Over the last 30 years, the American electorate has undergone a major realignment, driven primarily by polarization along educational lines. Degree-holding suburban voters, previously a solidly Republican group, have drifted to the left and towards the Democratic party, while white non-college voters have responded in kind by shifting strongly to the right and swinging Republican at increasingly high levels. Unsurprisingly, given today’s levels of partisan polarization, neither of these patterns show signs of abating anytime soon.
This was most clearly evident in 2016 and 2020, where the voting lines diverged more starkly along education than they ever had before, particularly among white voters. As per the Associated Press’ 2020 Votecast, although whites comprised 74% of the electorate and backed Donald Trump by 12 points, the picture under the hood is of two very different different demographics. Whites with a degree (31% of the electorate) voted for Joe Biden by 7, while Trump won non-college whites (43% of the electorate) by 25 points. The national 32-point divergence on educational lines is the highest it has ever been.
To investigate this statistic and its implications, we’ll first use AP Votecast numbers to map state-by-state approximations below of how white voters voted in 2020, and specifically of how the white vote differed by degree attainment. It should be noted that like all survey-based estimates, the exact margins will have some noise and some degree of error; however, they are, on the whole, fairly solid and allow one to draw directionally sound conclusions.
The maps above show some clear and strong regional differences that are worth noting. For example, whites in New England and along the Pacific coast back Democrats by fairly comfortable margins, but these numbers pale in comparison to the rates of support that Republicans rack up with whites in areas like the Deep South (e.g. Louisiana, Alabama, etc), where voting diverges sharply along racial lines. Meanwhile, in the far more racially homogeneous Midwest, Democrats keep the white vote much closer than one might expect, which in turn leads to consistently tight elections in states like Michigan and Wisconsin.
However, regardless of the state-level partisan baselines, one thing is abundantly clear: no state has been immune to educational polarization taking root at some level among white voters. In every single state, college-educated whites are more liberal and more Democratic than their non-college counterparts. With this divide only forecast to increase as time goes on, it is worth considering a variant of an interesting question first asked by FiveThirtyEight in 2016: how much room does either party have to grow in each state?
In this article, we’ll limit our analysis of this question to the white vote on the lines of educational polarization. As this phenomenon continues, the divergence among educational lines is likely to grow even more sharply. This might lead to several Biden-supporting non-college whites switching to the Republican party over time, while many Trump-supporting college whites may swing Democratic in response. Thus, we’ll try to simplify the question above to the following, on a state-by-state basis: “For every white, college-educated voter who voted for Trump, how many Biden-voting non-college whites are there?“
The map below shows our attempt at using the AP Votecast data to answer the question posed. Broadly speaking, redder areas indicate room for growth for the Republican party, bluer areas indicate areas with more Democratic potential, and purple areas indicate a virtual tie in terms of future growth.
Before proceeding, it is worth pointing out that educational attainment, while highly informative, is by no means definitive. As David Shor and Ezra Klein have both said, the metric really serves as a proxy for class polarization, but while it is increasingly predictive, it is not the sole determining factor in a person’s vote. College-educated whites in states like Alabama and Mississippi are staunchly Republican due to a combination of racial polarization and high rates of religiosity. In such states, these factors will very likely dampen the leftward movement they may take over the next several years. On the flip side, in states like Washington and Oregon, high rates of urbanization and social liberalism mean that non-college whites are extremely Democratic, and it is reasonable to expect that a mix of cultural polarization and the widening urban-rural divide might also somewhat insulate many of them from the significant rightward swings seen elsewhere. With that said, this metric has only increased in predictive power lately, especially among whites, and so it is increasingly useful in helping us forecast how the political landscape may evolve.
The brightest spots for the Democratic Party are likely to be found in Texas and Georgia, both of which have a very high concentration of suburban college-educated whites. While the demographic is still staunchly Republican in both states, it has also seen sharp shifts left over the last few cycles. Meanwhile, the heavily-white rural areas in these states already being deep-red means that there are not nearly as many Biden-voting, non-college white voters left for the GOP to flip. Democrats thus have a fairly high amount of upside among white voters in these two states, and if educational polarization continues to increase, they should continue to make sizable gains through flipping a substantial amount of college-educated whites who voted for Donald Trump. This would greatly boost their chances of turning the two rapidly-growing states reliably blue.
This, however, is countered by the benefits Republicans would be poised to reap in states like Wisconsin and Nevada, both of which were decided by exceptionally thin margins in 2020. Wisconsin is a regional anomaly in that it provides very high levels of rural Democratic support compared to how surrounding areas that are demographically similar vote, and as the forces discussed in this article continue, it is more likely than not that they begin voting more similarly to how their neighbors do. In Nevada, meanwhile, extremely low rates of education (just 32% of white voters have a degree), high rates of religiosity, and higher rates of social conservatism make it a prime play for Republicans moving forward. In both these states, the upside for Republicans among non-college whites looks to be substantially greater than the upside for Democrats among college whites is, and this could see both states shift significantly to the right as a result.
There is another state in which Republicans may look to play offense as well. Maine, though comfortably Democratic, has a combination of high Democratic support among non-college whites and rather low educational attainment statewide. Concerningly for Democrats, there are many low-college clusters here that have experienced strong rightward swings over the last decade, unlike New Hampshire and Vermont (both of which are significantly more educated and with far fewer low-college clusters). This makes Maine another state in which Republicans will hope to gain ground in, and Democrats will have to hope that its heavy secularization serves as a counter; else, it is easy to see it becoming much closer in the near future, to the point where a narrowly Republican national environment could see it flip.
Stepping beyond the obvious targets on the map, the picture becomes a bit more worrying for Democrats with regards to the Senate. This is because every state elects two Senators regardless of its size, and there are several closely-decided, heavily white states in which Democrats have a bit more room to fall than Republicans do, such as Michigan and Pennsylvania. Losing ground in those areas may make winning the Senate a good deal harder for them, and while one can credibly argue that Pennsylvania’s urbanization may serve as a valuable counterweight to polarization among non-college whites, it is not clear that the same case can be made for Michigan. Ensuring that the party doesn’t crater in its working-class towns will likely be critical to its long-term viability in both the state and the United States Senate as a whole.
Lastly, there are a lot of states in which Democrats and Republicans both have a roughly equal pool of voters to gain with, and perhaps the most interesting of these is North Carolina, where several suburbs continue to keep a statewide victory just out of reach for Democrats federally by delivering resoundingly large raw vote victories for Republicans. Democrats still maintain higher-than-expected levels of support among working-class whites in the west of the state, however, and one of the keys to predicting the state’s outcome in 2024 will very likely lie in figuring out which party makes the most inroads with the demographics swinging towards them.
It is unreasonable for either party to expect that they can asymmetrically withstand the forces that have driven the last thirty years of partisan and ideological realignment in the United States. The world in which Texas and Georgia both begin voting regularly for Democrats is more likely than not to be the same one in which Wisconsin, Nevada, and Michigan begin consistently voting Republican. This might be a trade that excites Democrats at first, because winning the presidency becomes extremely difficult for the GOP if they lose Texas and Georgia in the same election. But it would also make winning the Senate exceptionally tough for Democrats, given that it would only worsen the chamber’s existing R+5 bias. And in a world in which bipartisanship appears to be dying a very public death, you wonder whether either party would want to make that trade.
I’m a software engineer and a computer scientist (UC Berkeley class of 2019 BA, class of 2020 MS) who has an interest in machine learning, politics, and electoral data. I’m a partner at Split Ticket, handle our Senate races, and make many kinds of electoral models.