Declines in turnout among low-income people in the US – again

Voter turnout among lower-income people and communities dropped in 2024 relative to 2020, which was already lower than 2016. The gap between voting by the privileged/high income and voting by lower-income people widened (relative to 2020, which was already a bigger gap than 2016).

I’ve now added 2024 data to two different analyses for two papers I wrote (both with Ankit Rastogi) comparing income patterns in voter turnout in 2016 and 2020. Both of those papers indicated the income gap had widened, and turnout among lower-income people (or people in lower-income precincts) had stayed steady or even declined since 2016.

I suspected based on some analyses of Philadelphia data that turnout had declined again in 2024, but couldn’t be sure until I looked at both the CES and some precinct level data from Redistricting Data Hub. Now I’m sure.

Below are a few graphs showing what I found. (They are ugly, Stata-standard – I work in R when I’m aiming for pretty graphs, but for data cleaning and quick analysis Stata is my go-to.)

For both analyses, the X axis is household income percentile, the Y axis is predicted turnout. [Wonky bit: from a regression model that interacts year with every variable, and has controls for state & some demographics.]

First, here’s estimated turnout 2012 – 2016 by income group. You can see the predicted turnout from the lowest income people dropping, at the right side of the graph, in every election. Next if you look at the slopes of each line (ignoring the heights) – they get steeper from 2016 to 2020 to 2024.

For this one (with a great grad student, Lydia Orr), we used Census block-group data, mapped to precinct returns. (This is only for 4 states – AZ, GA, SC, TX.)

You can see basically the same pattern here – 2020 and 2024 are both steeper than 2016 (by a lot); it looks here like there was lower turnout in the poorest places in 2016 than in 2020, and lower in 2024 than 2020 also. In our analysis comparing precincts (rather than block groups) in ALL STATES it looks more like 2020 turnout in the poorest places was the same as or lower than 2016 turnout.

[Wonky bit: here, the X axis is the percentile rank of each block group’s median household income – the ones with the lowest median income are at 0; the Y axis is predicted number of voters in that census block group; there’s a control for the estimated number of potential voters in the precinct in the model, so differences in precinct size aren’t driving the results (I’ve also done it with a direct turnout estimate as the dependent variable.]

All of which to say, “Democracy is Losing the Demos“; there’s a pervasive disconnection between lower-income people and politics/voting/government/political power, and if we’re going to keep/restore/improve our democracy we better figure out how to make democracy and voting work better for poor and working-class people (I’ll have a public report about that coming out later this year.)