Democrats need the working class, and owe them most of their post-2016 wins
By John L. Ray (@johnlray)
Much has been said about the failure of Democrats to earn the working class vote. And yet in the post-2016 electoral environment Democrats have won races in working class states and districts from Alabama to Oklahoma. In PA-18, Democrats narrowly won a deep red district and in OH-12, they can within a point of a seat that had spent decades in safe red territory. How have Democrats fared with voters of different socioeconomic classes since the 2016 election in the Congressional and state legislative elections that have taken place since then?
To assess this question, I compared data on federal and state elections in the post-Trump era to Census demographics. In particular, I compared median household income and the racial composition of the districts to special and state legislative elections post-2016 tracked by DailyKos. I focused on two quantities of interest: change in partisan vote share since 2016 and household income.
Household income was of particular interest because some pundits have suggested Democrats haven’t made any ground among “working class” voters—a term often little more than a euphemism for middle-class white people. The term “working class” is constructed from ahistorical nostalgia that holds working white men as the engine of the economy. For this reason, I wanted to know how Democratic vote share has changed by racial group across different levels of household income.
There have been just over 300 Congressional and state legislative elections since November 2016, including the 100 seats that were up election in Virginia and the 40 up for election in New Jersey on November 7, 2017. While this is not a large sample, we’re able to see trends that could carry over to the November general election. The trends are observational in nature and are drawn from simple bivariate correlations of vote share change on household income across race/ethnicity groups, but provide working insight as to whether or not reality corresponds to narrative.
In the following graphs I plot change in vote share by median household income from 2016 to the most recent election across districts that have had elections since Trump took office. The first plot includes all races that have occurred since Trump took office, while the subsequent plots subset districts to those that are above the sample median in Black population, Latino population, and white population.
First, Democrats overall are outperforming recent history, to the tune of about a 6% improvement over Clinton vote share since the 2016 election and about a 7% improvement over elections in the same district for the same office prior to Trump (noting that, while most states have elections in even-numbered years, some states like Virginia hold elections in odd-numbered years). Second, the scale of that overperformance is higher in lower-income households. Districts with below-average household income with respect to the full sample (below roughly $57,000) have improved over Clinton’s vote share by about 8%, while those with above-average household income have improved over Clinton by about 5%.
That effect holds across race/ethnic groups, but is strongest among above-median white population districts and is not statistically significant in the traditional sense in above-median Black districts. This is the case because Democratic vote share is already very high in above-median Black districts–there’s not much more room to become vote more Democratic. In above-median Black districts Democratic vote share is between 60% and 70%, while in above-median white districts it is below 50%. So not only has Democratic vote share improved dramatically, Democratic vote share improvement is strongly correlated with income in above-median white districts. Without dwelling too much on regression statistics, these results suggest a $10,000 change in household income is associated with about a 0.8% change in predicted vote share improvement for Democrats over Clinton’s 2016 performance in a typical district. That relationship is negative: As districts get wealthier, they become less favorable to Democratic candidates in the post-Trump era.
In the following figures, I show the magnitude of Democratic vote share change from November 2016 to the most recent election. The red line shows the average Democratic vote share among each subset of the overall collection of districts in November 2016, and the blue line shows the average Democratic vote share in elections held in those districts since then.
The conventional wisdom is that Democrats lack appeal with working class voters. That narrative has a certain appeal (and there was some evidence for that narrative in 2016), but it does not seem to hold in the present era. Democratic vote share has improved sharply in white working class districts and the party is performing better with working class voters of color. Though pundits have focused on the wealthy suburban districts, these data suggest Democrats will be competitive across the board, including the working class districts that moved right in 2016.
John L. Ray (@johnlray) is a Senior Political Analyst at YouGov Blue and a Senior Adviser to Data for Progress.