538 has new state-level Republican primary polling averages

Trump leads in the early states, but Haley shows signs of strength.

By
Nathaniel Rakich
 and
G. Elliott Morris
October 9, 2023, 10:54 AM ET
6 min read

If all you look at is national polls of the 2024 Republican presidential primary, former U.N. Ambassador Nikki Haley doesn't look all that strong. As of Monday at 10 a.m. Eastern, she has only 7 percent support in 538's national polling average. But parties don't pick their presidential nominees using national primaries, so those national numbers don't give us a complete picture of the race.

For that, you need to look at state polls. That's why 538 is pleased to announce that we're unveiling polling averages for individual Republican state caucuses and primaries, which begin just a few short months from now in snowy Iowa and New Hampshire. And our state-level averages reveal that candidates such as Haley, former New Jersey Gov. Chris Christie and South Carolina Sen. Tim Scott are in better shape in those early states than they are nationally, giving them hope that an early win or two could still propel them to the nomination.

That said, let's not miss the forest for the trees: Former President Donald Trump still has a dominant lead in these early states — almost as big as his lead nationally. So it's still very possible that the 2024 Republican nomination won't be much of a contest.

The first state on the 2024 Republican primary calendar is Iowa, which will hold its famous first-in-the-nation caucuses on Jan. 15. And right now, the Iowa polls look pretty similar to the national polls. According to our average, Trump is in first place with 49 percent support, followed distantly by Florida Gov. Ron DeSantis at 17 percent. That's about 3 points better for DeSantis, and 9 points worse for Trump, than they are doing nationally (15 percent and 58 percent, respectively). Haley (9 percent in Iowa, 7 percent nationally) and Scott (6 percent in Iowa, 3 percent nationally) are also polling better in Iowa than they are overall.

Next on the calendar will likely be New Hampshire; although the Granite State hasn't yet officially announced its primary date, it has held the first primary in the nation for over 100 years. And according to our New Hampshire primary average, Trump is in great shape here too, with 45 percent support. But Haley is at a healthy 13 percent, ahead of DeSantis with 10 percent. And Christie, who is barely polling at 3 percent nationally, is at 9 percent in New Hampshire — perhaps because New Hampshire allows independents to vote in its primaries, and even Republican-leaning independents tend to be more skeptical of Trump than actual registered Republicans. (Christie has centered his campaign on his criticism of Trump.)

The next two Republican contests are scheduled for Feb. 8 in Nevada and the U.S. Virgin Islands. But we don't have enough polls of either of those contests to generate a polling average for them. (538 publishes a polling average for a race only when it has been polled at least eight times by at least three different pollsters. In the case of the 2024 Republican primary contests, we're also requiring that those polls be conducted after Jan. 1, 2023.) That takes us to the final early state: South Carolina, which votes on Feb. 24.

In our South Carolina average, Trump is at 46 percent support, Haley is at 18 percent, DeSantis is at 14 percent and Scott is at 7 percent. Of course, Haley and Scott are at an advantage here: Haley was the two-term governor of the Palmetto State, and Scott has been its junior senator since 2013. But the fact that Haley, in particular, is in second place in at least two of the first five primary states (or territories) suggests that she, not DeSantis, may be the last candidate standing against Trump next spring.

We've also rolled out primary polling averages for two other states that vote in March: California (March 5) and Florida (March 19). We'll add more states to this list on a rolling basis once they reach our eight polls/three pollsters threshold. In addition, we debuted two new approval trackers — one for the Supreme Court and one for Congress — as well as averages of Republican presidential candidates' favorability ratings specifically among Republicans. (For example, here is Trump's.)

Finally, when building these new polling averages, we made some technical tweaks to our methodology, which also affects our preexisting averages. Specifically, we made four big improvements, along with several code optimizations:

  • The first large change is an added aggregation-and-adjustment step for automatically detecting outlier polls and decreasing the weight our model puts on them.
  • Second, we added a step when calculating our polynomial trendline — one of the ingredients that goes into each average — that uses iteration to provide more stable averages when polls have higher variance.
  • Third, we now combine our moving average and polynomial trends into one super-trend based in part on how each component indicator would have predicted real-life poll results over the last five days.
  • And finally, we made minor adjustments to the way we optimize the settings of our model, ensuring the optimal balance between a fast-moving trend that predicts poll results well, but can jump around a lot, and a smooth, slower-moving trend that doesn't overreact to new data but might miss sudden shifts in public opinion.

    • For more details, check out our full, updated methodology page.