California shifted sharply to Trump in the 2024 presidential election. Even though Trump lost the state by a large margin, he did better than any Republican in 20 years. Yet in the same election, hotly contested US House races led to a net gain of three seats for the Democrats, leading to their second-highest seat total ever. How do we reconcile these two shifts? Is the state moving Republican or not?

Part of the discrepancy reflects the difference in timing between presidential and congressional elections. Trump’s vote gains are in comparison to the 2020 presidential election, while the Democratic seat gains came against the congressional elections of 2022. Trump gained about 4.5 percentage points (pp) statewide since 2020, but his party only gained 0.3 pp in congressional races since 2022.

In fact, between 2022 and 2024 Democrats actually gained vote share in 17 of the 43 contested seats (40%). The Democrats who claimed new seats had a slight improvement in one very competitive race (0.3 pp in District 13) and more substantial improvements in two others (4.6 pp in District 27 and 2.5 pp in District 45).

The results look more comparable if we use the 2020 election as the common baseline. Between 2020 and 2024, Democrats lost about as much House vote share in contested races (2.7 pp) as they did presidential vote share in those same contested districts (3.7 pp).

How, then, did Democrats do so well as Republican support surged? They made the most of the close races. They claimed all but one of the seats that had margins of victory under 5 percent. The three seats with the closest outcomes were also the three the Democrats flipped. The winners in these seats might easily have been different.

The presidential and congressional votes also had somewhat different drivers, despite being strongly correlated overall. The shift toward Trump was much larger in places where turnout declined the most: it was almost twice as large for districts in the top half of turnout declines (6.0 pp) as for districts in the bottom half (3.2 pp). And districts with larger Latino populations saw much larger shifts toward Trump as well.

The change in the House votes (using 2020 results reconstituted for the current districts; see the figure notes for details) also correlates with turnout and the size of the local Latino population. Yet the relationships are weaker than in the presidential case. The vote for Congress shifted Republican by about one point more in the most heavily Latino districts, compared to the 3.8 points more for president. And the congressional shift was 1.8 points larger in the districts with the biggest turnout drops, compared to 2.8 points larger in the vote for president.

These results point to a modest separation between congressional and presidential voting in California. Most people vote a straight party ticket, but in 2024 a meaningful number focused instead on candidates—enough to shift the congressional delegation one way as presidential support went the other. In the mid-20th century, votes for Congress were significantly decoupled from votes for president, with important consequences for the organization of political life and policy outcomes. We are a long way from returning to that world. But it will be important to watch the current separation to see if it grows over time.