Following up on the post that we linked to last week, here’s Yair’s analysis, using Mister P, of how everyone voted.
Like Yair, I think these results are much better than what you’ll see from exit polls, partly because the analysis is more sophisticated (MRP gives you state-by-state estimates in each demographic group), partly because he’s using more data (tons of pre-election polls), and partly because I think his analysis does a better job of correcting for bias (systematic differences between the sample and population).
As Yair puts it:
We spent the last week combining all of the information available — pre-election projections, early voting, county and congressional district election results, precinct results where we have it available , and polling data — to come up with our estimates.
In future election years, maybe Yair’s results, or others constructed using similar methods, will become the standard, and we’ll be able to forget exit polls, or relegate them to a more minor part of our discourse.
Anyway, here’s what Yair found:
The breakdown by age. Wow:
Changes since the previous midterm election, who voted and how they voted:
Ethnicity and education:
Urban/suburban/rural:
>
Yair’s got more at the link.
And here’s our summary of what happened in 2018, that we posted a few days after the election.
Should the percent of electorate times the vote margin come out to the national vote margin? I checked the urban/suburban/rural split and the age group split and they multiply out to an overall margin of ~5.5 to the Democrats. But the reports I’ve read of actual vote counts have the Democratic margin at more like 7 and likely to increase as western states keep counting mail-in ballots.
Alex:
From Yair’s post:
In general, I think it makes more sense to look at average district vote rather than total vote, as average district vote gives a better sense of nationwide support for each party, to the extent that we think of the people who chose to vote in a given district as representing public opinion within the district.
I couldn’t find a reference until they posted a new story today (good or bad, I get pretty much all my quantitative political info from 538), but the margin seems to be over 7 even when you take out uncontested races – see footnote 3 and text around it https://fivethirtyeight.com/features/everything-is-partisan-and-correlated-and-boring/ .
To your general point, that same article notes that 538’s partisan lean measure, which is the average voting record for a district (if I’m understanding you and them correctly), is sufficient to pretty accurately predict the 2018 results when you add on a generic polling average. That, in turn, is hardly any different from the total vote: both are ~ +7.5 Democrat right now.
So am I misunderstanding Yair’s point about dealing with uncontested races, or are those results still just notably below the actual vote counts?
Can we estimate the breakdown by IQ?
And there must be a genetic factor. (jk)
Is the “ethnicity and education” table missing non-white non-college voters or are the “Non-White College Men” and “Non-White College Women” rows supposed to be “Non-White Men” and “Non-White Women”? The share and margin change graphs following would imply the latter.
When you write “Percent of electorate” do you mean “Percent of voters”. Isn’t the electorate the total number of people eligible to vote?
“The breakdown by age. Wow:”
Why “Wow”? (I think I see why, but I’m just not sure, so I would like to see it spelled out.)
Terry:
Wow because the voting patterns for different age groups are so different. It didn’t use to be that way. It’s only in the past decade or so that we’ve had this pattern of young people strongly supporting the Democrats and old people strongly supporting the Republicans.
An interesting thing seems to be that, apparently, the “flower power” generation now is the most solid Republican age group? For someone who grew in the 1980s, this is a curious reversal – in these times, the cliche was the 20-something yuppie Reaganist versus the 40 y.o. ex-hippie “liberal” (some TV shows were even built on that premise, like “Family Ties”). Now, the former Republican yuppies of the 1980s will be the “45-64”, and the former radicals of the 1960s will be “+65”.
Well, perhaps the former hippies are “Urban 65+” and the former yuppies are “Suburban 45-64″…
Could you also use a synthetic US population to calculate turnout for arbitrary geographies and/or demographic groups? The main code I’m familiar with is SynthPop (https://github.com/UDST/synthpop) although I can’t tell if it works with recent versions of census data. There’s also a pre-built synthetic population here (https://fred.publichealth.pitt.edu/syn_pops) but it uses 2010 census data.