Their Crystal's On The Fritz

Crystal-gazing.

Finger on the pulse.

Finger in the wind.

Finger down my throat. 

Jennifer Rubin, polling the pollsters. Or trolling them - trolling the trollsters?

(pay wall)

Opinion
4 reasons to be skeptical about election polling

Pollsters should worry that their profession might soon be regarded as more like astrology than political science.

Reasons to take polls with a large grain of salt abound. First, the nagging sense that pollsters are “missing” MAGA voters remains, as was the case during both the 2016 and 2020 elections. In both cycles, pollsters essentially got the Democratic share of the vote right but missed many voters who supported Donald Trump.

Some have suggested that Trump voters are less likely to talk honestly with pollsters than Democratic voters are. Plus, pollsters get responses from only a tiny fraction of calls they place, so voters with the luxury of time and deep interest in politics skew the results.

In 2020, polls gave a false sense of security for many House and Senate Democratic candidates. But that polling failure did not occur in 2018, when Trump was not on the ballot. So is the worry of missing Republican voters valid this time around? It’s a source of serious debate.

A second reason for caution: Some pollsters have reacted to their previous errors by overweighting survey results in the opposite direction. It’s unlikely that Republicans have made up their deficit among women, for example, despite what some polls show. Similarly, it’s unlikely that younger voters will show up as a fraction of their proportion in the 2018 or 2020 elections. Polls that show otherwise might be an indication that pollsters have overcompensated.

Third, as early voting becomes increasingly popular, no one knows whether this behavior will affect voting outcomes or whether the past profile of early voters (heavily Democratic) will hold up. There is no doubt that early voting has become easier and more familiar for millions of voters, as the Center for Election Innovation and Research points out in a recent report. David Becker, who leads the center, tells me that “35 states offer every voter the choice to vote early or by mail, and another 11 offer early voting to all voters (requiring an excuse to vote by mail).” While Republicans have certainly tried to restrict voting, he said, doing so might be hard when almost every voter “will still find voting to be familiar, convenient and safe.”

So far, early voting is slightly ahead of the 2018 rate and about 37 percent of 2020’s rate. The Democratic percentage of early voting is higher than it was in 2018 and 2020, but it’s unclear whether this means Democratic turnout as a whole will keep pace.

Finally, Republican pollsters are flooding the zone with partisan polls, which polling averages pick up. Naturally, that means mainstream media outlets are seeing these numbers and concluding that Republicans are gaining steam. But are they?

Tom Bonier, a Democratic strategist and head of the progressive data firm TargetSmart, noted the phenomenon in a recent tweet thread. He argues, for example, that one GOP poll showing Mehmet Oz, Republicans’ Senate nominee in Pennsylvania, up by three points likely underestimated young voters, but “many media outlets reported that Oz had surged into the lead.”

Similarly, in Georgia’s Senate race, Republican nominee Herschel Walker has the edge according to media outlets, despite reports that he paid for abortions previously (which Walker denies). But media reports are likely skewed by the large number of GOP-associated firms with polls showing the Republican ahead. Democratic operative Simon Rosenberg tweets: “A polling aggregator of only independent polls has the election 3.3 pts more Dem than” the RealClearPolitics average.

Perhaps the common media take that Republicans are “surging” is misleading. After all, while the New York Times’s election pundits have argued that Republicans are gaining ground, its own polling shows Democrats leading in three of four critical House races and a batch of Senate races (including Georgia, Pennsylvania and Arizona). So which is right? The analysis, the polling or neither?

The bottom line is threefold: First, polling is far less predictive than media outlets often portray it to be, largely because we do not know who will turn out. CBS News has wisely provided a range of polling results depending on different turnout models. That seems far more honest and informative.

Second, much of the horserace coverage reliant on polling is crowding out substantive coverage of the races. Outlets should adjust accordingly.

Third, we’ll find out the winner only after the votes are counted (shocking!). But even that won’t tell us whether the polls were “right,” or whether they contributed to a false narrative that drove results.

In the end, the only sane approach to understanding the election is to ignore all the polling noise and focus on what really matters: turning out to vote.

All I want for Christmas is a chance to tell Nate Silver to shove a pinecone up his ass.

(And don't kid yourself, these GOP asshats are going to drag this shit out for quite a while. We'll be lucky to get most of it done by Christmas.)