Gerd Nonneman

Which voters did the Democrats lose in 2024, and which did they regain in 2026? I charted the change in the Democratic margin by income, education, age, and race and gender: first from 2020 to 2024, then from 2024 to 2026. The headline: Democrats campaigned on economic populism in 2026, and regained the low-income, non-white, and young voters they lost to Trump in 2024. The 2026 midterm results are strong correlational evidence that an economic populist campaign can regain voters from illiberal populists. They are also inconsistent with much of the cultural backlash thesis of Inglehart and Norris. Trump’s 2024 gains were among young, lower-income, and Latino and Black voters, and those same groups voted Democratic again two years later, once the party campaigned on the cost of living and corporate power. When identifiable social groups change their vote twice within six years, in a pattern that follows income and age, they are responding to their material conditions and lived experience. Cultural values simply do not change that fast. Income: in 2024 the Democrats lost 10 points among households under $50k and 6 points in the $50–100k bracket, while gaining 3 above $100k. Trump was the first Republican presidential candidate since Reagan in 1984 to receive more votes than the Democrat among working-class voters (households under $50k). This is what led many to say, somewhat prematurely but not without reason, that Trump had built a multiracial working-class coalition. That coalition has now fallen apart. In 2026 the Democrats gained 18 and 17 points in the two lower brackets. The 2026 gain was larger than the 2024 loss, and it occurred in the same brackets. The party gained a further 5 points above $100k, so economic populism did not reduce its support among affluent voters; i.e., no need to fear that economic populism would scare away Chuck Schumer’s mythical affluent suburban centrist voter. That fear is a myth. Chuck, you can be bold. Age: the Democratic margin among voters aged 18–24 fell by 20 points in 2024, the largest decline of any age group, and rose by 42 points in 2026. The 2026 gains are smaller in each older age group, with 5 points among those over 65. Race and gender: the margin among white men and white women was almost unchanged in 2024, and by some measures Trump even lost white voters that year. Trump’s gains were concentrated among Latino men (−34) and Black men (−6), and these two groups recorded the largest Democratic gains in the race and gender panel in 2026 (+27 and +23). According to cultural backlash theory, the far right increasingly attracts older, white, less educated voters who reject the progressive values of younger generations. Only the 2024 education data are consistent with that expectation: the Democrats lost 20 points among voters who never attended college. In 2026, however, the Democrats’ gains were nearly uniform across education levels (+16 to +19 in four of the five groups). In sum: the 2026 result was the work of the working class (low income), the young, and Latino and Black men, the voters Trump had won in large numbers in 2024, who voted Democratic again. The professional class contributed least to the 2026 landslide: the Democrats gained 5 points among households above $100k against 17–18 points below that line, and college graduates moved about as much as voters without a degree. Two footnotes. First, the data in the chart are changes in margins; the level of support in each group is not included. In cross-section, large differences by education, race and age remain: even in 2026, Republicans led among white voters without a college degree by 21 points, and the Democratic margin was far larger among the young than among the old. Elections, however, are won and lost through changes. Second, exit-poll data are correlational, and this time Trump was the incumbent responsible for inflation, so a significant part of the change is a vote against the party in power. At the same time, the realignment shows clear patterns, so it is not a mere midterm correction. Sources and method are in the figure note.

Which voters did the Democrats lose in 2024, and which did they regain in 2026? I charted the change in the Democratic margin by income, education, age, and race and gender: first from 2020 to 2024, then from 2024 to 2026. The headline: Democrats campaigned on economic populism in 2026, and regained the low-income, non-white, and young voters they lost to Trump in 2024. The 2026 midterm results are strong correlational evidence that an economic populist campaign can regain voters from illiberal populists. They are also inconsistent with much of the cultural backlash thesis of Inglehart and Norris. Trump’s 2024 gains were among young, lower-income, and Latino and Black voters, and those same groups voted Democratic again two years later, once the party campaigned on the cost of living and corporate power. When identifiable social groups change their vote twice within six years, in a pattern that follows income and age, they are responding to their material conditions and lived experience. Cultural values simply do not change that fast. Income: in 2024 the Democrats lost 10 points among households under $50k and 6 points in the $50–100k bracket, while gaining 3 above $100k. Trump was the first Republican presidential candidate since Reagan in 1984 to receive more votes than the Democrat among working-class voters (households under $50k). This is what led many to say, somewhat prematurely but not without reason, that Trump had built a multiracial working-class coalition. That coalition has now fallen apart. In 2026 the Democrats gained 18 and 17 points in the two lower brackets. The 2026 gain was larger than the 2024 loss, and it occurred in the same brackets. The party gained a further 5 points above $100k, so economic populism did not reduce its support among affluent voters; i.e., no need to fear that economic populism would scare away Chuck Schumer’s mythical affluent suburban centrist voter. That fear is a myth. Chuck, you can be bold. Age: the Democratic margin among voters aged 18–24 fell by 20 points in 2024, the largest decline of any age group, and rose by 42 points in 2026. The 2026 gains are smaller in each older age group, with 5 points among those over 65. Race and gender: the margin among white men and white women was almost unchanged in 2024, and by some measures Trump even lost white voters that year. Trump’s gains were concentrated among Latino men (−34) and Black men (−6), and these two groups recorded the largest Democratic gains in the race and gender panel in 2026 (+27 and +23). According to cultural backlash theory, the far right increasingly attracts older, white, less educated voters who reject the progressive values of younger generations. Only the 2024 education data are consistent with that expectation: the Democrats lost 20 points among voters who never attended college. In 2026, however, the Democrats’ gains were nearly uniform across education levels (+16 to +19 in four of the five groups). In sum: the 2026 result was the work of the working class (low income), the young, and Latino and Black men, the voters Trump had won in large numbers in 2024, who voted Democratic again. The professional class contributed least to the 2026 landslide: the Democrats gained 5 points among households above $100k against 17–18 points below that line, and college graduates moved about as much as voters without a degree. Two footnotes. First, the data in the chart are changes in margins; the level of support in each group is not included. In cross-section, large differences by education, race and age remain: even in 2026, Republicans led among white voters without a college degree by 21 points, and the Democratic margin was far larger among the young than among the old. Elections, however, are won and lost through changes. Second, exit-poll data are correlational, and this time Trump was the incumbent responsible for inflation, so a significant part of the change is a vote against the party in power. At the same time, the realignment shows clear patterns, so it is not a mere midterm correction. Sources and method are in the figure note.

En nu? 🤫