Analysis: Will political the political shifts we saw in 2020 continue in 2024?

[ProgressiveModerate]

In 2020, there were quite a few notable political shifts that defined the election. Mainly:

1. Trump making massive gains with Hispanic, Asian, and lesser degree black voters
2. Biden making gains with college educated voters
3. Biden making gains in fast growing metros and their suburbs, especially in the sunbelt
4. Biden making big gains in the northeast/Atlantic corridor
5. Trump squeezing a bit more out of many white working-class communities where he saw massive gains in 2016

I think oftentimes people assume that most shifts between 2 election cycles will continue. I tested this hypothesis by comparing the 2012-->2016 Presidential shifts to the 2016-->2020 shifts by Congressional district, and this was the result:


(Note positive values indicates a leftwards shift between cycles, negative values a rightwards shift).

The correlation is a lot weaker than many might assume.

Interestingly there are a lot of contrarian districts in the top left of this plot that shifted hard left in 2016 (despite the Country shifting right) and then hard right in 2020 (despite the Country shifting left) - most of these were heavily Hispanic seats in the sunbelt states like FL, TX, and CA.

Still, you can see the large majority of seats seem to fall into a somewhat weak positive correlation. Basically all the seats that fall into the top right quadrant of this graph (seats that shifted left in 2016 and 2020) were those suburban districts, many of which have experienced rapid growth over the past decade.

In the bottom right, you have districts that shifted right in 2016 but then left in 2020 - quite a lot of districts fall into this category, following the nation at large. A lot of the districts in this area are relatively rural/exurban/small-city districts in the northeast.

Then in the bottom left, there are districts that shifted right in both cycles. This is a pretty interesting collection of rural and urban districts, but they're all districts that are relatively and have low college attainment, alongside many having a stagnant or shrinking population in the 2020 census.

In writing this, I'm noticing a theme between the districts that shifted the same direction twice vs the districts that shifted different directions between elections - demographic change vs persuasion.

The districts that fall into the top right and bottom left quadrants (double leftwards shift and double rightwards shift) are seats that mostly have a demographic explanation that explains a significant part of there shift both cycles. In the case of seats like GA-07, TX-03, or VA-10, all 3 seats have seen rapid growth since 2010, large increases in their non-white population, and increases in college attainment. These factors likely explain a large part of the shift - looking at vote totals backs this up. In TX-03 for instance, Trump 2020 got 205k votes to Romney's 154k, however Biden got 152k to Obama's 68k - that's well over doubling raw votes for Democrats even as Republicans gained more votes - growth is clearly a huge part of this reason why this district went from Romney + 39 to Trump + 14 in just 8 years. All else being equal, one should expect many of these suburban districts to swing to Biden again. On the flip side, many of the seats in the bottom right quadrant are losing population, often population that's favorable to Democrats. NC-01 is a good example - over the past decade the % of people in the district who are black has declined (alongside the overall population of the district), explaining at least part of the rightwards lurch this seat has taken. If all else was equal in 2024, we'd expect this seat to swing right by default from population changes.

On the flip side, the districts that have shifted because of large amounts of persuasion are a bit harder to pinpoint and predict because persuasion is much more fluid than population change. TX-29 is a great example - it went from having a 15% shift left in 2016 to a 13% shift right in 2020 - this was largely due to the district's large Hispanic population. There's sort of an irony here because despite the national swing, there were far more leftwards in 2016 --> rightwards in 2020 districts caused by persuasion which shows how persuasion on a more local level may not align with the national picture. Trying to predict swings from persuasion is trickier because oftentimes these swings are far more aggressive and volatile than swings caused by demographic shifts (swings caused by demographics don't rarely seem to exceed ~12% at the Congressional District level between given cycles). One thing that generally seems to be true about these types of persuasion congressional districts is they tend to have very unique cultural dynamics (Cubans in Miami, Mormons in UT, WWC in the Mahoning Valley) that may make them more susceptible to targeted messaging.

To be clear, these two types of swings can happen at the same place at the same time; in VA-10 for instance there probably were a decent number of colleges educated Trump-->Biden vote flippers, or in TX-07 where in 2020 Trump's large gains with Vietnamese and Hispanic voters was able to offset the favorable Demographic changes in the district to Democrats.

I think a general lesson is yes, the Biden's suburban swings are generally a lot more stable than the gains Trump make with Hispanic and Asian voters going into 2024, but Trump also has more potential to make gains with Hispanic or Asian voters that were larger than 2020.



TLDR; there are two things that power swings; demographic change and persuasion. Demographic changes tend to be pretty consistent between cycles and hence swings caused by demographic changes should be pretty consistent between cycles. Swings caused by persuasion are much more of wild cards, with large inconsistencies between cycles, and often causes swings of very large magnitude making the persuasion swings much harder to predict.

[LabourJersey]

I'm unsure how many of these trends and shifts will remain.

I wonder how many of them are more the result of the pandemic - from decreased Dem turnout due to no ground game, from minorities concerns on lockdowns, etc - and how much is actually change.

Good analysis, though.

[DaleCooper]

The population decline is part of what contributes to the nihilism that you have to have in order to support Trump as a non-evangelical. Omaha and its suburbs are growing, but almost everywhere else in Nebraska is practically hemorrhaging population. If you go visit one of these dying towns, I can assure you that it'll be one of the most depressing experiences you've had all year. There's no opportunity. These aging people's kids have all moved away, and the few that have stayed are either alcoholics or worse. You almost wish that a tornado would just come blow the town away and put it out of its misery. They're not voting Trump hoping that he'll bring life back to their community, they're hoping he'll bring down the rest of the country. These people are past hope, they're just angry.

[seskoog]

I would say I think most of them continue, but you never know. Most shifts from 2008-2012 bore no resemblance to shifts between 2012-2016 and 2016-2020, with the only areas to swing left all three times being areas with rapid demographic change and the only areas swinging right all three times being rural. And the magnitude of each shift was different each time.

[ProgressiveModerate]

I'm unsure how many of these trends and shifts will remain.

I wonder how many of them are more the result of the pandemic - from decreased Dem turnout due to no ground game, from minorities concerns on lockdowns, etc - and how much is actually change.

Good analysis, though.

I think most of the voters who swung to Trump because of frustration with lockdowns have largely stuck with Trump at this point. If lockdowns were enough to flip your vote, then you had some real anger around then, and since then that anger has gone through an entire rollercoaster of right wing politics.

Dem turnout thing will be interesting to see - nationally the only substantial communities to see turnout declines relative to 2016 were urban black communities, communities where ground game is historically important. However, it's tricky because there could be other factors that explain these declines - for instance the Obama years being over so black voters generally being less energized.

[ProgressiveModerate]

I think one thing people really overestimate is that the 2016-->2020 swings with Asians and Hispanics are bound to repeat and by a similar magnitude. Both of these were clearly swings based of persuasion, and polling has historically struggled with both groups. From 2012-->2016 many of the Hispanic areas that had the biggest leftwards lurches had the biggest swings right in 2020 which makes me think the rightwards swing amongst Hispanics in 2020 may have been in part some sort of reversion to the mean - perhaps Trump's 2016 rhetoric around the border wall made him have an unusually weak performance for a Republican that cycle.

Trump 2020 still did worse than Obama 2012 in a lot of heavily Hispanic Congressional seats one might not expect such as CA-42, IL-04, TX-29, and so on.

It seems like most pundits, even left favorable pundits have some sort of Hispanic/Asian rightwards shift "baked in" to their 2024 forecasts - I'm not saying that can't or won't happen, but I feel like we don't question it enough.

[Kamala's side hoe]

I think one thing people really overestimate is that the 2016-->2020 swings with Asians and Hispanics are bound to repeat and by a similar magnitude. Both of these were clearly swings based of persuasion, and polling has historically struggled with both groups. From 2012-->2016 many of the Hispanic areas that had the biggest leftwards lurches had the biggest swings right in 2020 which makes me think the rightwards swing amongst Hispanics in 2020 may have been in part some sort of reversion to the mean - perhaps Trump's 2016 rhetoric around the border wall made him have an unusually weak performance for a Republican that cycle.

Trump 2020 still did worse than ObamaRomney(?) 2012 in a lot of heavily Hispanic Congressional seats one might not expect such as CA-42, IL-04, TX-29, and so on.

It seems like most pundits, even left favorable pundits have some sort of Hispanic/Asian rightwards shift "baked in" to their 2024 forecasts - I'm not saying that can't or won't happen, but I feel like we don't question it enough.

CDs aren't static over time- it's possible that some of those CDs went through major demographically change between 2012 and 2020.

I see no reason why Latino and Asian voters won't continue swinging and/or trending in the same direction as Black voters (towards the GOP), as they have in every cycle post-Obama. With (non-rural) Latinos and especially Asians I think this is mainly being driven by natural ideological polarization of newer waves of immigrants who grew up abroad- not persuasion of existing voters nor assimilation of native-born voters. (I don't really talk about politics with people IRL, but I can definitely confirm that 1.5 and 2nd+ gen Asians aren't assimilating into voting more R than their parents/elders.)

Anyways I agree that we probably aren't going to get 2016->2020 level R swings with either group, not even if Trump wins the PV and a second term.

[ProgressiveModerate]

I think one thing people really overestimate is that the 2016-->2020 swings with Asians and Hispanics are bound to repeat and by a similar magnitude. Both of these were clearly swings based of persuasion, and polling has historically struggled with both groups. From 2012-->2016 many of the Hispanic areas that had the biggest leftwards lurches had the biggest swings right in 2020 which makes me think the rightwards swing amongst Hispanics in 2020 may have been in part some sort of reversion to the mean - perhaps Trump's 2016 rhetoric around the border wall made him have an unusually weak performance for a Republican that cycle.

Trump 2020 still did worse than ObamaRomney(?) 2012 in a lot of heavily Hispanic Congressional seats one might not expect such as CA-42, IL-04, TX-29, and so on.

It seems like most pundits, even left favorable pundits have some sort of Hispanic/Asian rightwards shift "baked in" to their 2024 forecasts - I'm not saying that can't or won't happen, but I feel like we don't question it enough.

I see no reason why Latino and Asian voters won't continue swinging and/or trending in the same direction as Black voters (towards the GOP), as they have in every cycle post-Obama. With (non-rural) Latinos and especially Asians I think this is mainly being driven by natural ideological polarization of newer waves of immigrants who grew up abroad- not persuasion of existing voters nor assimilation of native-born voters. (I don't really talk about politics with people IRL, but I can definitely confirm that 1.5 and 2nd+ gen Asians aren't assimilating into voting more R than their parents/elders.)

Anyways I agree that we probably aren't going to get 2016->2020 level R swings with either group, not even if Trump wins the PV and a second term.

My point was many of these heavily Hispanic and Asian communities swung left in 2012-->2016 and still on net swung left 2012-->2020. Hispanics as a whole probably swung left 2012-->2016.