Independent presidential candidate Robert F. Kennedy Jr. announced Friday that he had suspended his campaign for the presidency and endorsed former President Donald Trump (though he said he will remain on the ballot in states that are not competitive at the presidential level). Kennedy’s bid failed to garner enough support to contend in any state, and support for him in national polls fell by nearly half after President Joe Biden dropped out of the running to be the Democratic Party’s nominee.

Our analysis of the polling data suggests Kennedy’s endorsement of Trump will have a minimal impact on the race. Kennedy, who has consistently polled around 5 percent since Vice President Kamala Harris became the presumptive nominee, was drawing roughly equally from both Trump and Harris, with that support coming from both traditionally Democratic and traditionally Republican groups. His endorsement of Trump may marginally help the Republican among white, male, and older voters. But the effect of his departure on overall support for either candidate will be small.

Kennedy’s candidacy has a small effect on our polling averages

So where will Kennedy’s voters go? There are a few ways to calculate a guess. The most straightforward method is to calculate an average of support for Harris and Trump using only the polls that allow voters to express support for Kennedy, and then compare those results to their average support in polls that do not include Kennedy as a response option. This allows us to guess where his voters might be going if they’re forced to pick other options.

538’s national presidential polling average is doing similar math when combining polls with different response options, so we can extract an estimated effect from the aggregation model as a first guess. Our model tells us that support for Harris is 1.3 percentage points higher in polls testing support only for Harris and Trump than in polls where Kennedy is included. In comparison, Trump’s support is 1.5 percentage points higher in two-way polls than when voters get the third-party option. Adding it up, this suggests that Kennedy’s presence in the race is drawing slightly more support away from Trump and increasing Harris’s margin by about 0.2 percentage points on average across the states.

To put that in context, the daily absolute change in 538’s margin for Harris in our national polling average is, on average, 0.21 percentage points. In other words, the direct effect of Kennedy dropping out, according to polling toplines alone, is about the same as the amount that polls tend to fluctuate based on the average day’s news or campaigning.

With Kennedy gone, heavily partisan groups will repolarize

Another way to guess who Kennedy is pulling more votes from is to look at voting patterns at the demographic level and compare results poll-by-poll instead of inferring aggregate trends (which is what our polling average does).

To do this, we collected all of the Harris-Trump polls that have been conducted since July 21 and included both two-way and multiple-candidate questions on presidential support. Then, we aggregated support for each candidate at the demographic group level using the published crosstabs for these polls, where available, and looked at how the margin between Harris and Trump differed within each of these groups when Kennedy was or was not included in the question.* (To be clear, this is not the same as our topline polling average: No fancy modeling has been applied to this data, just a simple average of all the polls.)

When Kennedy is absent from surveys, support for Harris rises most among Asian, Black and Hispanic voters,** as well as with nonwhite voters without college degrees and young voters. Trump, meanwhile, consolidates support with men, white and rural voters, and voters over 30. That means Trump is best situated to capitalize on Kennedy’s endorsement among the groups already friendliest to both of them.

For the most part, however, Kennedy is drawing few votes from either candidate and, once again, the effect of his withdrawal on the race is likely to be small. Harris’s margin over Trump changes by less than two percentage points in 16 out of the 18 groups we considered, and by less than one percentage point in more than a third.

In fact, the most notable trend in our demographic-level analysis is a repolarization of solid Democratic and Republican groups; that is, without Kennedy on the ballot, Harris tends to gain among groups that are typically solidly Democratic, while Trump does better with key Republican demographics. But vote intention among the more swingy groups — suburbanites, college-educated nonwhites, and women — changes little.

Exit this way

While running for president, Kennedy was often discussed in terms of whether he would be a spoiler candidate for either Trump or Biden/Harris. This was a distinct possibility: As of Aug. 23 at 6:30 p.m. Eastern, our election forecast for Trump and Harris gives Kennedy a 0-in-100 chance of winning the majority of Electoral College votes, but a 40-in-100 chance of third-party candidates winning more votes in any decisive state than the margin between Harris and Trump in that state. With Kennedy’s exit likely to lower those odds, that’s a big reason Trump is likely glad to have Kennedy’s support, even if he doesn’t bring many votes to the table.

When Ross Perot bowed out of the 1992 presidential race (before later reentering it), he said he did not want to send the election to the House of Representatives by siphoning off enough electoral votes to prevent any candidate from reaching an Electoral College majority. And while Kennedy also specified in his announcement Friday that he is withdrawing his name in battleground states where he might prove to be a spoiler, he struck a different (and, to be clear, not very realistic) tune by technically remaining in the race. “I encourage you to vote for me,” he said, specifying that his name will appear on the ballot in most states. “And if enough of you do vote for me, and neither of the major-party candidates win 270 [electoral] votes, which is quite possible … I could conceivably still end up in the White House in a contingent election.”

In order to be eligible to be elected president by the House, however, Kennedy would need to win at least one Electoral College vote from a state’s slate of presidential electors. That is likely an overly optimistic forecast (following an overly optimistic campaign). Toward the end of his press conference, Kennedy officially threw his support behind Trump. That may reflect a more prescient prediction of where his future in American politics lies.

Footnotes

*When pollsters released results among different populations, we preferred polls of likely voters over those of registered voters and registered voters over all adults. When a pollster released multiple different multi-candidate matchups, we included the matchup with the largest candidate field when calculating the margin including Kennedy. We included all national polls conducted after July 21 and released by Aug. 22 at 11:59pm.

**Some pollsters include a “Hispanic” demographic group in their crosstabs, and others include a “Latino” group. We included both categories in the Hispanic/Latino averages.