THE UNSTEADY CROWN OF A KING-MAKER:
What base rates can tell us about the power of Trump’s endorsement
Former President Trump lost his bid for a second term, but he is still freely wielding his political influence over this year’s midterm elections. As he told the Washington Post in an April interview, “I’m the king of endorsements.”
Indeed, since mid-May, Trump has endorsed sixty-six candidates, and fifty-seven of them won their primaries.
You could say a record of 57-9 suggests Trump’s endorsements must be enormously influential. And that is certainly what Trump himself is touting. On May 31, Trump wrote on his social media site, “We (all of us) had an incredible cycle of endorsement victories, a cycle like never before.” (Presumably, his 14-1 record on June 7 would serve to bolster such claims.)
But to put Trump’s influence into any kind of perspective, understanding the absolute number of Trump-endorsed candidates who won their races doesn’t help us much.
Fifty-seven winners is certainly a lot! But we need to go further and ask how many of those candidates would have won their races without Trump’s endorsement. Maybe that number is also fifty-seven. Or thirty. Or all sixty-six. Without this context, his win-loss record is impossible to interpret.
That means we need to look at the base rates.
A base rate, simply put, is how often something happens in a situation similar to the one you’re considering. Base rates provide a starting point, an initial best guess, for understanding what to expect and are crucial to great decision making and interpreting data.
As a simple example, if you were asked to make a forecast of how many days it will rain in September in your hometown, you can look up how many days on average it has historically rained in September and use that as your forecast for this coming September.
Knowing your base rates makes your forecasts and your decisions more accurate.
Base rates can help us to better understand the value of Trump’s support. If we can figure out how many candidates would have won their races without Trump’s endorsement, then we can see if Trump’s endorsement increases (or decreases) a candidate’s chances as compared to the base rate.
It turns out that the most predictive factor in a contested primary is whether one of the candidates is an incumbent.
Incumbents almost never lose primary races. According to a 2022 Stanford study citing data from an upcoming book by political scientist Steve Rogers, “98% of incumbents win their primaries.” This is similar to statistics from the Brookings Institution’s 2018 Primaries Project. In the 2014, 2016, and 2018 election cycles, between 42% and 50% of House incumbent Republicans faced primary challenges. Out of approximately 280 contested primaries in that period, just 7 of those incumbents lost.
This gives us a base rate of a 98% probability that an incumbent wins in a contested primary.
And that base rate tells you that a great strategy for picking winners in a contested primary is to just go ahead and endorse the incumbent. You will win almost 100% of the time!
It turns out Trump has pretty much followed that strategy, mostly endorsing incumbents.
In the primaries held from May 17 thru June 7, Trump endorsed forty-seven incumbents. Nineteen of those incumbents ran unopposed, obviously the same number who would have won in any unopposed election, so we can eliminate those races from consideration.
The only Trump-backed incumbent in the group who lost was Madison Cawthorn. That means Trump’s record in contested primaries in those fourteen states is 27-1.
Trump’s 21-1 record meant that he backed the winner 95% of the time, hewing pretty closely to the base rate.
Given that almost all incumbents win, including those backed by Trump, the fact that he went 21-1 isn’t that informative as it is hard to improve on a base rate of 98%.
The base rate for incumbents winning also tells you that a bad strategy for picking winners would be to endorse the challenger in a contested primary because you will almost always lose.
But maybe if you are a hugely influential political figure, you could buck the base rate with your endorsement. That would really tell us if someone is a king-maker.
So, let’s take a look at the non-incumbents Trump backed. While backing an incumbent doesn’t give much signal, maybe Trump’s endorsement can help a candidate overcome the long odds of challenging someone who is already in office.
In the most recent primaries, Trump went 0-5 in endorsing challengers to incumbents. This was conspicuously on display in the Georgia primary, where Trump endorsed opponents to Governor Brian Kemp (who refused to overturn that state’s 2020 election results), Secretary of State Brad Raffensperger (who certified the results), Attorney General Chris Carr, and Insurance Commissioner John King. (King was opposed by Patrick Witt, who was part of Trump’s post-election legal team in Georgia.)
Challengers face long odds and it appears Trump’s endorsement doesn’t do much to help a candidate overcome them.
Trump’s endorsees have fared a lot better in races for open seats, where there is no incumbent. In the most recent primaries, twelve of those fourteen candidates (including Sarah Huckabee Sanders, Hershel Walker, and Dr. Oz, who narrowly beat out Dave McCormick) won, with a pair of his Georgia House hopefuls coming in second but still advancing to primary runoffs.
All this doesn’t mean that Trump’s endorsement is for nothing, especially if the candidate is vying for an open seat. But comparing his win rate to the base rate does suggest that his endorsement is not particularly helpful beyond incumbency itself.
The concept of base rates, of course, isn’t just something that’s relevant to the public discussion of Trump’s endorsement power. These are questions we should be asking in evaluating any kind of data.
If you’re trying to evaluate the meaning of any statistic, one thing you should always be asking is, “Compared to what?”
That’s where base rates come in, helping put raw numbers into perspective.
 I’m excluding California. California holds an open primary, with the top two candidates advancing to November. All six of his incumbent endorsees advanced but four were unopposed by another Republican and the other two were essentially non-competitive, with those other Republicans receiving just 6% and 11% of the votes.