The rise of early voting data and the collapse of the polling averages
Midterm polling data is now rapidly becoming less reliable, thanks to a last minute deluge of Republican-commissioned polls muddying the waters. This is playing out even as the increasing popularity of early voting in key states is rapidly making early voting numbers more important. So you’d think the leading polling analysts would want to lead the way in early voting data analysis. But so far we’re largely seeing the opposite.
Leading polling analysts are now lining up to beg us all to write off early voting data, and by extension, to write off the rising crop of analysts who are trying to expertly analyze early voting data. Polling analysts are insisting early voting can’t tell us what polls can.
And who knows, maybe they’re right. Widespread early voting is an emerging trend post-pandemic. Meaningful data doesn’t go back very far, so there’s a smaller sample size. We won’t know until Election Day whether the 2022 projections based on early voting data are accurate.
But it is noteworthy that the polling analysts are all telling us to just ignore early voting data, as opposed to jumping in an helping us to figure out what early voting data really tells us. Because all this early voting data surely tells us something – whether we’re interpreting it correctly or not.
A cynic would say that the leading polling analysts are just trying to protect their position of authority by trying to suppress the rising field of early voting data analysis. If they embrace it at all, it might take off, and they might not be the top dog in the new paradigm.
On the other hand, what if early voting data really is even less reliable than the polls have become, and the polling analysts have dived in and seen this, and they’re just earnestly trying to warn us?
We’re going to find out soon enough. Polling analysts’ predictions have gotten incrementally less accurate each of the past few cycles, and are likely to be way off in 2022 due to them largely refusing to exclude the wave of GOP-commissioned polls from their calculations.
But will the early voter data analysis prove to be more accurate, or even less accurate, than polling averages? That’s the big question. If early voting predictions nail the 2022 results, that’ll be the new field. If it’s way off, is early data useless or just analyzed wrong?
In any case, when the elections results come in, judge the accuracy of the polls by the final polling averages, not any individual polls. And judge the early voting data predictions coming from experts like TargetSmart, not from every random clown on Twitter with a calculator.
In other words, compare the very best of the polling analysis to the very best of the early voting analysis, so we can find out which field is more viable at this point. Then we can address the fact that 90% of the analysis being offered in any given field is always crap.
Full disclosure, I’m rooting for early voter data to turn out to be accurate. For one thing, early voter data is more positive for the Democrats than the polls are. And also, with polling analysis going down the tubes, I’m hoping something reliable emerges in its place.
But really, although I’m a numbers guy, I’m not a participant in either field. I just like knowing that smart, trustworthy are crunching reliable data of some kind, and giving us some guidepost for knowing which races to put resources, etc.
And in a few days we’ll get election results that show us precisely what kind of data, and whose analysis of it, was on the right track. So that’s what I’ll look at when it comes to trusting polling analysts, early voting data analysts, or neither, in the future.
And whatever the results tell us, I would hope that everyone who’s made (or is looking to make) a name of themselves in either field, is willing to look at the accuracy of their own predictions and reevaluate their positions accordingly.
In the meantime, the best way to predict the outcome of an election is to create it. Spend this weekend volunteering for candidates. Phone bank. Knock on doors. Even if you wouldn’t usually do such things, and don’t see yourself ever trying it more than once, this is an opportune time to give campaign volunteering a try for a day or two and make a difference.
