Suffice it to say I was off base on turnout. I thought we would be on track to top the turnout of the 2003 Mayor’s race, but in the end we fell short of 2015’s level. I’d have lost quite a bit of money on that proposition if I’d bet on it. So what happened? How did I overestimate turnout this year. Let’s start with some numbers. This is a look at how early voting has gone in Houston since 2003 (we have less complete data before then):
Year Early Total Early%
2003 83,225 298,110 27.92%
2005 49,889 189,046 26.39%
2007 36,710 123,413 29.75%
2009 62,428 178,177 35.04%
2011 46,446 121,468 38.24%
2013 80,437 174,620 46.06%
2015 134,105 268,872 49.88%
2017 60,974 150,174 40.60%
2019 109,044 244,979 44.51%
2021 114,060 229,036 49.80%
2023 136,886 247,198 55.37%
For the years 2003 through 2015 plus 2019, these numbers represent the total number of ballots cast in the Harris County part of Houston. That’s mail ballots plus early in person ballots, with the “Total” column being final turnout for that year. In 2017 and 2021 there were no Houston elections, so I went with the overall Harris County numbers. I decided to include those years to see if there was something happening that I had overlooked. Finally, the numbers for 2023 are the votes in the Mayor’s race, which means it does not include ballots that skipped the Mayor’s race and thus is smaller than it should be. The Election Night returns don’t include such numbers for early and Election Day voting; we’ll get those when we have the final canvass. For these purposes, that’s OK.
As we have discussed before, early voting is a much smaller share of the final vote in odd-numbered years than it is in even-numbered years. In 2020, nearly 88% of all ballots were cast before Election Day; in 2022, that figure was 68%. In the odd years before 2023, that number had never exceeded fifty percent. It came close in 2015, then it dropped back in 2019. I figured this year would be more like 2015, maybe sneaking above fifty percent for the early vote. We got a bigger gain in the early vote than that, and I suspect that what we saw this year will be the new normal. That put a damper on my post-early voting projections, but I had always allowed for the possibility that early voting could be a bigger share of the final total than it had been before. So this was a mild surprise, but not a shock.
What was a shock to me was how little the Houston share of the Harris County vote was. Once again, some numbers:
Year Houston Total Hou%
2003 298,110 374,459 79.61%
2005 189,046 332,154 56.92%
2007 123,413 193,945 63.63%
2009 178,110 257,312 69.22%
2011 121,468 164,971 73.63%
2013 174,620 260,437 67.05%
2015 268,872 421,460 63.80%
2019 244,979 389,494 62.90%
2023 253,087 451,203 56.09%
2003 was a weird year in that the state propositions were all voted on in September. This was a deliberate move by the Lege to make it easier to pass the tort “reform” proposition, whose prospect they feared could be negatively affected by the anticipated high-turnout Mayor’s race in Houston. We have since gotten rid of uniform election dates outside of March (for primaries only), May, and November. Go figure. The non-Houston turnout that year came from the Metro referendum; had it not been for that, Houston would probably have been over 90% of Harris’ turnout that year.
2005 was also a weird year, in that then-Mayor Bill White was basically uncontested, but the Double Secret Illegal Anti-Gay Marriage constitutional amendment was on the ballot. That drove a lot of non-Houston turnout, which is why the city’s share of the vote in 2005 was so small. 2007 was also a snoozer of a city election but there wasn’t much else going on, so the percentage fell back into what would become the norm.
Depending on where you start the comparison, the Houston share of the turnout has either been on a steady decline (if you start in 2011) or has been overall flat (if you start in 2007). However you look at it, there was nothing to suggest to me that this number would decline so precipitously. Houston had an open-seat Mayor’s race with a ton of money being spent plus a high profile ballot referendum. There was a Harris County referendum, but it wasn’t contentious, and none of the state propositions stood out as being unusually salient. And Houston’s share of the registered voter total was about where it was in 2019, at around 45% according to the County Clerk’s office. (I inquired about this on Monday.)
And yet, the Houston share of the total vote dropped by almost seven points. From my perspective, this meant my estimate of the Houston early vote way overshot the mark – instead of about 155K, it was only 137K. I’d have made a more modest prediction about the final total, giving real odds of falling short of 2015, if I had realized that.
But I didn’t, so here we are. That prompts two questions, the first of which is what if anything can I take away from this. Right now my answer is, man, I dunno. I just don’t see anything in the data or the conditions that would have suggested this. I’ll definitely take it into account when I’m thinking about the 2027 election, but we’re way too far out from that for me to worry about it. This is a new data point. We’ll take it and go from there.
The other question is why was Houston turnout so weak compared to 2015? The Chron provides some anecdotal data and some editorial hand-wringing. Neither is all that compelling to me, but I’ll address a couple of things. From the first article:
Mark Jones, a Rice University political science professor who has studied Houston’s electorate, said older Houstonians are likely to have longer-standing community ties, and more are homeowners who deal with city services directly. This trend is more skewed now than in federal election years, he said.
“We don’t have a partisan battle in the same way that we have one in gubernatorial elections or presidential elections,” Jones said. “So that’s one mobilization tool that isn’t there.”
Voters in municipal elections also tend to feel they need more information to vote competently, and many are not aware enough to go to the polls since campaign information is less available. Houston’s transient population also leaves many people feeling less connected to the city, Jones said, and many do not realize how directly the city government’s decisions affect their lives.
All of that is true, but it was all true in 2015 as well, so it doesn’t answer the question of why the decline. From the op-ed:
It could be that the anticipated runoff kept people waiting until Dec. 9 to cast a ballot. It could be that the crowded mayoral race is led by two frontrunners who have been staples of Houston politics for decades: where some see experience, others see baggage or stale ideas. It could be the lack of high-profile races, a problem the city of Austin rectified by moving their mayoral elections to presidential election years. It could even be that the switch to four-year mayoral terms lessens the engagement between candidates, incumbents and constituents, as Rice University political science professor Bob Stein suggested on Houston Public Media. Smith thinks part of it could be the long list of mayoral candidates: 17 on the ballot. But across the country, mayoral races tend to see alarmingly low turnout — less than 15%, according to a 2016, 50-city study from Portland State University.
Another explanation for people staying home is apathy, of course. It’s human nature to be drawn to presidential elections, with all their hype and saturating ad campaigns and delicious red meat allegiances. Nonpartisan local races aren’t sexy, but chances are, they’ll have far more impact on your life than whoever is sitting in the White House. That’s what’s frustrating: many people who failed to cast a ballot in this election won’t fail to complain about all the things this election will impact, one way or another: crime, trash pickup, illegal dumping, stray dogs, affordable housing, homelessness, irresponsible development — you name it.
Austin’s elections used to be in May so their turnout was quite a bit lower than ours. The switch to four-year terms might be an issue, but I don’t think we have enough data yet to say for sure. Plus, take a look at some of those vote totals from the two-year era, which really stink in comparison outside of 2003. And if low turnout in Mayoral elections is a nationwide problem, then why are we spending so much time worrying about our particular case? It may just be how things are.
I do think there’s one difference between 2015 and this year, and the Chron sort of gestures at it in the second sentence of their op-ed. In 2015, Sylvester Turner was widely viewed as the frontrunner in the race, but it was not clear who was in second place and thus in the hunt for the runoff. As such, supporters of Bill King, Adrian Garcia, Chris Bell, Steven Costello, and Ben Hall could reasonably believe it could be their guy. To put it another way, if you didn’t like the top guy, or even the top two or three guys, you could vote for the one you did like and still believe that it could propel him into the December overtime contest.
This year, not so much. Once SJL entered the race, which prompted the exits of Chris Hollins and Amanda Edwards (now running for Controller and CD18, respectively), it was seen as a two-person contest regardless of how many other entrants there were. (There were 18 candidates, by the way, not 17. And there were 13 in 2015, so it’s not like this year was a huge outlier.) Of the seven “serious” candidates, two ended up struggling to get to one percent of the vote. It was really hard to see how a vote for a not-Whitmire not-SJL candidate could be meaningful. Some people still voted for those candidates, of course; about 55,000 people pushed the button for someone other than one of those two. But if one of the frontrunners was not your thing, maybe you didn’t care enough about the other races to bother.
Along those lines, there wasn’t that much action in the district Council races, though 2015 wasn’t a banner year for those either. 2015 did feature three open-seat HISD Trustee races plus a fourth race in which a former Trustee knocked off an incumbent Trustee, while this year we had two very low profile races. Maybe that had some effect as well. I tend to think the Mayor’s race drives the turnout and anything other than a referendum is secondary, but this could have reduced turnout a little on the margins.
I’ve rambled on here, so let me wrap up with a few words about the 2023 runoffs. Having a runoff in District D helps SJL a little, while the District G runoff does the same for Whitmire. Neither one should have much effect – these are both already high turnout districts, and having a Council runoff shouldn’t move the needle more than a smidge – but if it’s a close race they’re nice for each candidate to have. I think the all-Dem Mayoral runoff helps the Dem candidates in the At Large and Controller runoffs. Now that we have D-versus-R runoffs in those positions, I hope the HCDP and some other political players will get involved.
Finally, I’m a little surprised there was never another public poll to go with the two that came from UH. It’s not that our Mayoral races draw a ton of polling interest, but I did expect more than that. Maybe we’ll see other outfits have a go at the runoff. As much as I nitpick those polls, I’d still rather have more data points than fewer.