October, 2018:

Here are the totals for Monday, and here are the daily totals from 2010, from 2014, and from 2016, as well as a spreadsheet with totals from 2004, 2008, 2012, and 2016. The running tallies:

Year    Mail    Early    Total
==============================
2010  42,795  214,778  257,573
2014  57,929  163,275  221,204
2018  78,590  494,712  573,302

2008  46,085  376,761  422,846
2012  57,031  429,186  486,217
2016  85,120  555,383  640,503

The in person total yesterday was roughly what it was on Friday, which is to say on the high end for Week 1 but not a step up. My guess is that today and Wednesday will be similar, Thursday will be about the same or a bit higher, and Friday as per usual will be the busiest day, maybe fifty percent or so higher than the totals we’ve seen so far. Again, roughly speaking, that puts us in range for 850K to 900K for the early voting period, perhaps a bit more than the “45% in the first five days” scenario I outlined here. Could still be more, likely won’t be less. We’ll all then guess what next Tuesday’s turnout will be. Have you voted yet? If not, when do you plan to hit the polls?

I’m just going to quote this bit from this story about how the Senate campaigns are interpreting the early vote turnout so far.

Derek Ryan, a GOP data consultant who previously worked for the state party, said there are a couple metrics among the 15 counties that could be heartening each candidate. In Ryan’s analysis, Republican primary voters currently have a 90,000-vote advantage over their Democratic counterparts in early voting — a margin that is “definitely going to help Cruz out considerably,” Ryan said.

O’Rourke, meanwhile, could be boosted by early voters who have not voted in a general or primary election over the last eight years — currently 8.5 percent versus 5 percent for the entire 2014 period, according to Ryan’s analysis.

“The campaigns are seeing the same numbers that we are,” Ryan said. “Cruz is probably focusing on these primary voters. Beto’s probably optimistic about the ones that don’t have any primary election history.”

Two additional pieces of context to add here. One is that it’s always helpful to have a point of comparison. What kind of primary voter advantage did Republicans (presumably) have in 2014? My guess is that it was greater than it is now, but it would be nice to know that. We can also tell a bit more about those people with no primary history; I’m sure Derek Ryan knows that, he may just not want to do that kind of analysis in public. There’s also the question of when each party’s voters tend to come out. In Harris County at least, the first five days tend to be Republican, the weekend belongs to the Democrats, then the last five days generally trend in the Dems’ direction from the baseline of the first week. From what I know, this pattern has held true so far, at a higher Democratic level than in 2014. Whether that will continue in this highly atypical year is anyone’s guess.

Anyway. Here are the totals for Sunday, and here are the daily totals from 2010, from 2014, and from 2016, as well as a spreadsheet with totals from 2004, 2008, 2012, and 2016. The running tallies:

Year    Mail    Early    Total
==============================
2010  40,553  181,084  221,637
2014  57,546  137,137  194,683
2018  77,347  429,009  506,356

2008  45,361  314,252  359,613
2012  53,131  362,827  415,958
2016  80,681  486,060  566,741

We are now well past the cumulative EV total for 2010. I expect we’ll exceed the 2008 and 2012 totals by the end of the week; if past patterns hold and the final day or two of early voting have the highest individual day totals, we could exceed 2012 by a lot, and maybe approach 2016. Historic patterns have held for the first week, so I’d say the odds are they’ll hold for the second week. We’ll know soon enough.

This is the point in the early voting process where early voting hours expand, and as a result daily EV reports come in later. That may affect my ability to present the latest data each day, so I’m going to break the pattern today and engage in one of my favorite exercise, which is to use the data we have so far and make some wild guesses about where we may end up. Let’s take a look back at the first five days of early voting from the past elections we’ve been tracking, and see what fraction of the final EV total they were, and then how much of the complete vote was cast during the EV period. We begin with a table:

Year  5 Day EV  Final EV  5 Day%
================================
2010   164,190   447,701  36.67%
2014   158,399   379,282  41.76%

2008   260,105   746,061  34.86%
2012   313,405   777,067  40.33%
2016   452,124   985,571  45.87%

I’ve separated the Presidential years from the non-Presidential years because we generally have very different electorates in each, and as such the behavior of one crowd may not be that predictive of the other. This year sure seems more like a Presidential year, so we’ll take all the numbers into account. The other factor, as you can see above, is that there has been a steady shift towards more and earlier early voting. Week 2 of early voting is always busier than Week 1, though that is becoming less the case. My guess is that we’ll see a pattern more like 2014 or 2016, but we can take a broader range of possibility into account:

380,266 at 35% = 1,086,474
380,266 at 40% = 950,665
380,266 at 45% = 845,035
380,266 at 50% = 760,532

I have a hard time believing we’ve already seen half of the early votes, but it’s possible. I think the third possibility, which would be approximately what we saw in 2016, is the most likely, though as with all things this year I hesitate to be too definitive. Note that outside of the last scenario, the early voting total will surpass the entire turnout for any off year in Harris County. The question here is not whether we’ll break records, it’s by how much.

The other side of this equation is projecting final turnout from EV turnout. We go once again to the historic data:

2016 = 73.61% early
2014 = 55.13% early
2012 = 64.53% early
2010 = 56.03% early
2008 = 62.76% early

Again we see a distinction between the Presidential and non-Presidential years, and again we see a trend towards more of the vote being cast early, 2014 notwithstanding. So again, we consider a range of possibilties:

760,532 at 75% = 1,014,082
845,035 at 75% = 1,126,713
950,665 at 75% = 1,267,553
1,086,474 at 75% = 1,448,632

760,532 at 65% = 1,170,049
845,035 at 65% = 1,300,053
950,665 at 65% = 1,462,561
1,086,474 at 65% = 1,671,498

I’m basically assuming this will be more like a Presidential year in terms of when people vote. It makes no sense to me that we’ll have nearly half the vote cast on November 6, so I’m not going to calculate a 55% scenario. Even with the most conservative projections, we’re on pace to top one million, and beating past Presidential years is within range. Final turnout in 2008 was 1,188,731, and it is certainly possible we could top that. I have a hard time wrapping my mind around the idea that 2016’s mark of 1,338,898 could be exceeded, but I can’t rule it out. Ask me again after early voting is done. Like I said, it’s not a question of whether we’ll break records, but by how much.

UPDATE: The Saturday EV totals came in a bit before 9. Google Drive is being unresponsive so I can’t give you a link, but I can tell you there were 8,646 mail ballots received, 79,641 in person votes cast, and the overall total is up to 468,549, which is more than the entire EV turnout of 2010. As the man once said, hold onto your butts.