Archive

Show more
Labels
Posted by

Thirty-two points


Thirty-two points. That is what the New York Times is trying to sell us as the swing in independent women from Democrats to Republicans in its latest poll. Nothing has occurred in the past month to engender this swing. There have been no Democratic scandals. The economy is doing as well as it had before. Inflation is holding steady. But we are to believe that for no apparent reason, independent women shifted BY THIRTY-TWO POINTS from Democrats to Republicans.

I've written this before: our polling models are broken. For example, here are the various House forecasts, as collected on Wikipedia:



As you can see, there is quite a divergence, with only Decision Desk and RCP casting the Republicans as guaranteed to take the House. RCP, of course, tilts very right, and Decision Desk is usually all over the place. But what stuns me is the assuredness that in the wake of Dobbs, anyone can say with surety what's going to happen. The Economist seems to get closest to the unprecedented nature of this election, showing Democrats and the GOP more or less deadlocked.

The sample you get is what is going to influence the poll you put out. And to me, a sample which produces a 32-point swing in one direction or another with nothing really having changed in the political landscape raises huge alarm bells. I think Stuart Rothenberg says it best:


Polling used to be an accurate gauge of public opinion. But that was in a different universe. A universe with land lines and without caller ID. A universe in which people always picked up the phone. A universe in which the media landscape wasn't so fragmented and cut-throat that anything which generates clicks is seen as good for the bottom line.

The 2016 election put the fear of God into pollsters. They missed Trump's election—even though they got the overall popular vote right. They thus overcorrected for 2018, and nearly missed the blue wave of that year's House elections. Frightened again, they again overcorrected for 2020, and missed the GOP doing better than it was thought it would. And now we're seeing this again with 2022. The actual special elections which we've had have shown Democrats overperforming relative to the polling. New registrations since Dobbs have shown Democratic strength. Even the polling shows no red wave, but a closely fought contest. Chuck Grassley wouldn't be up by only three points in Iowa, no liberal bastion, in a wave election.

And yet, we're supposed to believe a 32-point shift in one key demographic because inflation is supposed to rule all. Right now, pollsters are blind shamans trying to discern the future by casting divining bones. The polling industry is chasing its tail trying not to look foolish, rather than trying to accurately gauge the public mood. Another reason why this is a different universe than the one in which polling was predictive is due to this country's growing diversity. Perhaps polling was easier when the population was 85% white. That's no longer the case.

Polling with such a wild swing is worse than useless. It does real harm. But it sure would make the media's job easier if those kooky Republicans were to win the House and Senate. The stories would write themselves, and the country defaulted on its debts and four Hunter Biden investigations were ongoing.

Post-script

When I got home last night, this tweet, part of a larger thread, leapt out at me:
This encapsulates the problem we have.

The media has abdicated responsibility for reporting reality. What it does now, at lead in the political press, is to conduct surveys to see what people think, and report that as news. That, needless to say, is more than useless. It's actively harmful. The media serves as a feedback loop for what people "feel", even if those feelings are not backed by empirical data. I suspect this is the case because, again, in a fragmented media environment, every eyeball and click means the difference between profit and loss. And of course people love having their views validated by organizations like CNN and the New York Times. To hell with the purpose of the media, which is to report and analyze accurately. It's much easier to spend money on polling everything under the sun, and then reporting on those polls as if they're the news.

This polling frenzy began during Barack Obama's presidency, when the novelty of the nation's first Black president pushed the media to conduct weekly, if not daily, polling to see how "America" felt about a Black man in the White House. And now we are where we are.

Then there's this:
Essentially, we are subject to a tyranny of data, and the data is all over the map, leaving us confused and anxious. Vox populi, which polls are supposed to provide, instead tethers us to dread. Polling, far from quantifying the mood, has no idea what the public mood is. It's time to end the era of mass polling. It's bad for democracy, and bad for mental health.
Labels: