Two polls taken of North Carolina voters right after the May 17 primary showed Republican Ted Budd leading Democrat Cheri Beasley in the pivotal 2022 race for U.S. Senate. They also show it’s too soon to count Beasley out.

East Carolina University’s latest survey, concluded three days after the primary, had Budd leading Beasley by eight points, 47% to 39%. The John Locke Foundation’s latest Civitas Poll, conducted a couple of days later, showed a narrower race with Budd at 44% and Beasley at 42%.

The first thing to say about these findings is that given the sample sizes of the two polls (635 for ECU and 600 for Locke), the variance between the two findings might be attributable to sampling error.

The second thing to say is that the two samples differ in a key respect: ECU surveys registered voters as a whole while Locke’s polling vendor Cygnal seeks to exclude registered voters who are unlikely to vote. Screening for likely voters in both fraught with peril and necessary if you’re trying to depict how an election might turn out if it were held at the moment the poll is being taken.

How might screening for likely voters explain Budd’s smaller, statistically insignificant edge in the Locke poll? To answer that question, let’s look at another question common to both surveys: the generic-ballot test.

ECU asked it this way: “If the November 2022 midterm elections for U.S. Congress were held today, would you vote for the Democratic Party’s candidate or the Republican Party’s candidate for Congress in your district?” Respondents picked the Republican 47% of the time and the Democrat 44%.

The Locke poll placed its congressional question after a similar question about state legislature, which found a six-point advantage for the GOP. The next question read like this: “And regardless of how you would vote at the state level, if the general election for U.S. Congress were held today, would you vote for the Republican candidate or the Democratic candidate?” Precisely 50% of respondents picked the Republican, with 43% picking the Democrat.

In other words, the ECU poll showed Budd matching generic support for Republican candidates (47%) but Beasley trailing a generic Democrat by five points. Meanwhile, the Locke poll showed Beasley nearly matching generic support for Democrats but Budd trailing a generic Republican by six points.

Assuming these contrasting patterns are more than just statistical noise, I think Locke’s likely-voter screen may be the explanation. Although Budd won the GOP primary by a comfortable margin, it was a contentious race in which 41% picked a different candidate. Voters in primaries tend also to vote in the fall, as well, so screening for likely voters may have had the effect of removing from the sample some “soft” Republicans, independents, and conservative Democrats who recognize Budd’s name (because of months of advertising) but aren’t necessarily going to cast general-election ballots. That served to magnify the slice of likely GOP voters who picked Pat McCrory or Mark Walker and weren’t yet ready to switch their allegiances to Budd.

As for Beasley, because she didn’t have a competitive primary, many Democrats may not yet recognize her name (yes, I know she’s run statewide as a judicial candidate but that doesn’t mean she’s well-known). A poll of registered voters would include these “soft” Democrats and Democratic-leaning independents while a poll of likely voters would exclude some of them.

Given everything else we know about the 2022 cycle — such as President Joe Biden’s manifest unpopularity in our state and overall Republican advantages in enthusiasm and fundraising — Budd is clearly in a better position than Beasley to win the U.S. Senate race. It’s no sure thing, however.

Some small but potentially decisive share of Republican-leaning likely voters aren’t yet in Budd’s column. Because former Supreme Court Justice Beasley isn’t some hard-edged partisan with lots of preexisting negatives, Budd can’t assume this will happen automatically. He needs to bring these voters actively into the fold.

John Hood is a John Locke Foundation board member. His latest books, Mountain Folk and Forest Folk, combine epic fantasy with early American history.