Public sentiment isn’t much better. Polling released earlier this year shows that a majority of Republicans and 4 in 10 Democrats view members of the other party not as political rivals but, instead, as enemies. That’s not a sign of a partisan thaw.
And yet, unexpectedly, there are small signs that partisanship in the House might have plateaued, if it hasn’t really improved.
One measure we can consider is the average ideology of sitting members of Congress. The team at Voteview uses a proprietary methodology to assign ideological scores to members of the House and Senate, allowing us to track how liberal or conservative individual representatives are. (The short explanation of how it works is that each member is assigned a score between -1, very liberal, and 1, very conservative, that generally measures support for government activism.)
Over the past 50 years or so, the average ideology of the two parties has grown apart. It’s been about two decades since the most conservative Democrat was more conservative than the most liberal Republican (displayed as the white gap between the parties on the graph below).
(That five-Congress spike in the most conservative Democrat reflects former Georgia representative Larry McDonald.)
What should stand out on that graph, though, is that both the ideological ranges each party’s caucus covers (the light-colored bars) and the average ideological value of the caucus haven’t moved a lot in recent Congresses. Neither, therefore, has the difference between the two parties’ average scores. It’s up slightly — but only slightly.
I’ve noticed that pattern in the past, so it immediately came to mind when Cook Political Report unveiled its assessment of its Partisan Voter Index by House district after the 2020 election. The PVI score of a congressional district is a measure of how Republican or Democratic the district voted relative to the country overall in the past two elections. Cook recalculates the scores after presidential elections or after the redrawing of congressional lines (as after a census). It correlates with the Voteview scores of the representatives who represent those districts, as you might expect, with more Republican districts generally being represented by more conservative elected officials.
Comparing the PVI score of a district with the ideology of the district’s representative gives us the plot below. We added in another metric: how much the district’s PVI score shifted since the 2020 iteration of Cook’s analysis.
As you can see in the graphs at the bottom, there’s not a clear pattern of districts moving left or right. About 60 percent of districts that shifted to the right already had Republican-leaning PVI scores in Cook’s 2020 assessment. About 44 percent of the districts that shifted to the left were Democrat-leaning last year.
What’s particularly interesting about Cook’s analysis is the districts that sit in the middle of that graph. There are, by Cook’s count, 78 districts in which the PVI score is within five points — swing districts in its vernacular. It’s the solid-colored circles on the graph below.
Notable here is how many of those districts are now represented by more liberal representatives. This is in part a function of the Democrats winning the House in 2018, a flip that depended on winning more-purple suburban areas.
Over time, Cook’s David Wasserman and Ally Flinn write, the number of swing districts has shrunk. After the 1996 presidential contest, there were 164 swing districts, more than a third of the total. By the 2017 iteration, that figure had declined to 72 districts, about 17 percent of the total.
In the new analysis? Back up to 78 swing districts.
That’s still lower than the number identified in the 2016 iteration (which followed redistricting in several states). But, again, it shows a metric of partisanship that hasn’t changed significantly in the past several years.
What’s also worth pointing out here (as Wasserman and Flinn do) is that this loss of swing districts is not obviously a function of redistricting broadly. There are two post-census redistricting efforts indicated above, the new PVI assessments in 2002 and 2012. There was a slight downtick in the number of Democratic and swing districts after that assessment, but the pattern by 2013 was consistent with the period before the lines were redrawn.
“Of the net 86 ‘swing seats’ that have vanished since 1997,” they write, “81 percent of the decline has resulted from areas trending redder or bluer from election to election, while only 19 percent of the decline has resulted from changes to district boundaries.”
The question, though, is what happens moving forward. Are we about as sorted as we’re going to get? That sorting ties into ideology because more densely partisan districts can afford to elect more ideological primary candidates. (Georgia Rep. Marjorie Taylor Greene’s district is R+28, for example.) Over the past four PVI allocations, the average lean of districts hasn’t changed a lot, with Democratic-leaning districts averaging a 16-point advantage in 2020 and 2021. Republican-leaning districts went from a 12-point to a 13-point average advantage, up slightly from an 11-point edge in the assessment after the 2012 election. Does this contribute to the relatively static ideologies displayed in the House?
Having not been introduced to American politics yesterday, I will not proclaim that partisanship in the House has peaked. Maybe we’re measuring the wrong things; maybe these measures aren’t as important as they might seem. Maybe the upcoming redistricting that will accompany the release of the new census data will reshape things again.
It is nonetheless worth noting that these measures have stalled. It’s possible, if unlikely, that congressional partisanship has peaked. It’s possible (in a very “anything is possible” sense) that the trend could reverse.
You are nonetheless advised not to hold your breath.