We are happy to introduce our demographic vote preference tool. This tool allows you to compare how a given demographic reported they were going to vote in 2020, versus in 2024. You can check off various permutations of gender, education, and income.
2020 survey data wave 10: 2020-09-04 to 2020-09-30, N=23050; wave 11: 2020-10-02 to 2020-10-23, N=19570
2020 survey data wave 33: 2024-08-30 to 2024-10-08, N=25518; wave 33.5: 2024-08-19 to 2024-10-26, N=26298
What is most striking to us is that education polarization between the parties continues, but in particular, it has spread to non-white demographics where the relationship had been inconsistent in 2020. Thus, while for white voters support for Harris among those with college or more education has increased by about 5 points, and among those with high school or less decreased by less than 2 point, the gap between the college or more versus high school or less has increased by much more for African American, Hispanic, and Asian American respondents. Among African Americans, those with a high school or less education, support for Harris relative to Biden has dropped by 10 points (versus a 2 point increase among those with college or more education); a 15 point drop among Hispanics with high school or less (versus just a 2 point drop among those with college or more); a drop of 12 points among Asian Americans with high school or less (versus just a 4 point drop among those with college or more). (Note that our survey includes fewer Asian American respondents, and thus these numbers should be interpreted with extra caution.)
Interestingly, the net effect of educational polarization has been a decrease in racial polarization because white respondents have higher levels of education, on average, than non white. Harris has increased support from white respondents, relative to Biden (by about a point), while losing support from non-white respondents (dropping 6 points among African Americans; 9 points among Hispanics; and 6 points among Asian Americans). This reflects increased uniformity in the education gaps across racial groups.
The combination of education polarization and an increased gender gap across all racial/ethnic groups has resulted in much increased vote shares for Trump among men of color. Thus, for example, we estimate that Trump’s vote share among African American men will be around 27% (20% in 2020); among Hispanic men 47% (35%); and among Asian American men, 41% (32%).
A final interesting thing to note is the shift in support across age groups. The 30 and unders were overwhelmingly supportive of Biden in 2020, by nearly 2:1 (66:34); now they are more evenly split (55 to 45). This is part of a very substantial rightward shift in younger American voters that we’ve noted before. The shift among younger people occurs for women and men, but is much more pronounced among the latter (a 14 point drop in support for the Democrat among younger men versus a 7 point drop among younger women). The shift among young people is counterbalanced by smaller increases among people older than 30 of about 1.4%. This shift among older people, in turn, is exclusively among those with college or more education (a 5% increase, versus a 1.7% decrease among those individuals with high school or less).
An important note on methods: We include the option to present results for 2024 with and without weights for 2020 votes for Biden/Trump. We believe that weighting by prior vote will yield a more accurate prediction for 2024, and it reduces our overall estimates for Harris’s margin over Trump by about 4 points, to 5 points. Whether this is good news for Harris depends on whether we undershoot Trump’s performance again. A comparison between Biden’s performance in 2020 and Harris’ performance in 2024, not weighting by prior vote, suggests that she would get one percentage point less than Biden did (which would have likely been enough to swing the 2020 race). And, our best prediction, when weighting by prior vote, is that the margin for Harris will be the same as it was in the actual vote for Biden in 2020- about 4 points. (But: really, as we discuss in our prior post, our survey and methods aren’t designed to produce the best estimates of the election outcome.)
We recommend, when using this tool, for 2024 estimates, to not weight by 2020 vote, to allow a comparison using the same weighting methodology. (We haven’t yet produced estimates for 2020 using the 2016 vote, so you currently can’t make that comparison.) This allows an apples to apples comparison, and should capture the direction of the shift of the demographics of the coalitions of support in 2020 and 2024. Our discussion above is based on this comparison.
