Alabama and the limits of Republican factionalism
Tonight, I'll be paying close attention to the results coming in from Alabama's GOP presidential primary.
That's not because there is much doubt about the outcome. Against stiffer competition in 2016, Trump carried every county in this Republican-dominated state, winning by 22 percentage points statewide. In 2024 primaries so far, Trump has tended to do well not only in places where he did well eight years ago, but also in places where Ted Cruz, the leading candidate among more conservative and evangelical voters, did well. Since Cruz finished second in Alabama in 2016, it's not a very promising state for Haley.
But even without much question about who's going to win statewide, Alabama serves as a bellwether for the state of today's Republican Party. It's the state with the highest population share of evangelical Protestants nationwide, a focal constituency of the contemporary GOP. Alabama has also been home to several competitive GOP primaries in recent years that can jointly tell us a lot about competition in the party today. Even after Trump's rise to the top of the party in 2016, precinct-level returns in Alabama's Senate primaries make clear that there aren't consistent pro-Trump and anti-Trump blocs going head-to-head in election after election. Instead, GOP candidates in recent Alabama primaries have put together somewhat idiosyncratic geographic coalitions.
Consider 2017, when appointed U.S. Senator Luther Strange, who had the backing of both Trump and the GOP establishment, competed against judge Roy Moore in a special election for the GOP's Senate nomination. Despite accusations of child sexual abuse, Moore prevailed over Strange in the primary (before losing to Democrat Doug Jones). New research suggests that in that primary, precinct-level support for Moore was higher in places where Trump and Ben Carson had done better in 2016, but the correlations are pretty modestly sized. In other words, knowing where Trump did better in 2016 really didn't have much predictive power in the next year. Still, Marco Rubio's 2016 vote share was negatively associated with Moore's, meaning those two candidates drew support from different places.
2020 may provide a clearer test: Trump's first supporter in the U.S. Senate and subsequent attorney general, Jeff Sessions, was running to take back his old seat, but Trump and Sessions had since fallen out and Trump instead endorsed the eventual winner, Tommy Tuberville. The places that had backed Trump in 2016 were somewhat less supportive of Sessions in 2020. Still, these correlations were pretty modest, and Trump's 2016 support also wasn't strongly correlated with backing either of Sessions's major opponents, Tuberville or Bradley Byrne.
And in the 2022 Republican Senate primary, Trump support was slightly negatively associated with 2022 support for both Rep. Mo Brooks, the conservative whom Trump unendorsed during the campaign, and eventual winner Katie Britt, who he did endorse. Trump's 2016 vote was only somewhat positively correlated with support for Mike Durant, a helicopter pilot who was shot down in Somalia as part of the battle depicted in "Black Hawk Down." But like 2017 and 2020, the 2022 Republican primary was not just a rerun of 2016.
Now, with Trump himself on the ballot, it's a different story. I expect that Trump will do very well in the parts of Alabama that he dominated in 2016 (read: most of the state), and that Haley's vote margins will be slightly stronger around Birmingham and Huntsville, just as Rubio did eight years ago. This was true in New Hampshire, for example, where Trump's 2016 precinct-level vote share was correlated with his 2024 vote share at 0.63, a strong correlation given the very different competition he faced in the two years. But, as a close look at Alabama primaries in recent years shows, Trump's 2016 supporters haven't necessarily formed a cohesive voting faction. During the Trump era, when Trump is not on the ballot himself, voters haven't always hewed closely to the divisions he has fostered.
—Dan Hopkins, 538 contributor