overly favorable for Democrats. Obviously, that’s not always the case. As for the more recent cycles... In 2014 the generic ballot polling average understated Republican support by 3.3%. And most recently in the 2018 midterms Democrats were underrepresented by 1.1% in the polls. So, here’s what we know, which is also a bit of narrative buster for some on the right. The average generic ballot polling is predictive of which party fares best, however, is off by an average of just over 2% against the actual results, and notably Democrats have been the most likely to be underrepresented in the midterm polls. Also, in addressing another part of the question, the most recent midterm polls have been the most accurate of the past three cycles, so there’s no indication they’ve become less accurate more recently.
As always there are two sides to stories and one side to facts. Those are the facts. None of that guarantees where the ball will land with the polls and the outcomes this time around, but I am confident the generic ballot poll record for predicting the party which will win the most seats will be extended to a perfect 23-0.
Each day I feature a listener question sent by one of these methods.
Email: brianmudd@iheartmedia.com
Gettr, Parler & Twitter: @brianmuddradio
iHeartRadio: Use the Talkback feature – the microphone button on our station’s page in the iHeart app.
Photo: Getty Images