This Chron story about early voting through Sunday made me grind my teeth.
One week in, early voting turnout has surpassed the levels seen in the last two Houston mayoral elections — but not overwhelmingly so — perhaps a surprise given the prominence of front-runners John Whitmire and Sheila Jackson Lee and the national attention the race has earned.
About 101,000 Harris County residents have already voted early this election cycle, either by mail or in person, compared with about 63,000 Harris County residents one week into the 2019 early voting period and about 82,000 residents during the same time period in 2015, the last time there was an open mayoral race.
But raw vote totals do not tell the whole story. There are also significantly more registered voters in the county: nearly 2.6 million registered voters in Harris County this election cycle, up from 2.1 million in November 2015.
If you consider turnout as a percentage of total registered voters, turnout is actually lower this year than in 2015. It sits at just 3.9% of total registered voters in Harris County this election cycle, compared to about 4% during the 2015 cycle.
Well, I agree that the whole story has not been told. This is because “Harris County” and “Houston” are two separate entities, and so one has to be extra careful when making comparisons. Let’s take a deeper and more complex look at the numbers, because I believe the picture being painted here is misleading.
To be clear, article author Matt Zdun is comparing Harris County turnout in 2015 and 2023, and I don’t have a quibble with that. But his references to the open Mayoral race, which was also a feature of 2015, and expressing “surprise” that turnout isn’t higher than it is – even though it definitely is in raw numbers – gives the impression that the Mayor’s race this year is somehow falling short. What I’m saying is I don’t believe you can come to that conclusion with the numbers provided.
What I mean by that is perhaps best expressed as a chart, which shows the number of registered voters in Houston and Harris County, and the percentage of those voters in Houston:
Year Hou RVs Harris RVs Hou Pct
2015 979,401 2,054,717 47.67%
2019 1,085,813 2,329,277 46.62%
2022 1,133,155 2,543,162 44.56%
2023 1,154,157* 2,590,121 44.56%*
The asterisks for 2023 are because those are estimates. I used the percentage from 2022 to estimate the current number of RVs in Houston based on the November 2023 RV figure for Harris County. The real number is likely a bit smaller, but I have no idea how much smaller. Note that if I used a smaller RV figure for Houston, the effect I’m about to show would be greater.
My point here is that the growth in Harris County voter registration is about 30% from Houston, and 70% from not-Houston. As such, given that there’s nothing more exciting for non-Houston people to vote on now than in 2015, the rate of growth of turnout should be a bit slower than the rate of growth of RVs, since most of those RVs aren’t voting in this Mayoral race.
Let me throw another chart at you, this one providing a different turnout estimate based on an estimate of the share of the early vote that’s actually coming from Houston:
Year Hou vote Harris vote Hou Pct
2015 268,872 421,460 63.80%
2019 244,979 389,494 62.90%
The point here is that the turnout report we get doesn’t specify how many of those voters were actually in Houston. But we can estimate that based on those final percentages above. Doing that, using the numbers given for 2015 and 2019, and using 63.0% for 2023 based on the increased voter registration disparity but somewhat mitigated by the open Mayoral race, and we get the following; yes, one more chart:
Year Hou EV Hou RVs EV TO
2015 52,411 979,401 5.35%
2019 39,461 1,085,813 3.63%
2023 63,604 1,154,157 5.51%
I have made at least three assumptions in calculating these numbers. One, my estimate of Houston’s 2023 registered voter total. Two, my estimate of the 2023 Houston share of the Harris County vote. And three, that the final ratio of Houston-to-Harris voters is the same for early voting. Any of these could be off in any direction and by any amount, so my entire construction here is held together by duct tape and dental floss.
But that’s basically my point here. These numbers are more subtle and complex than they are being portrayed by Matt Zdun. My assumptions may be wrong, but I think they’re all very reasonable, and if anything I’ve erred on the side of making that final comparison less favorable to 2023. The bottom line, which I’ve used a lot of words to get to, is that you can’t just conclude that early voting turnout is “down” compared to 2015, not based on just the Harris County turnout and RV totals. Maybe it is and maybe it isn’t, and there’s definitely an argument that it isn’t. Don’t draw conclusions you can’t fully justify.
All this still overlooks two other points. One is that an increase in early voting doesn’t necessarily mean an increase in overall voting. As we’ve discussed before, it may be that what we are seeing is mostly a shift in voting behavior, which has very much been the case in even-numbered years but not in the odd years, at least so far. I have not seen any analysis of the voter roster, which might indicate if what we’re seeing is mostly old reliables with no noticeable bump in new voters. I’m not in a position to analyze that myself, so I’ll just keep caveating myself.
And two, there’s a non-trivial number of Houston votes that get cast in Fort Bend County, too. In 2019 there were 4,187 such votes, and in 2015 it was 4,847. Again, not a huge amount – I have no idea how many so far – but they can be consequential, as the 2015 Mayoral runoff made clear. It wouldn’t shock me if Fort Bend adds in another 6-7K total votes in the end. Comparing Harris to Harris year-over-year is fine – I do it all the time, and it keeps things consistent. Just don’t forget that it’s not the whole story.
With all that out of the way, here’s today’s EV update:
Year Mail Early Total
2015 23,650 73,905 97,555
2019 9,699 66,255 75,954
2023 9,816 106,731 116,547
The final EV totals from 2015 are here and the final EV totals from 2019 are here. The daily EV report is being posted online now here, but I’m still grabbing a copy each day and saving it for my purposes. The Day Eight file is here.
I’ve already said plenty, so I’ll just note that the totals for this Monday and the Week 2 Monday from 2015 are nearly the same – this Monday was slightly ahead, but by like a hundred or so – which means that if I did the same turnout calculation as above, 2023 would be slightly less ahead of 2015. But even if it had slipped behind, the same point would hold. You’ve got to really do the work to be able to say anything about the turnout.