What Happened on Tuesday?
Media Myopia and Misreading CrossTabs
Some Red Tsunami, huh?
It’s been days since the Senate was decided and days more before we have a final outcome on the House. But even after a week, certain things are becoming clear. More research will no doubt put more nuance to what we know at the moment.
First, let’s lay off the pollsters (at least most of them — the partisan leaning polls were as bad as always.) I read today that the New York Times/Sienna polls fell consistently within the margin of error on all major contested Senate races.1 Aggregator averages appear to have been skewed by those late-stage partisan lean polls. Things were always close on the generic ballot question (which party would you favor).
Second, the projections based on polls always require a model against which the data is evaluated. Does the makeup of the electorate look like the 2018 midterms or the 2004 midterms? Simply assuming that the same configuration holds in this cycle that operated in prior cycles creates error.
Exit polls show that this electorate was sharply divided by age. Those under 45, Democrats won about 55% of the vote. For those 45 and over, Republicans won the same percentage. Another intriguing data point I saw was that the under-30 vote offset the over 65 vote almost exactly.
Third, the Dobbs decision on abortion played a key role in both turnout and supporting Democrats. This is especially the case in those states where abortion was literally on the ballot regarding constitutional amendments (either enshrining rights or abolishing abortion). In Michigan, Kentucky, and Montana (plus the earlier vote in Kansas), this was huge.
Fourth, the dominant media narrative was superficial and incomplete. It normally went something like this:
Midterm races are always a challenge for a president’s party, and especially so when people are so concerned about inflation and gas prices. Whenever a party has faced such headwinds they have lost 40 seats in the House.
The first thing to notice is that we’re talking about a very small number of prior elections to make inference. For all the comparisons of today’s economic environment with Jimmy Carter’s, there have only been 11 midterm elections from 1978 to 2022. Not a lot to see big patterns.
The media narrative was flawed in an even larger way. They presumed that issues potential voters identified as serious issues would directly motivate their votes.2 It would motivate them to vote against the incumbent party who could be held responsible for inflation and gas prices. This theme was picked up by all the conservative commercials that talked about "Joe Biden's inflation" and tried to argue that somehow deficits and spending caused inflation.3
Democratic strategist Joe Trippi (campaign manager for Howard Dean’s 2004 campaign and now co-leader of The Lincoln Project) had a fascinating Twitter Thread yesterday. It’s worth quoting in full:
So the outcome of this election and why so many got it wrong is what happens when data misinterpreted happens because pundits rely on data without listening to people. I learned more in focus groups since 2016 than from any poll. 1/4
Smirk at focus groups. But you hear things from people that informs the data. Like groups I saw of suburban undecided white men. All complaining about inflation and gas prices. 2/4
And when we asked what or who was responsible for higher costs. Not one person in 4 groups blamed Biden. “COVID” “supply chain” “war in Ukraine” were there answers. Was inflation their top concern? You bet. But they were smarter than the media gave them credit for 3/4
We call it cross pressure. You are worried about inflation and may even blame Biden. But you also think MAGA is too extreme and you are sick of the chaos. But the herd looks at those inflation numbers and KNOWS how you are gonna vote. Predicts the Red Wave and swings & misses.
The media narrative assumed that the voters would blame Biden and the Democrats for the economic situation. The voters, in Trippi’s analysis, recognized that inflation was a global phenomenon and that supply chain issues and the Russian incursion were more responsible for our current situation than the last round of stimulus funds approved in March of 2021.
Finally, democracy matters. It is notable that nearly all of the election deniers on the ballot lost (save for Indiana and a slim possibility of Kari Lake). Tom Nichols tweeted this in sharing Trippi’s analysis:
Interesting. “Inflation is out top concern” does not mean “and therefore I’m voting for kooks”.
One final point that speaks to why it’s hard to maintain a clear view on this issue: Conservatives apparently cannot read crosstabs for what they are. Crosstabs allow one to look at how subsets of the population voted. Given what I said above about the youth vote and concerns over Dobbs, it shouldn’t have been surprising that single women voted overwhelmingly for Democrats last week.
I feel bad for picking on Jesse Watters because it’s so easy. But he, and many other conservative voices, have taken this descriptive data and tried to turn it into causal data. In other words, unmarried women didn’t vote for Democrats because that subset of the population differed on issues of policy. It was because they were single. If only they were married, they’d have voting patterns like the other married women.
This is, of course, ludicrous. If those single young women all got married tomorrow, they’d still have the same policy interests that they had when they were single (and the percentage of married women voting Republican would go down).
Babylon Bee editor Joel Berry made Watters seem ponderous. He tweeted the following on Friday:
Unmarried women in America are lost, miserable, addicted to SSRIs and alcohol, wracked with guilt form abortion, and wondering from partner to partner.
If conservatives insist on reacting like this, watch those Single Women support Democrats percentages skyrocket in coming years.
It is, of course possible, for conservatives to conduct a 2012-type autopsy (which was ignored) and figure out policy positions that would be better suited to appealing to young people in general and young women in particular.
On the other hand, maybe they’d rather repeal the 19th and 26th Amendments granting (white) women and 18-21 year olds the right to vote (Alexandra Petri explains the latter). Maybe we should add “property owners” while we’re at it to get back to the vision of The Founders.
It’s still a problem that people can’t grasp the idea of margin of error. If candidate X is up by 4 points with an MOE of 5 points, then X’s outcome may actually fall between being down by 1 to being up by 9 (at the time the poll was taken).
I have to note here the circular logic of media reporting on polls. They write stories about how bad inflation is for the average household (buying four gallons of milk a week). Then when polls show people identify inflation as a major concern (because they’ve read all those stories), the feel confirmed in their narrative and retell the story with new data. Same thing happens with crime and the border.
Senator Mike Lee quoted Milton Friedman to this effect in one of his debates. Nobody checked on whether that claim was borne out empirically.