What I’m Seeing: Uniform Swing and Thermostatic Politics
Charlotte Swasey, November 6th 2024
All caution about over-interpreting with limited data applies here. I’m using no exit poll results, sorry exit polls but you’re just not very good.
Uniform Swing
Election forecasters talk a lot about this concept, but I think it gets little mention elsewhere. It refers to the idea that elections are heavily driven by national factors that show up as smallish movement to one direction in every race. Essentially, if the national environment has gotten -1.5 points less Democratic, you’d expect to see that -1.5 point shift show up in every county, rather than seeing a random set of shifts influenced by local factors.
As far as I can tell, that looks to be what happened last night. One of the most clear indicators of uniform swing is movement in places where there wasn’t an active Presidential campaign occurring- Kentucky was an early (grim) indicator of this last night. In preliminary data, we’re seeing a roughly +6 margin shift towards Republicans in non-battleground states, with a slightly smaller one (+3 in margin maybe) in battlegrounds. Nationally, Biden/Trump was 51%-47% in popular vote, and this election looks to be Harris/Trump 48%-51%. That’s a shift from Dems +4 to -3, approximately.
There are some exceptions to this swing- NY, for example, looks bad in a way that this doesn’t fully explain, as does Florida, and suburban Atlanta may have moved against trend. Uniform swing does not purport to explain every result in full. What it does tell you is that you’re probably seeing a national problem.
[If you haven’t looked at the very good Washington Post graphic, you ought to, it visualizes this phenomenon really well.]
A Note on Battlegrounds
This election looks closer in battleground states than in non-battleground ones (something like a difference of R +3 vs R +6). If that’s right, that point is around the size of effect you’d see for a very successful campaign.
For people who don’t work in politics, that might seem hilariously tiny, but we know that campaign effects are *really small*. Two points would be a genuinely impressive effect for a campaign to have in states it was targeting, one point is very standard.
To me, this indicates that the problem isn’t with the campaign itself. This looks more like a national environment turning against Democrats than a bad campaign. Of course, campaigns are involved in shaping that national environment, but I don’t see anything to indicate problems with the on the ground mechanics that the campaign ran.
Exceptions from Uniform Swing
There’s clearly something over and above the national trend going on in Hispanic areas. El Paseo county in Texas saw around Trump’s margin improving by around 20 points from 2020. For a more rural example, Santa Cruz County Arizona saw the margin shift 15 points towards Trump. Florida as a whole is alarmingly bad, and Osceola county swung Republican by a similar amount to Santa Cruz. I don’t think exit polls give us enough information to pin down the exact state of Hispanic voters, but I would count this as pretty good evidence that there’s a larger-than-national swing against Democrats among Hispanics.
Urban areas also look unusually bad. California takes forever to count election results, so not much useful info from there yet, but New York had a shockingly bad night, with Queens swinging around 21 points margin Republican from 2020. The results in urban Nevada also aren’t great, with Clark County’s margin shifting 9 points Trump. It’s possible this is a symptom of a slide with POC, an extension of the Hispanic results discussed above, but it could also be something about urban areas in particular.
Turnout
Since votes are still being counted, I don’t feel great making sweeping pronouncements about turnout details. However, from what we’ve seen so far, this doesn’t look like a massive drop or increase in turnout. It looks fairly on par with 2020. We can dig into differences in particular areas once the data settles down a bit, but I wouldn’t expect a huge turnout story here.
What Uniform Swing Means
If we are in fact seeing fairly similar movement against Democrats everywhere, that indicates there’s something going on nationally that voters didn’t care for. I am fully ignoring issue polls here and pulling from political science fundamentals, because exit polls are a terrible way of measuring issue priority. [Issue polling is also quite bad a lot of the time, but that’s a longer explanation]
I think my best explanation right now is thermostatic response to Democratic governance. There’s a theory that voters behave “thermostatically”, essentially moving right when the government moves left, and vice versa. This matches well with what I’m seeing right now- Biden governed as a historically left presidential and the Democratic party as a whole moved historically left. A rightward swing from voters in response to this would make complete sense under thermostatic theory.
The other half of thermostatic theory is that it posits that this movement by the electorate is a way of setting policy to what they prefer, much like you turn the thermostat in your home up or down in response to temperature shifts. Under this theory, you’d read this result as the electorate telling us we’re too left.
Note that this doesn’t tell you literally anything about issue specifics- the sense from voters that a government is “too left” is entirely vibes, aggregated together from actual issue positions, media coverage, whatever they’re seeing, etc. This leaves a lot of room to figure out how to respond, since voters are usually quite disengaged on actual policy.
Fundamentally, these results mean there’s something Democrats have done or are doing that voters just do not like. It’s really unlikely to be an effect of the campaign, or of events that popped up, or of something the candidate said. This is pretty grim, and we should take a moment to reevaluate what we’re doing. We’ll know more as more data comes in, but this certainly feels like a rebuke from voters that we should take extremely seriously.
