A couple of weeks ago we told you we were paying for a poll. The reasoning was simple: heading into the August 4 Democratic primaries, the only public numbers on the Governor's race were campaign internal polls pointing in opposite directions (the Holscher campaign's from January telling one story, the Skoog campaign's launch poll telling a very different one). That's not an evidence base.

So, through our nonprofit Civic Clarity, we commissioned a genuinely independent survey of likely Democratic primary voters. No candidate, party, or outside group funded it or directed it. We designed the questionnaire and the analysis ourselves and hired Change Research, an established polling firm, to field it. The whole thing came together faster than we expected…fast enough that the survey was in the field before we'd even finished telling you it was coming!

The results are now in, and we're doing what we said we'd do: publishing the topline numbers, the methodology, and the full crosstabs for anyone to inspect. This isn't unfamiliar terrain for us; Civic Clarity is run by a two-person board, and between us we bring a decade of running a global survey-research agency and years in the education-testing world working alongside PhD psychometricians on how survey questions are built. That background is mostly useful as a source of humility: we know how much a careful poll can tell you, and how easily a careless one misleads. What follows tries to stay on the right side of that line.

Among likely Democratic primary voters, here is where the Governor's race stands. These are the "collapsed" numbers — after undecided voters were asked which way they lean:

Two things jump out, and they pull in opposite directions.

The first is that Cindy Holscher leads decisively. Her advantage is far outside this poll's margin of error: this is not a close race at the top. Whatever else the data show, they don't show a tossup.

The second is that the electorate is remarkably unformed. Even after we pushed “leaners” (people leaning towards a candidate but not fully committed), 44% of likely primary voters hadn't settled on a candidate. Before we pushed them, a 55% majority were undecided, with Holscher at 31, Corson at 8, and Skoog at 5. Nearly half the people who say they'll vote in August have not committed to anyone. That is an enormous bloc of movable voters, and it is the single most important piece of context in this poll. A big lead and a wide-open field are both true at the same time.

The obvious objection to a frontrunner's lead this early is that it's just name ID: the challengers simply haven't introduced themselves yet, and the gap will close as they do. The data complicate that story.

Among voters who have actually heard of each one, the share already backing them runs roughly: Holscher 56%, Corson 20%, Skoog 13%. In plain terms, voters who know Holscher pick her at more than four times the rate that voters who know Skoog pick Skoog. So while name recognition is part of the picture (fewer voters have heard of the challengers), it isn't the whole picture. Holscher isn't just better known; the voters who know her like what they see and commit to her. That's a harder thing for a challenger to overcome than a simple awareness deficit, because it can't be closed with ad spending alone.

A fair counterweight: the support that exists is soft. Of the voters who named a candidate, only about 36% said they're "definitely" voting that way; 64% could still change their minds. Combine that with the 44% undecided and there is, on paper, plenty of room for the race to move.

Holscher's coalition is not just larger; it's more likely to actually show up.

Her supporters are meaningfully more engaged than the electorate as a whole. On a combined measure of how closely voters follow Kansas politics, Holscher backers run about 16 points above the average; 33% of them say they follow "very closely," against 25% of all likely primary voters. They're also more motivated to vote: 91% rate their motivation an 8, 9, or 10 out of 10, compared with 82% overall, and 76% are maxed out at a 10, versus 66% of the electorate. The undecided bloc, by contrast, skews toward lower-attention, lower-motivation voters.

This matters because August primaries are low-turnout affairs, and low-turnout electorates are disproportionately made up of the most engaged voters — the very people already breaking for Holscher. So a realistic turnout model makes her position somewhat stronger than the raw topline suggests, and the challengers' path narrower, because a large share of the persuadable undecideds are exactly the people least likely to cast a ballot. The movable vote is real—It's just concentrated among the least reliable voters.

Cindy Holscher is the frontrunner on every dimension we measured. She leads across gender and education groups, runs modestly stronger among women and college-educated voters, and carries the most committed, most engaged base in the field. Her net favorability among those who've heard of her is strongly positive. We found no obvious issue- or favorability-based vulnerability in the data that a challenger could plainly exploit.

Ethan Corson sits in a clear but distant second. He's been running for nearly a year, and that shows: his awareness-to-support conversion (about 20%) is far better than Skoog's, and he holds a consistent second place. The honest framing is that Corson is the established alternative to Holscher—but a meaningful part of his edge over Skoog is simply a head start.

Curt Skoog lands third, with an asterisk that matters. He entered the race on June 1, only about ten days before we went into the field. His weak topline is therefore a cold-start number, not a ceiling. And there's real evidence he's underexposed rather than rejected: 47% of voters have never heard of him, but among those who have, his net favorability is positive (+11 statewide). His problem, in other words, looks like reach, not damage: the voters who form an impression of him tend to form a good one.

Here's the constraint that has to travel with that optimistic read, though: as of this poll, that goodwill isn't converting into votes. The 13% awareness-to-support conversion is the lowest of the three. So the favorable interpretation (late entry, positive impressions, room to grow) is genuinely available, but it comes welded to a hard question: can he turn being liked into being chosen, in roughly eight weeks, against a frontrunner whose own supporters are already locked in? The current data show that conversion isn't happening yet. Whether it starts is the open question of his candidacy.

One structural point: the path to denying Holscher the nomination almost certainly runs through consolidating the non-Holscher vote behind a single challenger. With both Corson and Skoog drawing from a similar lane, a split field is the scenario that most helps a frontrunner win with a plurality. That's just the arithmetic of the race as it currently stands; whether it changes is up to the candidates and the voters.

We asked likely primary voters to name their top issues (up to three). The ranking:

One important caveat about what this does and doesn't measure. This question captures salience: what voters say matters to them, not which candidate "owns" an issue. We deliberately did not test candidate-by-issue strength, so please don't read any of this as "Candidate X is stronger on healthcare." The data can't support that claim.

What is notable is that Holscher's supporters prioritize the same issues, in roughly the same order, as the electorate overall. There's no issue cleavage splitting this primary. It's being decided on candidate factors, not a policy fault line.

We also tested the Democratic U.S. Senate primary, the more crowded of the two races. The collapsed numbers:

Adam Hamilton (the late, big-name entrant) leads, and his lead over the rest of the field is clear. But the headline here is the 55% undecided, even after leaners. If the Governor's race is unformed, the Senate race is barely formed at all: more than half of likely voters haven't landed on anyone, and the candidate in first sits at 18%. Hamilton's net favorability among those who've heard of him is positive (+18), and 47% haven't heard of him at all—a profile that, like the Governor's race, reflects how recently and unconventionally the front of this field took shape. Prediction markets had implied near-total certainty about this race. The electorate, as of mid-June, is a long way from certain.

Christy Davis is currently running second, on voter alignment between with Holscher:

Some of you have seen the Skoog campaign's own launch poll, which showed him substantially stronger than our numbers do. That gap is real, but both polls can be legitimate while measuring different things.

The key difference is what each survey asked, and when. The Skoog campaign's poll was recruited with the campaign's own messaging, and respondents saw an introduction to the candidate before they were asked the ballot question. That measures something closer to a candidate's ceiling under a favorable introduction: support after voters have just been told a campaign's best case. Our poll asked the ballot question cold, with no introduction, which measures something closer to where the race stands right now. Neither approach is wrong. But they are not directly comparable, and the gap between them is not evidence that either poll is flawed. If anything, the distance between "Skoog after a positive intro" and "Skoog cold" is itself a measure of the work his campaign has to do to move voters from the first state to the second.

A number you'll want for context: how many people actually vote in this thing. Our projection, based on historical turnout in contested Kansas Democratic primaries, is roughly 160,000 to 175,000 ballots — about a third of the state's roughly 495,000 registered Democrats. Treat that as an estimate with a real range (call it 140,000 on the low end, 200,000 on the high end), not a precise figure.

2018 is the closest contested-primary benchmark, but unlike 2018 this primary also carries a statewide constitutional amendment on judicial selection, which adds real upward uncertainty to any turnout estimate; 2022's abortion amendment shows how sharply a ballot measure can lift primary turnout, though that vote was a far higher-salience case.

Polls are most dangerous when people stretch them past what they measured, so let’s be explicit about the limits.

This poll did not test the general election. There is chatter in Democratic circles arguing that a particular nominee would or wouldn't fare better against the likely Republican in November. We have no data on that. We did not ask a single general-election matchup. If you see the electability argument made, it isn't coming from us, and we'd ask you not to attach our name to it. We can't confirm it, deny it, or quantify it.

A lead in June is not a result in August. With 44% undecided and soft support among those who've chosen, this race retains real uncertainty. But "uncertain" is not the same as "about to flip." The honest read is that the path for a challenger is contingent and uphill; it requires consolidating the field and turning out exactly the lower-engagement voters who are hardest to turn out. We're trying to describe the race, not handicap it in anyone's favor.

We don't have a horse in this. Capitol Bee and Civic Clarity are strictly nonpartisan. We are not endorsing, we are not advising any campaign, and we ran this poll to put the same independent information in front of every voter, donor, and candidate at once. We fully expect at least one campaign to scrutinize these numbers; good! Every figure in this piece traces directly to the published crosstabs, and we'd rather you check our work than take it on faith.

The most useful way to explore all of this is on SimKansas, where we've posted the full results in an interactive format alongside our scenario models — which are projections, not predictions:

You can also read the raw vendor documents directly: the topline results, the methodology statement, and the full crosstabs.

If you'd like to help fund the next one: we're weighing an independent look at the state Supreme Court retention question and a general-election poll once the primaries resolve — you can contribute here. And if you'd rather not, sharing these results so they reach voters who need them helps just as much.

The whole value of this poll is that you can check it, so here's how it was built.

Change Research surveyed 1,022 likely Kansas Democratic primary voters from June 11–15, 2026. Respondents were recruited by text message sent to cell phone numbers drawn from the voter file for individuals who qualified for the sample (registered Democrats, plus the unaffiliated voters who indicated they will vote in the Democratic primary), and were directed to a self-administered survey on the web. The survey was commissioned by Civic Clarity; questionnaire design and analysis were ours, and Change Research handled fielding and programming. Results were weighted to the expected primary electorate on age, gender, race/ethnicity, education, and region. The modeled margin of error is ±3.02% for the full sample, and larger for subgroups.

Two things to keep in mind when reading the numbers. First, subgroups carry substantially larger margins of error than the topline. The supporter bases for Corson and Skoog in particular are small, so any breakouts of who their voters are should be read as directional, not precise; the Holscher coalition and the undecided bloc are the only groups large enough to read with much confidence. Second, we've drawn an explicit line between the initial ballot (before leaners, with about 55% undecided) and the collapsed ballot (after asking undecideds which way they lean, 44% undecided). We've used the collapsed numbers as the headline but reported both, so the size of the genuinely undecided bloc stays visible.

A word on why we built it from the voter file rather than asking you, our readers, to take it. A poll recruited from people who already read a politics newsletter isn't a representative sample of Kansas primary voters; it's a sample of unusually engaged news consumers, who skew differently on age, partisanship, and turnout. Sampling from the voter file starts instead from a known population of actual registered voters, which is what lets the result be weighted to mirror the likely electorate. That's the standard the American Association for Public Opinion Research points to, and it's why we kept our own readers out of the data on purpose. No method is perfect — non-response and likely-voter modeling introduce real, unmeasured error on top of the reported margin — but starting from the voter file is the part that makes everything else defensible.

Capitol Bee is a nonprofit civic journalism project of Civic Clarity, Inc., a 501(c)(3) organization. We don't have a paywall because we believe transparent government reporting should be accessible to everyone. State legislative bills are analyzed at BillBee.ai. Election forecasts live at forecast.capitolbee.com. Glossary of terms here.