Back in the day, daily TV ratings coverage was a very big deal. Each weekday morning, The Hollywood Reporter would write up the Nielsen overnights and try to find the most interesting story within the data, and this kind of coverage was a popular feature across all industry trade publications. But as the streaming and on-demand era took hold, how well a show performed on its very first night mattered less and less. So in 2020, THR moved away from writing up the overnights in all but the most compelling of cases. The numbers were becoming meaningless.
A similar thing seems to be happening in election polling, yet major media outlets still invest heavily over the course of months in polls that hint who might be ahead in Arizona, or what the spread might be in the national popular vote. And then their polls are combined with other polls to form a collective of polls that might be saying … something.
The polls got it very wrong in 2016.
The polls got it wrong in 2020.
The polls got it somewhat wrong in 2022.
Will the polls get it wrong in 2024? Who knows. It’s hard to even be sure what the polls are saying, exactly. Poll analysis master Nate Silver has chided pollsters for pumping out suspicious data sets, suggesting pollsters might be “herding” (which is apparently a nice word for “cheating”) to achieve an effective tie outcome to avoid looking wrong later.
There are reasons modern polling is struggling: Shy Donald Trump voters. Shy Kamala Harris voters? The difficulty of reaching voters in the cellphone age. Like with TV ratings, the landscape has changed and the numbers have grown fuzzy.
Polling has tremendous value for the campaigns themselves, of course, which engage in private polling. They need to know if their message is working and how to allocate resources to various states. This type of polling makes total sense.
But why does the general public need to see hundreds of polls in the media during an election cycle? What does that achieve, exactly? Particularly weeks and months in advance? I would argue polls are — and this is my excuse for writing about this in The Hollywood Reporter — an entertainment product. They’re not real news. They’re semi-scientific voodoo; so ephemeral that one study found they are only 60 percent accurate and are becoming less reliable. Yet they’re sponsored by some of the most prestigious media institutions, places where 60 percent accuracy — even 90 percent accuracy — wouldn’t come anywhere close to meeting the threshold for publishing a regular news story. But since these institutions are all invested in polls and getting traffic from poll stories, and we’re not, it’s easy for me to say: Polling is silly. It doesn’t do anybody any good — and might actually do bad.
Because what do the polls do? Before early voting begins, they just kinda seem to stress everybody the fuck out. And don’t we all have enough of that? Once early voting starts, if the polls show our preferred candidate is way ahead, then we might be less likely to vote (they’ve got this in the bag anyway). If the polls show our preferred candidate is way down, we might also be less likely to vote (eh, they’re going to lose anyway). If the polls say — as the polls this election cycle have maintained — that the race is super-duper tied, then they don’t actually do the one thing you want from a poll — tell you who’s winning.
This last scenario is the one polling defenders point to — that a close poll can motivate people to vote. The problem is this assumes the poll is correct when they’re often not, so it might have exactly the opposite effect. And it still doesn’t justify the months of public polling before voting begins. And, since everybody ignores polls when they turn out to be right, yet blast polls when they’re wrong, they arguably contribute to eroding public trust in media since, fairly or unfairly, the misses make a lot more noise than the hits (the polls in 2022, for instance, got a lot right, but they still get lumped in with 2016 and 2020 for the things they didn’t). And we’re not living in a time where “getting some stuff wrong” should be okay.
And just imagine what the last few months would have been like if you hadn’t seen a single poll story, and instead just got, you know, news coverage and opinions about the candidates. Still awful, sure. But less awful? Instead, the polling community (The Polling Industrial Complex? Big Polling?) seem to be obsessed with figuring out how to make polls more accurate after each embarrassing public failure — what additional weighting and methodology and tweaks they can do to fix all this. As if it’s very important to fix all this, without ever stopping to wonder: Wait, do we need to continue doing this at all? “So preoccupied with whether or not that they could, they didn’t stop to think about whether they should,” etc.
So whomever emerges victorious today, moving forward let’s all agree to vote with our attention and no longer read poll stories. If we all do this, then they will go away and bother us no more. Granted, from surveying my friends, most of whom did not answer the phone, the chances that this will work is maybe 2 percent, and that’s within the margin of error.
Read the original article here