Since the election, we’ve made it our mission here at FiveThirtyEight to uncover the variables that best explain why things shook out the way they did. We’ve already shown you the strong correlation between partisanship and U.S. House results. We have also pointed out how Republicans won a huge share of rural seats, while Democrats won every single urban seat and many, many suburban seats. Today, we’re taking a look at how well educational attainment predicted the 2018 House results, especially among non-Hispanic white voters.
In 2016, educational divides emerged as one of the top explanations of voters’ choices: White voters without a bachelor’s degree made up the Republican base, while a coalition of nonwhite voters and white college graduates formed the Democratic base. The 2018 midterms seemed to continue what we saw in 2016: Districts with bigger black populations, Hispanic populations or college-educated non-Hispanic white populations tended to vote more Democratic, while non-college-educated white voters remained strongly loyal to the GOP. We found a clear negative relationship (R = -0.72) between the Democratic margin of victory in a district and the share of the district’s population age 25 or older who are non-Hispanic white and lack a bachelor’s degree — a group that pundits often call the “white working class.”1
As we noted, this trend was already evident in 2016. Indeed, a district’s share of non-Hispanic whites without a bachelor’s degree was slightly more predictive of how it voted in the 2016 presidential election than in the 2018 U.S. House election: The correlation coefficient between Hillary Clinton’s vote margin and the percentage of the district that was white people without a bachelor’s degree was -0.79 in 2016, compared with -0.72 in 2018.
However, the relationship was less pronounced in the 2012 presidential election — the correlation was weaker (R = -0.64) and the partisan gulf was smaller between districts that had a high percentage of white voters without a degree and districts where that percentage was lower.2 Put another way, the new demographic coalitions in the Democratic and Republican parties had not yet fully coalesced.
In terms of white voters’ educational attainment predicting election outcomes, 2018 represented a middle ground between 2016 and 2012. Voters neither snapped back to their pre-2016 preferences nor remained quite as divided by education as they were when Trump himself was on the ballot. It’s a confirmation of other analysts’ findings that the 2018 election results represented a blend of 2016 and 2012 partisan baselines. And it answers a question that we posed before: Would Democrats be able to build on their 2016 gains in diverse, upper-class suburbs to make states like Texas competitive? Based on the 2018 election results, it looks as if the answer might be yes. Education as a major factor in vote choice is here to stay.