Early & Absentee Voting Megathread - Build the Freiwal

[DataGuy]

These numbers in Texas do look pretty insane, but I'd be curious to see what more rural counties look like. If turnout is extremely high in ruby red Republican territory as well, that doesn't bode nearly as well for Beto.

Some rural counties in Nevada are reporting, and turnout seems higher than Saturday, like in Washoe. Republicans were able to cut into the Democrat margin a bit as a result. While Clark will likely more than offset that, it does present a possible problem for Democrats: High Republican turnout could limit the impact of high Democratic turnout.

To your comment on Texas, turnout was in fact record-shattering even in small, deeply Republican counties. From the Texas Tribune:

You must be logged in to read this quote.

https://www.texastribune.org/2018/10/22/texas-early-voting-turnout/

That's a percentage increase of 369% over 2014. (Midland County voted for Trump 75% to 20% in 2016.) By comparison, first-day turnout increased by almost 90% in Dallas County and by 176% in Travis County.

Both Dallas County and Travis County do of course have far more raw votes than Midland, but if smaller Republican counties across the board also see such astonishing spikes in turnout, it could negate any benefit for the Democrats.

[DataGuy]

According to the Texas Secretary of State's early voting report for yesterday, turnout equalled or exceeded 2016 turnout in both Trump counties and Clinton counties.

My formulas were suggesting amazingly high vote tallies that even I had a hard time believing, but maybe it'll be right after all. Take it from a local: it's absolutely crazy here. I saw it myself.

[DataGuy]

"Early voting heaviest so far in large Texas counties won by Donald Trump," from the Corpus Christi Caller Times:

https://www.caller.com/story/news/local/texas/state-bureau/2018/10/24/early-voting-heaviest-so-far-texas-counties-won-donald-trump/1752924002/

I'm definitely an early vote skeptic, but I will be keeping an eye on these numbers throughout the week to see if they hold up.

[DataGuy]

Most of the Week 1 is in from Nevada, so it's a little easier now to compare it to the Week 1 vote from 2016. Everyone claims that, unlike other state, the Nevada early vote really is predictive. Here's what it looks like in aggregate right now compared to 2016:

Week 1, 2018

Democratic: 42.22% - 112,517
Republican: 38.84% - 103,496
Other: 18.94% - 50,472

https://www.nvsos.gov/sos/home/showdocument?id=5950

Week 1, 2016

Democratic: 44.50% - 151,020
Republican: 35.84% - 121,651
Other: 66,728

https://www.nvsos.gov/sos/home/showdocument?id=4543

The gap between registered Republicans and Democrats is actually smaller this year, which might seem counterintuitive. This race will clearly come down to independents. And before everyone automatically assumes they'll break for Rosen, it's worth noting that every recent poll has shown Heller leading among that group.

I still say it's a textbook tossup.

[DataGuy]

Most of the Week 1 is in from Nevada, so it's a little easier now to compare it to the Week 1 vote from 2016. Everyone claims that, unlike other state, the Nevada early vote really is predictive. Here's what it looks like in aggregate right now compared to 2016:

Week 1, 2018

Democratic: 42.22% - 112,517
Republican: 38.84% - 103,496
Other: 18.94% - 50,472

https://www.nvsos.gov/sos/home/showdocument?id=5950

Week 1, 2016

Democratic: 44.50% - 151,020
Republican: 35.84% - 121,651
Other: 66,728

https://www.nvsos.gov/sos/home/showdocument?id=4543

The gap between registered Republicans and Democrats is actually smaller this year, which might seem counterintuitive. This race will clearly come down to independents. And before everyone automatically assumes they'll break for Rosen, it's worth noting that every recent poll has shown Heller leading among that group.

I still say it's a textbook tossup.

You forgot to add in absentee ballots.

Anyway the %s look better for Rs as compared to 2016 but remember that Rs outperformed expectations on Election Day massively in 2016 + won Indys by a surprising margin. Neither of those is likely to happen this year, so Rs HAVE to do better with the early + absentee vote to have a chance - thus far they are doing that, so tossup seems reasonable for the statewide races.

I was just using in-person turnout because I think it's a slightly better measure of enthusiasm, but counting the mail-in ballots it looks like this:

Week 1, 2018

Democratic: 41.36% - 128,500
Republican: 39.89% - 123,951
Other: 18.75% - 58,256

Week 1, 2016

Democratic: 43.71% - 167,913
Republican: 36.65% - 140,792
Other: 19.63% - 75,408

[DataGuy]

Here is the early voting data from Texas yesterday, relative to the number of registered voters in the county and the number of votes cast on the same day in 2016. I broke it down by Trump and Clinton counties, to get a general idea of which regions might have been more "enthusiastic."

Trump Counties

Tarrant County: 3.44% of reg. voters, 93.7% of 2016.
Collin County: 4.61% of reg. voters, 99.5% of 2016.
Denton County: 4.45% of reg. voters, 94.0% of 2016.
Montgomery County: 3.43% of reg. voters, 84.1% of 2016.
Williamson County: 2.41% of reg. voters, 102% of 2016.
Galveston County: 2.83% of reg. voters, 82.5% of 2016.
Nueces County: 1.59% of reg. voters, 78.0% of 2016.

Clinton Counties

Harris County: 3.41% of reg. voters, 105% of 2016.
Dallas County: 2.53% of reg. voters, 110% of 2016.
Bexar County: 2.84% of reg. voters, 88.7% of 2016.
Travis County: 2.82% of reg. voters, 109% of 2016.
El Paso County: 1.45% of reg. voters, 85.2% of 2016.
Fort Bend County: 4.37% of reg. voters, 107% of 2016.
Hidalgo County: 1.85% of reg. voters, 90.5% of 2016.
Cameron County: 1.29% of reg. voters, 108% of 2016.

Averages

Trump County averages: 3.25% of reg. voters, 90.5% of 2016.
Clinton County averages: 2.57% of reg. voters, 100% of 2016.


Trump counties did better than Clinton counties in terms of basic turnout percentage, but Clinton counties clearly outperformed Trump counties when compared to the number of votes cast on the corresponding day in 2016. That suggests that relative enthusiasm was significantly higher yesterday in Democratic areas as opposed to Republican areas, which is something of a reversal from earlier last week. I'll be looking over the next few days to see if the momentum really has shifted, but I would say that yesterday probably wasn't a very great day for the GOP in Texas.

[DataGuy]

Great weekend for Democrats in Florida. They have cut down the Republican advantage to 2.2%. It was 6.4% at this time in 2014.

Republicans: 1,151,593 (42.3)
Democrats: 1,092,547 (40.1)
NPA/Minor: 482,252 (17.7)
Republican edge is 59,046 (+2.2%)

But aren't Democrats supposed to do better on weekends anyway? If they couldn't eliminate the GOP lead over their best days, how likely is it that they'll ever catch up in the early vote?

[DataGuy]

I took a look at the amount the final RCP average was off the actual results in 2016 in some of these states.  This is what I found:

FL   +1.0 R
IN   +8.3 R
MO +7.5 R
TX   -3.0 R
AZ   -0.5 R
NV   -3.2 R

If this were added/subtracted from the current RCP results, the Republicans would take IN, MO, TX and AZ (by 0.2). The Democrats would take FL and NV.

Ah yes, because polling error is always the same every election. Obama totally didn’t overperform polling in nearly every battleground state in 2012, after all.

It's not sound to assume that polls will always miss the exact way they did in 2016, but at least in the House of Representatives, history does suggest that the incumbent House party is favored to overperform the generic congressional ballot. On average, the party controlling the House on Election Day has overperformed the GCB by 2.1%. Controlling the House of Representatives actually has a much stronger correlation with overperformance than controlling the White House.

I analyzed that trend recently: https://uselectionatlas.org/FORUM/index.php?topic=304364.0. We'll see if it holds this year.

[DataGuy]

Here is the Texas turnout report for Sunday.

Trump Counties

Tarrant County: 1.29% of reg. voters, 85.7% of 2016.
Collin County: 1.69% of reg. voters, 98.8% of 2016.
Denton County: 1.75% of reg. voters, 90.4% of 2016.
Montgomery County: 1.53% of reg. voters, 83.4% of 2016.
Williamson County: 1.52% of reg. voters, 89.1% of 2016.
Galveston County: 1.10% of reg. voters, 81.8% of 2016.
Nueces County: 0.50% of reg. voters, 74.7% of 2016.


Clinton Counties

Harris County: 1.47% of reg. voters, 95.7% of 2016.
Dallas County: 1.21% of reg. voters, 109% of 2016.
Bexar County: 1.36% of reg. voters, 83.4% of 2016.
Travis County: 1.61% of reg. voters, 116% of 2016.
El Paso County: 0.77% of reg. voters, 116% of 2016.
Fort Bend County: 1.28% of reg. voters, 96.5% of 2016.
Hidalgo County: 0.64% of reg. voters, 91.3% of 2016.
Cameron County: 0.07% of reg. voters, 282% of 2016.


Averages

Trump County average: 1.34% of reg. voters, 86.2% of 2016.
Clinton County average: 1.05% of reg. voters, 124% of 2016.


The % of 2016 numbers might be distorted somewhat by Cameron County, a very small border county whose mere 200-vote increase over 2016 translated into 282%. But even excluding Cameron, it's still 1.34% of reg. voters and 86.2% of 2016 for Trump counties versus 1.19% of reg. voters and 101% of 2016 for Clinton counties.

That means the % of reg. voter gap between Trump and Clinton counties is even smaller than Saturday's (in the Democrats' favor, to be clear) and the % of 2016 gap expanded (again, in the Democrats' favor).

As some of have noted, these might just be weekend numbers, but if these trends hold for today and the coming days I think the GOP just might be in serious trouble.

On the bright side for the GOP, TargetSmart's model for 10/28 still has them leading by 15 points.

[DataGuy]

@DataGuy, would it be reasonable to assume these numbers could be more damaging for the GOP in downballot contests than in the Senate race? Thinking Congress and Legislature especially, as I’m pretty confident in all those Harris Co. Republican execs losing

The majority of people don't give a great deal of thought to downballot races. Unless a downballot candidate is especially well-known in their own right, they're more often than not at the mercy of top-ballot candidates' coattails (or lack thereof). In Texas, I think the implications for downballot candidates will largely depend on who has the stronger influence: Ted Cruz or Greg Abbott? If people generally cast their downballot votes the same way they cast their Senate vote, that's bad news for the GOP.

But if the Senate race is really just an aberration mostly driven by unique personality factors and does not represent a permanent partisan shift, people might match their downballot votes with their gubernatorial votes and pull some struggling GOP candidates just over the finish line. In other words, if the Senate race is a one-off exception for many Republican-leaning voters solely because they don't like Ted Cruz or because they think Beto is just so specially great.

We'll see what happens, but unfortunately for the GOP most polls seem to be showing a stronger correlation with the Senate race and suggesting that Abbott is in fact the exception to the overall trend.

[DataGuy]

The majority of people don't give a great deal of thought to downballot races. Unless a downballot candidate is especially well-known in their own right, they're more often than not at the mercy of top-ballot candidates' coattails (or lack thereof) the voter's feelings about the president. In Texas, I think the implications for downballot candidates will largely depend on who has the stronger influence: Ted Cruz or Greg Abbott? If people generally cast their downballot votes the same way they cast their Senate vote, that's bad news for the GOP.

FTFY

Interesting study, but I have my doubts that it's entirely true. I recently did a detailed analysis of final results vs. the GCB and found that, in the context of House races, which party controls the House of Representatives itself on Election Day has a considerably stronger impact than who controls the White House.

Not to mention that "voters' feelings about the president" might already be somewhat baked into the Senate race.

[DataGuy]

To be honest, all this talk of "firewalls" strikes me as supremely overconfident. Did you know that, in 2016, the Clark County "firewall" was about 73,000 votes, yet Clinton only ended up winning the state by 2.4% and 27,000 votes?

Yes, this is a midterm year, but it's a very unusual one and turnout in NV is expected to be about 80% of 2016 turnout. People are almost acting like this is a presidential election. To be roughly equivalent to 2016, the so-called firewall in Clark would have to be about 58,000 votes. And we're already declaring a Senator Rosen based on what might be a 35,000 or 40,000 vote margin?

The fact is that Democrats are significantly underperforming their 2016 "firewall" both in Clark and statewide, and that still holds true when it's adjusted for this year's expected turnout. Even with their much larger EV padding, Clinton and Cortez-Masto barely won the state. I am not saying that Heller's going to win, but what I am saying is that it's foolish to "call" anything based on the early vote.

To me, it's still a tossup. Yes, Rosen could win. Perhaps there is a reasonable argument for tilting this race in her favor. But to rate this "Safe D" or even "Likely D" is not realistic. This race will still be decided by independents, who could easily go either way.

*Corrected some numbers to reflect exclusively early vote (accidentally added in Election Day vote), but the point still stands.

[DataGuy]

Ralston said in a post this morning that he still considers NV-SEN a tossup, so apparently he agrees with me on that. I still think people are obsessing too much over the Clark "firewall" and not looking at the state as a whole. Democrats are leading statewide by only 12,000 votes, which is a far cry from 2016's ~45,000 EV statewide advantage. Their percentage lead is about 2.6% right now, and I know Ralston once remarked that either party has a legitimate shot as long as the margin stays below 3%.

Although if there's one more poll showing Rosen ahead, I would tilt this race in her favor.

[DataGuy]

Ralston said in a post this morning that he still considers NV-SEN a tossup, so apparently he agrees with me on that. I still think people are obsessing too much over the Clark "firewall" and not looking at the state as a whole. Democrats are leading statewide by only 12,000 votes, which is a far cry from 2016's ~45,000 EV statewide advantage. Their percentage lead is about 2.6% right now, and I know Ralston once remarked that either party has a legitimate shot as long as the margin stays below 3%.

Although if there's one more poll showing Rosen ahead, I would tilt this race in her favor.

Comparing overall #s is useless considering turnout was higher in 2016. Not to mention, there's still 3 days left.

Sure, it's not a direct apples-to-apples comparison, but turnout this year is expected to be about 80% of 2016's turnout. That means to that roughly match 2016's statewide advantage, Democrats would need a margin of roughly 36,000 votes. Still falls short. And if Democrats have barely reached a 12,000 vote margin more than 75% of the way through early voting, they are not gaining an extra 24,000 votes over the final 25%.

[DataGuy]

Republicans are pretty much screwed on Tuesday.

I will likely make my final predictions tomorrow, unless there are some final polls on Monday.

But I'm more going by past Dem wave election trends, special election results and early-voting trends this time and ignore the polls, such as the R-leaning ones in TN and TX (which I think will be really close races).

The overwhelming majority of predictions based on the early vote fail miserably. My model, the final predictions of which I will be posting soon, will ignore the early vote. We'll see what method fares better this time around.

[DataGuy]

Republicans are pretty much screwed on Tuesday.

I will likely make my final predictions tomorrow, unless there are some final polls on Monday.

But I'm more going by past Dem wave election trends, special election results and early-voting trends this time and ignore the polls, such as the R-leaning ones in TN and TX (which I think will be really close races).

The overwhelming majority of predictions based on the early vote fail miserably. My model, the final predictions of which I will be posting soon, will ignore the early vote. We'll see what method fares better this time around.

Other states maybe, but I feel like ignoring early in NV may be a mistake.

Even without the EV, the other data in NV suggests the race is leaning in Rosen's favor. So at this time, I don't think the EV would make that much of a difference in my prediction anyway.

[DataGuy]

Cumulative Texas early vote report, in-person + mail-in/absentee.

Clinton Counties

Harris County: 36.59% of reg. voters, 87.56% of 2016.
Dallas County: 39.66% of reg. voters, 96.33% of 2016.
Bexar County: 37.71% of reg. voters, 87.77% of 2016.
Travis County: 47.43% of reg. voters, 97.45% of 2016.
El Paso County: 30.51% of reg. voters, 92.46% of 2016.
Fort Bend County: 46.14% of reg. voters, 93.30% of 2016.
Hidalgo County: 31.68% of reg. voters, 81.93% of 2016.
Cameron County: 26.37% of reg. voters, 84.96% of 2016.

Trump Counties

Tarrant County: 41.49% of reg. voters, 90.41% of 2016.
Collin County: 49.39% of reg. voters, 94.85% of 2016.
Denton County: 45.96% of reg. voters, 95.29% of 2016.
Montgomery County: 41.42% of reg. voters, 88.08% of 2016.
Williamson County: 48.97% of reg. voters, 100.01% of 2016.
Galveston County: 42.41% of reg. voters, 88.84% of 2016.
Nueces County: 32.96% of reg. voters, 90.36% of 2016.

Averages

Trump Counties: 43.23% of reg. voters, 92.55% of 2016.
Clinton Counties: 37.01% of reg. voters, 90.22% of 2016.


Overall, Trump counties saw higher turnout both in terms of registered voters and relative to 2016's numbers.

Drilling into specific regions, one of the most energized counties was Travis County, which is bad news for the GOP since it's heavily Democratic. But on the flip side, Republicans also have reason to be optimistic due to lackluster enthusiasm in overwhelmingly Democratic South Texas. That suggests that perhaps the Hispanic voter surge, which is said to be coming every election year, might not too big after all.