One brief comment on this Houston Landing story about undervoting, and then I will have some numbers for you.
Tens of thousands of Harris County residents stopped voting well short of the end of their Nov. 5 ballots, skipping dozens of races and passing up the opportunity to change the outcome of all but one countywide contest.
Known as undervoting, the number of people that left races blank varied by contest, ranging from .6 percent to a little more than 9 percent.
That .6 percent represents 9,822 voters who bypassed the presidential contest between Republican Donald Trump and Democrat Kamala Harris, the first race on the ballot.
The gap generally grew wider as voters moved down the ballot, ending with 144,571 who did not make a selection on the question of whether to raise the property tax rate for the Harris County Flood Control District. The winning margin was 39,198 votes.
[…]
Numerous contests were decided by a few thousand votes but skipped by more than 100,000 voters.
For example, in the 177th Judicial District, which flipped from Democratic to Republican control this year, Emily Munoz Detoto won by 5,864 votes. Meanwhile, 110,630 people who cast a ballot left that race blank.
In the contest for Harris County tax assessor-collector, Democrat Annette Ramirez beat Republican Steve Radack by just under 34,100 votes. More than 90,000 voters, however, skipped that race.
“It’s part of the life of a politician,” Radack said. “If you’re further down the ballot, you’ve got to know there’s a certain number of people that don’t care or don’t know.”
Perhaps not surprisingly, the number of undervotes exceeded 600,000 in uncontested races, with voters presumably deciding their support was not needed.
Only the race between Democratic Harris County Sheriff Ed Gonzalez and Republican Mike Knox would have been unaffected by the undervote. Gonzalez won re-election by more than 93,000 votes. The number of people who did not vote in that race was about 82,000.
The possibility that a race could flip if everyone had voted in it is only partly a function of how many undervotes there are. For the outcome to change, the trailing candidate would have to win a greater share of a smaller number of votes than they have already received. In a race with a tiny margin, where in a vacuum one may assume that the undervoters are basically a coin flip, there can be a decent albeit still minority chance that enough of those coin flips favor the laggard. In cases where the margin is greater, those odds diminish rapidly, to the point where you’re more likely to be bitten by a shark that is then struck by lightning. The math is just brutal.
All of that assumes that the undervoters are a random sample of the electorate as a whole. That may not be true – it may be the case that more undervotes came from precincts that heavily favor one party or the other. It’s possible to do that work and assign an expected value for the undervotes in question, but we’re still assuming that the undervoters are representative of their precincts. We’re also assuming that they didn’t have a reason beyond “I didn’t feel like it” to skip that particular race. I think those are reasonable assumptions, but they’re still assumptions. At some level, it’s all a guess.
Anyway. I promised you some numbers, so here they are.
2020 2024 2020% 2024% Winner
======================================
83,611 98,078 5.05% 6.26% D
86,623 102,744 5.23% 6.55% R
91,780 104,785 5.54% 6.68% R
88,352 102,809 5.33% 6.56% D
91,886 108,455 5.55% 6.92% D
93,171 110,584 5.62% 7.05% D
94,804 111,935 5.72% 7.14% R
108,196 112,997 6.53% 7.21% R
105,880 115,383 6.39% 7.36% R
105,259 118,450 6.35% 7.56% D
104,103 117,669 6.28% 7.51% R
105,551 118,909 6.37% 7.59% R
102,111 122,356 6.18% 7.81% D
104,456 124,838 6.31% 7.96% R
124,167 7.92% D
106,708 6.81% R
115,320 7.36% D
97,556 112,717 5.89% 7.19%
The first column is the undervote total for each of the contested district and county judicial races in 2020; the second is the same for 2024. Similarly, the next two columns are the undervote rate expressed as a percentage of total turnout. I didn’t list the names of the races because they don’t line up (and they really don’t matter). The numbers are in the order that the races appeared on the ballot – that Landing story has a chart for 2024, so you can figure out which number goes with which race for each. Note that they wrote this before the official canvass was released, so they will differ slightly from my figures. There were three more contested judicial races in 2024 than there were in 2020. The last column is just a note of which party won that race in 2024. The last line is the average for each column.
There were more undervotes this year than in 2020. Not a huge amount, but there was an increase and it happened in the context of fewer overall votes being cast. The undervote rate jumped by more than a point as a result. What that happened this year when there were basically as many races as in 2020 is unclear to me. Maybe the change in voting machines made a difference – maybe it took longer to vote and people were a little more likely to lose interest as a result. You can see that generally speaking, the undervote rate increases as we get farther down the ballot. There are a couple of complicating factors here, though.
One is that the rate doesn’t uniformly increase as we go down. Notice that in both 2020 and 2024, the rate dipped, then went back to increasing, with the fifth race having a higher rate than the third. In both years, that fourth race featured Democrat Michael Gomez. In 2020, there were several races that came after the high-water undervote rate of 6.53%. Those races featured, in order, Democrats Natalia Cornelio, Julia Maldonado, and Leslie Briones, and Republican Linda Garcia. It seems that the presence of a Latino candidate in the judicial races can increase participation in that race by a little bit.
Another factor is that just before the last two races, which is the end of the District Court elections, is the District Attorney race, followed by two County Court races. The undervote rate in the DA race was 5.45% in 2020 and 6.16% in 2024. Well below the average in each case, and also in each case the judicial races that followed got a bounce as well. In 2024, that first race after the DA again featured Republican Linda Garcia, so there were two causes for the bounce. I’m just saying, there are more dimensions to this than one might think.
There are other races that then come after the last of the judicial races. If fatigue were the big issue, you’d think that the undervote rate would keep increasing. But that’s not what happens.
Year Race Undervote Pct
====================================
2020 County Aty 100,251 6.07%
2024 County Aty 106,526 6.80%
2020 Sheriff 79,737 4.83%
2024 Sheriff 84,256 5.37%
2020 TaxAssessor 80,107 4.85%
2024 TaxAssessor 92,263 5.89%
Those races all have lower undervote rates than the judicial races before them, and with the exception of County Attorney in 2020 they’re all under the average for the undervote rate. People are still paying attention way down at the bottom of the ballot. They presumably feel like they have more information about these races than the judicial ones, but I don’t think it’s the case that they think they have more information about the earlier judicial ones than the later ones. I don’t have a good answer for this, I’m just telling you what the numbers say.
I should note that I ran the numbers for this in 2022 as well, where again there were a few scattered Republican victories, but I never got around to finishing and publishing it. I’m going to try to do that now. Spoiler alert, the pattern is very similar.
One last thing, while the overall undervote rate was higher in 2024 and the range from top to bottom was higher in absolute numbers, the undervote share didn’t increase more proportionally in 2024 than it did in 2020. The 6.53% undervote rate in 2020, the highest one for that year, is 29% higher than the low score of 5.05%, while in 2024 the high undervote rate of 7.96% is 27% more than the low undervote rate of 6.26%. Just in case you were wondering.
Anyway, this is all a lot of words and numbers for a truly niche topic, but I figure that’s what you come here for. Let me know if you have any questions.
In Harris County, Harris = 802,370 votes, Trump = 720,046 votes. In the HC races, most GOP candidates either matched or exceeded Trump’s numbers while the DEM candidates dramatically underperformed Harris’ numbers (~60,000 less votes). Colin Allred’s HC vote total exceeded some HC DEM judicial candidates’ vote totals by ~100,000. So, it appears a significant number of local DEMS voted for Harris and Allred, but then jumped ship and voted for the HC GOP candidates. This also happened in 2022 (our local DEM candidates significantly underperformed Beto’s Harris County numbers).
I hope our local Democratic leaders understand why they are losing local support and take corrective action. If this trend continues, DEMS may lose some county leadership races in 2026.
Just to provide more background: In the 2022 Harris County Judge race, Lina Hidalgo got 42,495 fewer votes than Beto O’Rourke. At the same time, Hidalgo’s GOP opponent Mealer got 44,288 more votes than GOP Greg Abbott. There was definitely some local Democratic ticket-splitting going on. It looks like the same thing happened locally this year.
https://www.harriscountydemocrats.com/post/harriscountydemocrats-com-comments-on-the-2022-general-election-results
This does not explain how Lillian Henny Alexander won by 40,000 plus votes in the 507th district court race . There was an extreme margin here and I don’t see any analysis on why . This race is before Sean Teare’s race and the last judicial race .
Edwin, there isn’t a good answer to that question. Every year, there’s a range of votes that judicial candidates get. There’s no clear correlation to the ups and downs. Sometimes, Latino candidates do better than the baseline, but not all the time. I agree that Judge-elect Alexander’s total was impressive. I can’t point to a single reason why.
Greg, as noted, the same behavior existed in 2020. My explanation then was that there were nominal Republicans who crossed over to vote for Biden instead of Trump. I think that explanation holds this year as well, though perhaps at a lower volume. The big difference otherwise was the voters who didn’t turn out this year, and that was a national phenomenon.
Charles, I agree some moderate Republicans voted for Biden/Harris but then voted Republican in the down ballot races. Still, Trump wasn’t on the ballot in 2022 and our local Dems still struggled to barely win.
As far as our Dem voters staying home, there’s a message there. They are pretty much saying, “none of the above.” It’s definitely time for the Democratic Party to look in the mirror.
Interesting; I often hear people saying they don’t want to vote “because [they] don’t have all of the facts” but I didn’t know how much factual backing there was to that idea so I’m glad to see this discussed here. I wonder how we’d go about measuring the number of voters who would have turned out if “x” (e.g. if the Dem platform had pushed further left).
We spoke last Monday at the DNC event but I got caught up speaking with other patrons afterwards and couldn’t chat with you – I’ve tried to contact you directly about since, but I can’t find your email anywhere on the site and I couldn’t track you down on LinkedIn. Assuming you see my email when I submit this form, could you reach out to me?
Alexander?
It’s all Greek to me, the etymology profession, not to mention wikipedia.
It literally means “defender of men.” What irony. In light of several good incumbent male judges being shown the door in Harris County primaries.
Hypothesis: Being a woman marginally helps in judicial races, esp. in family court, but most voters wouldn’t know that the 507th is a family district court or even know the difference from civil. Alexander says she has had hundreds if clients. Perhaps that gave her an edge in name recognition.
Female status can be inferred from first name (better cue in low-information races than race, which requires a visual).
As for Hispanic surname, is that plausibly a plus in Harris County? What about Bexar County? And if Spanish surname is a plus, is German/Askenazi/Ayrian a liability? (Schaffer, Engelhart), or non-Spanish surname in general?
Then there is the married female issue? For example: Does Mayfield make a candidate more electable than Ibarra? Does the combo help? What about Lucci? Is an Italian last name Latin enough to confer a Latina edge?
I don’t have the answers, but there is food for thought and data to be examined.
When asked why he ran against Medina, he replied, “My last name is Devine, and his last name is Medina.”
Guess who won? Guzman did better when running for the state supreme court, but not so well against Paxton.
“Guzman subsequently came in 3rd in the Republican Primary, receiving 17.5% of the vote and failing to qualify for the subsequent runoff.[17]”
Meme–the dynamics of a Hispanic surname in a Republican primary statewide may be very different than in a general election within Harris County. I recall that Steven Wayne Smith defeated incumbent Supreme Court Justice Xavier Rodriguez 53.5% to 46.5% in a statewide GOP primary in 2002. But Al Gonzales beat Rod Gorman by 58% to 42% in 2000 for the state supreme court in statewide GOP primary results. Given the drift of Hispanic voters to the GOP, a Hispanic surname may be even less of a disadvantage at present in a statewide GOP contest.
2000, those were George W Bush years, and R were actively courting Spanish surname voters. Trump was also courting this time around and Cruz did in the valley. I didn’t see Democrats courting Latino voters other than the generic TV ads.
Re: “Tens of thousands of Harris County residents stopped voting well short of the end of their Nov. 5 ballots, skipping dozens of races …:
How does that even make sense as a factual description? Either they stopped or the skipped and went on to make choices in other races (and referendum issues) further below. Which is it? It could be both, of course, but for different voters: some stopping, others skipping. If the actual ballots are available for inspection, this could be examined more systematically.
Re: “dynamics of a Hispanic surname” (Mainstream), Eva Guzman performance in different races (Meme)
Good analytical point made regarding same candidate’s vote-getting performance at different times and in different races. Demographic and biographic candidate characteristics are constant (though not necessarily the campaign strategy and effort), opponents typically vary in different elections, and electorates also vary (a lot) primary vs. general and state vs. local/district. As for Spanish-named candidates, there are many, so a more ambitious data-driven multivariate analysis (high N) would seem feasible. Examination of candidates’ comparative performance in uncontested elections (raw vote count, obviously not the 100% vote share) allows for inferences under conditions where the opponent as a variable driving voting is removed. Differential vote totals among unopposed candidates will accordingly be driven by candidate-specific attributes, including name (cue for ethnicity and to some extent race) and first names (cue for candidate sex) and even by merits factors (such attorneys voting for or skipping judicial candidates they have professional experience with or reputational awareness of).
Devine vs. Medina
Interesting quote, assuming it’s authentic, but what a candidate thinks could at best have a bearing on decisions to run and who to challenge. More likely, incumbent Medina’s legal troubles while in office played a role in his defeat, but since there was no resulting conviction, it would seem fairer not to repeat the particulars as a general good practice, which also applies in cases where there has been an expunction of records or acquittal.
The problem with the expunction remedy has always been that the private media is not bound by the expunction order, so a person once accused of something may never be able to shed the taint, and that could – in the case of office-seekers — affect their electability.
Regarding ‘Medina’ as a lexical stimulus, it’s the name of an ancient city in Saudi Arabia, but how many voters would connote so? But Wikipedia says “It is one of the oldest and most important places in Islamic history.” Well-educated voters may be aware, but they would also be likely to be better informed about statewide candidates, and less likely to rely on decision heuristics. Still, one could ask, was the incumbent’s surname a liability? Does it mean that Medina suffered a religion-based disadvantage running against a Christion challenger, not to mention one homophonetically divine? We will never know. It’s pure ad hoc speculation. Anecdotal.
@Wolfie…
TL:DR.
Pingback: Texas blog roundup for the week of November 25 | Off the Kuff