Is the Undervote going to be decisive in 2026?

For the past 14 years California has been experimenting with a version of the Open Primary in which the top two vote getters, irrespective of party, advance to the General Election.

This was sold as a feature, not a bug.

The idea was that in a legislative or congressional race in a heavily Democratic district, the Republicans wouldn’t have a shot at electing one of their own, but they could be the deciding vote to elect the more acceptable Democrat. And, conversely, in a heavily Republican seat where Democrats don’t have a shot, they could at least ensure that the more acceptable Republican gets elected in November. The theory rests on a presumption of highly rational behavior from the out-party voters who don’t see a member of their party on the November ballot.

We have many intra-party runoffs in this November general election. This includes 20 contests at legislative and congressional levels, and one on the statewide ballot.

A perfect example of an intraparty contest in which the out-party can be the deciding vote is the 24^th State Senate district. In the primary, Republican support was largely split among two candidates, and totaled 31% of the vote. Republican candidates won all the precincts in Malibu, Torrance and Rancho Palos Verdes Estates, and Democrats dominated in the Santa Monica and West Hollywood portions. This is a seat Republicans couldn’t ever win in the General, but they could have a lot of power in a race between the two top Democrats who advanced to the General, John Erickson and Brian Goldsmith.

On the Republican side we have a barnburner in the 40^th Congressional District where two Republican incumbents, Ken Calvert and Young Kim are facing off in a race where Democratic candidates earned 43% of the vote in a district that was purposefully drawn to be as Republican as possible, in order to make surrounding districts more amenable to Democrats.

So, in these examples, it should be that large chunk of the out-party voters that decide the contest, and in the minds of the authors of the Open Primary, the voter thinking should be simple. They should be asking themselves “which one is the closest to my views?” and voting for that candidate.

But, in practice it’s not that simple. Having seen dozens of these contests, we can now see three different ways in which voters in the out-party can determine their votes.

The Rational Thinkers will attempt to pick the most “moderate” candidate that best aligns to their views. But, that can be a challenge. In the two examples above, how are Republicans supposed to determine who is the most moderate between the Nancy Pelosi-backed Brian Goldsmith and the former Mayor of West Hollywood John Erickson? Or, how do Democrats rationally choose between two Republicans they have generally reviled for decades – Young Kim and Ken Calvert? For some this is the decision akin to choosing to drink battery acid or rat poison – not a real choice with any perceived up-side.

The next group is the Spiteful Voters, those who will vote for someone they don’t align with at all, just to send a message. Think of that 2018 contest between incumbent Senator Dianne Feinstein and then-Senate Pro Tempore Kevin DeLeon. In that race, many were surprised to see DeLeon winning every rural Republican county.  That wasn’t because those Republicans were suddenly supportive of his progressive agenda, they were just protest voting against Feinstein. They weren’t rationally thinking that KDL was A-OK, they were casting a vote out of spite for Feinstein.

And finally, and often the largest segment, are the Opt-Out Voters – those who’s viewpoint was recently well elaborated by Republican blogger Jon Fleishman in his article Sometimes the Best Vote Is Not to Vote at All. These voters cast a vote for governor or president, they vote on the propositions and down-ticket races, and then they leave the same-party contest, sometimes the biggest race in their area, blank. Election officials call this an “undervote.”

The recent analysis produced at https://PoliticalData.Vote gives life to the data in past contests showing just how large this segment of the vote can be and what segments of the electorate are the ones opting out.

At the outset, calculating the undervote is pretty easy. It’s just the number of ballots cast in a jurisdiction minus the number of votes recorded in a specific contest. If 100,000 ballots were cast in a Senate district and only 84,000 of them contain a vote for a Senate candidate, the undervote is 16%.

In a normal Democrat-versus-Republican contest for Congress or the Legislature, the undervote runs in the low single digits. In 2022, the median contested D-versus-R race for Congress saw 2.9% of ballots skip the contest. In 2024, with a presidential race pulling more casual voters to the polls, the medians drifted up to 5-7%. Put two candidates of the same party on the ballot and the number jumps to 13-17%, regardless of office and regardless of year.

Table 1. Median undervote by office and matchup type, 2022 and 2024 general elections (contested races only)

The same pattern shows up on the statewide ballot. In 2022 every partisan statewide contest was a Democrat against a Republican, and the undervote ranged from 1.9% to 4.9%. But the Board of Equalization District 4, where Democrats Mike Schaefer faced Democrat David Dodson in a district with a large Republican electorate, we saw 25.6% of ballots skip the race.

But the more useful analysis is looking at who is skipping. To answer it, we introduce some statistical methods. To calculate this, each precinct’s undervote share is analyzed against the composition of the people who actually cast ballots in that precinct, creating a projection of what populations had the greatest share of that undervote. The result is a mathematical formula that estimates how often different subgroups – like Democrats, Republicans or renters or Latinos or younger voters, left the race blank.

The regression estimates that in the 2022 BOE 4 seat with over 25% dropoff, that high undervote was caused by more than half of Republican voters leaving that race blank. Some legislative and congressional intraparty contests had even greater undervote. Looking at just the races where the out-party was more than a quarter of the electorate we see an average of 41% undervote, with a high of 68%.

Selected same-party legislative contests, 2022 and 2024, ranked by estimated out-party skip rate

In the 2026 General there are eight intra-party congressional runoffs, with particularly hot races in CA 4 (Thompon v. Jones), CA 7 (Vang v. Matsui), CA 11 (Wiener v. Chan), CD 14 (Wahab v Hernandez), CA 34 (Gomez v. Torres) and the aforementioned CA 40 (Kim v. Calvert). There is also the unique Dem v. Independent contest between Kevin Kiley, who recently registered as an Independent and Democrat Richard Pan which should also have significant undervote among Republicans given there is no registered Republican on the ballot.

There are 12 intra-party runoffs for legislative seats, with competitive contests in the aforementioned SD 24 (Erickson v. Goldsmith), SD 26 (Hernandez v. Rascón), AD 12 (Elward v. Lucan), AD 66 (Seo v. Deen) and AD 68 (Penaloza v. Lopez).

In a contest like the CA 7 race between incumbent Doris Matsui and her challenger from the left, Sacramento City Council member Mai Vang, the campaigns have very clear and opposite strategies. Matsui would like to have the out-party voters support her as the more acceptable of the two, while Vang would be happy to benefit from anyone’s protest vote, or be good with a ton of undervote in the more conservative parts of the district that were added under Prop 50.

In the CA 40 race between Kim and Calvert we have seen an amazing transformation of the campaign messages.  In the Primary, Ken Calvert was telling voters that he was supported by Trump, and Kim was a secret liberal who opposed Trump. Now, with these Democrats potentially in play for Kim, the Calvert campaign suddenly finds it important to remind voters that she’s actually a Trump toady, and the previous attacks of her not backing Trump are just evidence that she’s an untrustworthy chameleon.

The Calvert strategy in particular highlights a risk in trying to aggressively to capture these out-party voters.  According to reports, the Calvert campaign is afraid that trying to win over any of these voters as the more acceptable of the two Republicans could hurt them with their MAGA base. If, as an example, he tried to do ads highlighting that he voted for the 2022 Respect for Marriage Act supporting gay marriage, while Kim opposed it, it could potentially help him gain a few Democratic votes, but he would probably lose a lot more Republican votes if they knew he was promoting that vote.

There are 12 intra-party runoffs for legislative seats, with competitive contests in the aforementioned SD 24 (Erickson v. Goldsmith), SD 26 (Hernandez v. Rascón), AD 12 (Elward v. Lucan), AD 66 (Seo v. Deen) and AD 68 (Penaloza v. Lopez).

And we could see something similar to that in the SD 24 race between John Erickson and Brian Goldsmith where Erickson said after the election that he would reach out to the voters who had supported other Democrats in the primary, but had no interest in those voters that supported the two Republicans. Essentially, the theory of the case in this contest could be that it’s more important to shore up the progressive base than risk being seen as kowtowing to the interests of the more conservative parts of the electorate, especially if those conservatives are much more likely to skip this race in the General.

The Insurance Commissioner race is another interesting contest for analyzing the undervote. Jane Kim and Ben Allen, both Democrats, finished first and second in June. Republicans cast about 29% of the ballots in the June primary and Republican candidates earned 37% of the vote. In the primary, the contest already had a 9.2% undervote, the highest of the partisan statewide races. If Republicans skip at the 50% median rate, that alone adds roughly 15 points – a margin that is well outside of what is the likely vote spread between the two candidates.

Finally, there is the extremely competitive L.A. Mayor’s race, which isn’t a true Dem on Dem race, given that it is ostensibly non-partisan. But this contest is very much like these other intra-party races in that Republicans who backed Spencer Pratt in the primary are the effective “out-party” voters who are going to be faced with the same decision.

The LA Mayor’s race has had more public polling than other contests, and one shocking finding from the IGS Berkeley poll in that contest showed that among the Pratt voters, Nithya Raman was rated unfavorably by 87% of his primary voters. That’s a shockingly high number, but still lower than the 91% disapproval of Karen Bass. While we don’t have the same figures for other races, you could presume that Republican voters in these competitive Dem on Dem races, and the Democratic voters in the CA 40 race between Kim and Calvert have a similar viewpoint of their counter-party options.

Spencer Pratt, who consolidated the city’s Republican and anti-Bass vote, finished third with 25.5%, and as of this writing, his position in this race has been to skip voting entirely. He’s actually calling for an undervote which could make it harder for Bass or Raman to capture many of his voters.

The undervote is not trivial. In these same-party runoffs it can be the largest single bloc of voters, bigger than the margin in most intra-party races. These campaigns have a tough job determining if they can simply avoid these voters, or if the better path is trying to convince them that their rat poison is just a bit sweeter and easy to swallow than their opponent’s battery acid.