The Washington Post's Lenny Bronner recently shared a series of analyses of the Iowa Republican and New Hampshire Republican primary contests, estimating the flow of each Republican primary candidate's voters from 2016 to 2024. While this model was fairly accurate in predicting the eventual result of the South Carolina Republican primary, it overestimated Haley's share of the Michigan primary vote by roughly 10%. Having compiled county subdivision-level results from the Michigan primary, I fit the voterflow model for Michigan townships specifically, aiming to identify differences between Michigan and the first three contested primary states using an ecological inference method.¹

Unlike the New Hampshire and South Carolina primaries, which saw roughly equal or greater turnout compared to the 2016 primaries, the Michigan primary saw somewhat depressed turnout. It follows that Haley suffers a greater differential turnout loss compared to Trump. 

The depressed turnout in the Michigan primary may be attributed to several different factors, most prominently, it is evident that Trump is the presumptive Republican nominee, and barring extraordinary circumstance, will stand to face President Biden in November. While this was arguably true for the New Hampshire and South Carolina primaries, Haley was largely still seen as a viable alternative to Trump until her loss in her home state to Trump. Michigan also saw a more active Democratic primary compared to previous contests due to an Uncommitted campaign to voice disapproval of Biden's handling of Gaza; it had been the case that many Democrats voted in the Republican primary to protest Trump in past primaries due to the low salience of those Democratic primaries. 

The Republican Party in 2024 is——in a way completely unpredictable in early 2016——indisputably the party of Trump. While Trump faces less opposition in this primary as a de facto incumbent, his consolidation of the party base can also be explained in part by members of his party less tied to his political coalition leaving the party; I speculate that the greater share of Kasich '16 voters voting for Trump '24 compared to Rubio '16 voters is a result of a greater share of Kasich voters leaving the Republican Party than Rubio voters, rather than greater ideological similarities between Trump and Kasich compared to Rubio.

Below is the complete voterflow from 2016 to 2024.