Polling Review: a spectacular failure
Wow! What a week! And what a spectacular failure of the pollsters. There is going to have to be a very deep review of the polling process to figure out how to change things for the current American emotional makeup. Something!
I’ve been searching for reviews on what went wrong during the polling because I think most polls indicated a very close race and it ended up not being neck and neck. There was a clear winner.
I was most interested in how Nate Silver of 538, Allan Lichtman’s 13 keys and Ann Selzer Iowa polls did, so let’s see how they did.
Nate Silver 538 polling – the closest
Nate Silver actually does a meta-analysis of all polls to arrive at the national level and also the state level. He doesn’t do the actual polling but instead do a study of all of the polls, placing greater weight on those pollsters who are more reputable and have a better history of success. I think he does some kind of adjustment when he does this meta-analysis.
His prediction was “the data tells me that this is going to be a toss-up”, but his gut said that Trump was going to win. He did say not to trust anybody’s gut, including his, and it turned out his gut was right. Out of the three pollsters, he came the closest.
Okay, I’m reading an article that said that Nate took down the polling result from the 538 site, predicting a Harris win. Okay, if that is true, then he was not the closest. And the rest of the article indicated that if he had to take a guess, since the polls were so close, he would predict Harris. So, I don’t know which is true.
My research indicates that I remember the earlier prediction in October, and he changed it just before the election, so that is the discrepancy.
Right now, I can’t find any assessment of why his meta-analysis failed but I would hazard a guess that if the many polls he studies are off, then he would be off too.
Allan Lichtman’s 13 keys – failed
Allan Lichtman stated very confidently that Harris would win. His keys are not polls but are instead an analysis of how the white house governs. His thesis is people vote based upon how the white house governs; polls are just noise.
That logic makes a lot of sense, and he has historically been very successful, maybe since 1984, with only one miss (the 2000 Gore-Bush contest that was allegedly decided with a heavy thumb on the scales from the Supreme Court).
If I understand correctly, his assessment of where he went wrong lies not with how he interpreted the keys but on the unique situation in this election season: Biden’s late season drop out, Harris obtaining the nominee status without undergoing the primaries, and the extreme disinformation environment.
Will he change his keys? He says no because the keys have worked for roughly 41 elections, including those before he began the process. At present time he won’t be changing.
Ann Selzer Iowa polling – off too
Ann Selzer’s polling is focused just on Iowa; thus, its implications really does not impact the national level other than some states might be similar to Iowa. When her poll came out Saturday, just before the election, showing Harris winning by 3 points, it sent shock waves and frankly, excited the Democrats. I saw one just bouncing in his chair with glee as he was reporting on her poll.
Her polling is regarded as the gold standard since her polls are rarely wrong…except this time.
Her process makes sense: she doesn’t rely on the historical voting patterns, most specifically the last vote 2020, to determine who is likely to vote. I believe she tries to poll the likely voters to get a feel of how they may vote. She tries to assess who is likely to vote in this current environment, not based upon 2020 environment. A lot of other pollsters use 2020 election cycle to determine the likely voters.
That made sense…but Iowa went to Trump, and it was a resounding win.
As far as I can tell, her initial assessment of where she might have gone wrong is that those small set of undecided voters broke for Trump at the last moment or some RFK Jr voters moved towards Trump, again at the last minute.
Or maybe the findings of her Saturday poll (where Harris beats Trump) activate the Trump supporters.
Other pollsters – abject failure
I think most of the other pollsters predicted a very close race but probably leaned towards Harris. It is hard to tell because I was not tracking those other pollsters. Certainly, there was no indication that Trump would win by a healthy margin.
Now, I haven’t heard of causes for their failure, but I do have some ideas:
- To aid in the determination of likely voters, they may have depended on the historical 2020 voting patterns to develop their pool of likely voters. But 2024 probably turned out to be very different: Latinos and to some extent Black men moved towards Trump, and young bros turned out – probably after the Joe Rogan podcast.
- Pollsters have their own weighing methods which might have become outdated.
- They may have also been nudged toward predicting in the same realm as other pollsters so as not to stick out. Their reputation is on the line, so they don’t want to stick out too much.
- Some voters may not have been truthful and said they were voting to Harris, knowing that Trump was a very unpopular candidate. The bad halo effect of being perceived as a Trump voter probably induce them to keep their mouth shut or to lie.
Concluding remarks
So, the election cycle has ended with a Trump win. Now we are going to face another set of tensions which has already broken out: some text messages went out to blacks telling them to report to some “plantation” to become slaves and the ugly bros came out chanting, “your bodies, my choice”.
Yep, it looks like it is going to be ugly.
