February 19, 2018

Early Voting Points to a Tight Election

The election has already begun, voting has started, and guess what: polls which predicted a Hillary landslide are dead wrong. I can say one thing for certain: Hillary will not win in a landslide. And I think there’s a 50% chance she won’t even win. The early voting results paints a mixed, competitive picture of both good news and bad news.

First, the good news. Trump is up in Iowa without any adjustments and Republicans are doing better in early voting than they were in 2012 in Iowa. Georgia is not a battleground state, with Trump up 3 in the RCP average without any adjustments. Maine CD2 is probably going to go to Trump, but it’s unlikely to matter.

The bad news is out west. Trump is matching Romney’s results in Nevada, and Colorado is simply a lost cause. Trump will win Arizona, Romney won it by 9.03%, and Trump is doing well enough in early voting that he’s on track to win it, but he’s doing worse than Romney. Also, Virginia is probably not really a battleground state anymore. It’s gone.

But the really potential good news – and the way Trump can win the election, is a turnout boom in Ohio, Pennsylvania and possibly Michigan. This is where the data is exciting and is going to decide the election. So lets start with that.


Ohio doesn’t release data on party registration for early voting, but they do release data by county, and the early voting results suggests a sea change for 2016.

At this point in 2012, counties that voted for Obama had requested 899,204 (63.83%) ballots, whereas now they have only requested 859,565 (60.32%) ballots. A decline of 39,639.

At this point in 2012, counties that voted for Romney had requested 509,479 (36.17%) ballots, whereas now they have requested 565,351 (39.68%) ballots. An increase of 55,872.

This constitutes a 7.02% swing in the proportion of ballot requests from Democrat counties to Republican counties. That alone would be massive – however, remember that the real partisan effect is muffled because these aren’t party numbers.

Cuyahoga county (where Cleveland is) for example went 68.8% for Obama, which means 31.2% went to Romney. Ballot requests are down 17% in Cuyahoga, even with the Republicans in Cuyahoga who, presumably, are requesting ballots MORE than they were in 2012. And in Republican counties, the rise in ballot requests is also presumably dampened by the Democrats in those counties who are requesting fewer ballots.

I.e. if the counties were all either 100% Democrat or 100% Republican, then we would expect the R-D absentee ballot request gap to narrow 7%.

But if the counties are, on a weighted average, 66.6% to 33.3% in favor of one party (either Democrat or Republican), then we would expect that the Republican-Democrat gap thus far in early voting is actually 21%.

But here’s another problem: Democrats turning out for Trump. Because it’s clear that in Ohio Trump is generating much more enthusiasm, given the rise in ballot requests in Republican counties. And so those Democrats who support Trump are probably more likely to turn out than Republicans who are anti-Trump, because Hillary generates no enthusiasm for a GOP crossover. Though this is just speculation.

Why am I so interested in Ohio? Because right next to Ohio are Pennsylvania and Michigan.

Pennsylvania doesn’t have early voting.

Obama won Ohio by 2.98% in 2012.  Obama won Pennsylvania by 5.38% in 2012. Early voting in Ohio is forcing me into a scary prediction of a 4-11 point Trump win in Ohio. If the GOP-Dem turnout gap ends up being something like 10% more in favor of the GOP than 2012, and in addition Trump wins independent voters and does better in crossover voters,  Ohio could be a blowout.

But Pennsylvania is right next door and only 2.4% more Democrat than Ohio. If Trump steamrolls Ohio, it seems to me that Pennsylvania would likely be in play. If Trump wins Ohio by 5%, then would we not expect him to win Pennsylvania by 2.6%? Perhaps I am overestimating how linked the two states are. But at least we can say “it bodes well”.

Also, in the primaries, Ohio and Pennsylvania had similar turnout changes for both parties. Ohio Republicans had an increase of 78.8%, Ohio Democrats a decrease of 47.9%. Pennsylvania Republicans had an increase of 87.9%, Pennsylvania Democrats a decrease of 30.7%. And given the relation between changes in primary and changes in general election turnout, we should expect Ohio and Pennsylvania to change similarly.

Ohio and Pennsylvania are ground zero of the proposed “monster vote”. If the “monster vote” exists, it will be most concentrated in these two states and to a lesser extent Michigan. If it doesn’t exist well then nevermind.


Current early voting in Michigan shows Republicans leading in ballots returned. In addition, Republicans have returned fewer ballots as a percentage of total ballots they received, meaning that not only are Republicans beating the Democrats in absentee voting, but they have a bigger upside as well:

Michigan did not have early voting in 2012. So we don’t have any baseline to compare this to. However, the fact that Republicans are doing significantly better in early voting than Democrats in Michigan, where Obama won in 2012 by 9.5 points, is certainly surprising, especially given that Democrats usually do better in early voting.

This is certainly a positive sign, but it’s hard to tell what it means without a prior baseline. But given what’s going on just south in Ohio, there’s reason to believe this is a sign that Michigan will be more competitive than in the past. However, at Obama winning by 9.5 points in 2012, even the most optimistic swings in the midwest based on the Ohio data would merely make Michigan competitive.

But frankly, it’s a bit unimportant, because in the scenario in which Michigan is competitive, Trump has almost certainly comfortably won Pennsylvnia, at which point the election is over.


Early voting in Wisconsin indicates a Clinton win. Wisconsin doesn’t have party registration voting statistics, but they do have county data for 2016. However, they don’t have county data for 2012.

In the 2012 election, the counties that Obama won made up 54.42% of the vote, while the counties Romney won made up 46.88% of the vote. This translated to a total vote count of 52.83% for Obama and 45.89% for Romney.

In 2016 early voting as of October 28 in Wisconsin, the counties Obama won made up 55.37% of the early vote, while the counties Romney won made up 44.6% of the early vote. And to the extent this translates to election day, it looks like Trump will do close to how well Romney did in Wisconsin, which was to lose by 6.94 points.

However, in Ohio, counties that Obama won in 2012 were a bigger chunk of the early vote than they were of the overall vote:

2012 Obama Ohio counties as % of overall vote – 58.09%
2012 Obama Ohio counties as % of early vote by October 26 – 63.83%

2012 Romney counties as % of overall vote – 41.91%
2012 Romney counties as % of early vote by October 26 – 36.17%

And so in Ohio, we saw an 11.48% swing from the early voting result at this time in 2012 to election day in terms of county percentage of votes. And that translated into a real vote total percentage of 47.69% for Romney and 50.67% for Obama.

If Trump gets an 11.48% county vote swing in Wisconsin, that would result in 50.13% for Clinton, 49.84% for Trump, ignoring any potential that Trump would do better with independents and with crossovers than Romney did vs. Obama. But this is extremely speculative.

There was apparently data on race for early voting in Wisconsin in 2012 and 2008, an in that, it shows that the early voters are slightly whiter than they were in 2008 or 2012, which is good news for Trump.

Wisconsin is a bit of an enigma given the lack of prior baselines – Trump could be massively overperforming in early voting in Wisconsin and we would never know it because we don’t know how well Romney or McCain did. However, given how massively Democrat Wisconsin has been, and the dubious early voting results, I expect Hillary to win the state by 4.5 points.


Week 1 2012:
R – 110,094 (36.34%)
D – 139,281 (45.97%)
O – 53,610 (17.69%)
Absolute Gap: 29,187
Percent gap: 9.63%

General Election:
Obama – 52.36%
Romney – 45.68%

Week 1 2016:
R – 120,304 (35.69%)
D – 150,484 (44.65%)
O – 66,272 (19.66%)
Absolute Gap: 30,180
Percent gap: 8.96%

General Election Prediction:
Clinton – 50.5%
Trump – 46.5%

Now these numbers are just party registration, it’s possible that Trump is doing much better among democrats and independents than Romney did. Moreover, party registration =/= party ID, meaning that the voters could identity much more as independent. And of course, D doesn’t necessarily mean Clinton, R doesn’t necessarily mean Trump. Certainly the conservative treehouse thinks that Trump and Hillary are neck and neck given assumptions about how many ID independents there are and how many D’s and R’s are crossing over to the other side.

This is all plausible, but I don’t really buy it. If it looks like 2012 and quacks like 2012, it’s probably going to go like 2012.

That said, this is certainly not “eye-popping” as Hillary would have us believe.

North Carolina

Republicans are doing better in the total aggregate of early voting in North Carolina than they did in 2012, and North Carolina went to Romney by 2.04%.

Romney won it in 2012, Trump is doing better than Romney in 2012, NC is probably going Trump. Not much else to say.


In early voting – absentee plus in-person voting, Obama won by 5 points in 2012, and according to Politico in 2012 was ahead by 3 points On November 1st.

As of October 29 2016, Republicans had 1,450,760 absentee + in-person early votes, while Democrats only had 1,427,314. This is a lead of 23,446 votes, and is a 0.65% lead.

And there doesn’t seem to be any major trend.

On October 26, Republicans had 1,014,206 absentee + in-person early votes, while Democrats only had 1,002,481. The lead then was 11,725 votes, and was a 0.47% lead.

And so Obama was ahead in early voting by anywhere from 3-5 points, and ultimately only won the state by one point. Trump appears to be ahead by around half a point, with no weakening of the trend. If Trump extends the lead on election day as Republicans usually do, and does better than Romney in crossover and independents resulting in the slight early voting turnout advantage being an even bigger edge in votes, Trump can win Florida handily, and my prediction is Trump wins by 2.


What I notice in Democrat early voting analyses is they spend a long time talking about Colorado, Nevada, Arizona, Maine, New Hampshire. If they’re stupid they’ll say Hillary is winning in North Carolina.

They don’t talk about Ohio or what that may mean for Pennsylvania.

Previously we’ve talked about the “monster vote” and non-traditional voters coming out to vote in the primary, and it’s hard to say how it’s going to pan out on election day. My belief had always been that, if a “monster vote” existed, it would manifest more on election day than in early voting due to them being more first-time voters and unfamiliar with early or absentee voting. Republicans generally do better on election day anyway, but I predict Trump will, to some degree, have a bigger gap between his election day results and his early voting results than most Republicans due to this bloc of people who voted in the GOP primaries which hadn’t voted in the prior 2 general elections.

Certainly the polls which showed Hillary up 5-7 points have been shown to be completely ludicrous as real votes actually come in. The early voting thus far paints a picture of a very tight, competitive election, vindicating the LA Times and Rasmussen.

But really these are all just inferences based on party registration and location. It’s possible that Trump is totally blowing Hillary out in states where it seems close based on party registration turnout; I don’t think the reverse is likely, that Hillary is doing better than Obama among registered Democrats or Independents.

And we will see very soon on election day if Trump has a chance, because it’s going to come down to Pennsylvania, ground zero of the monster vote should it exist, and right next to Ohio where Trump appears to be steamrolling. If he loses Pennsylvania he’s probably lost the election. There’s a chance he could win Wisconsin, as Wisconsin is a bit of an enigma, but really I think that of the three – Michigan, Wisconsin, Pennsylvania – Pennsylvania is the most likely and if Trump doesn’t win it he won’t win either of the other two.

Trump losing Pennsylvania but winning Wisconsin would be the scenario in which winning Maine CD2 would be important; otherwise Trump would have to pray that the Republican Senate would vote for him, something that is by no means guaranteed.

And since Trump is doing badly in the mountain west, he can’t rely on good results from Colorado or Nevada. Everything else is basically in line, it comes down to Pennsylvania.

Facebook Comments
  • Daniel Balint

    Informative, as always.

  • Kris Polk


    • Gilberto Carlos

      > Full of Mexicans
      > Crimson red

  • A lot of the early voting occurred before Comey reopened the investigation of Hillary. Are you factoring this into your numbers?

  • The early vote in Nevada points to a 1-point Clinton lead so far:


    • Ryan Faulk

      That’s making assumptions about how registered R’s, D’s and I’s are voting in Nevada, and that’s based on polls that have, in general, been wrong about early voting.

      • Do you have any better ideas? Polls in Nevada are much closer this year than in 2012. Plus, there’s less of a Black turnout issue than in the South.

        • Ryan Faulk

          No the polls, at least at the end of week 1, were about on par with 2012 in terms of party registration voting. The idea that Trump is doing better than Romney is not based on the party registration of early voters, but on assumptions about who they’re voting for, which is based on polls.

  • Dr. James Russell, PhD

    This confirms my intuition (usually correct) that PA is the pivotal state in this election, more-so than any other. I’m very whitepilled on his chances here personally.

  • There’s some reason to believe PA and OH might not be as linked this year as in 2012: Trump’s pattern of support in the PA and OH primries resembles the McCain overperformance over Bush in 2008 (terrible in Lancaster County, good in Western PA and Luzerne counties, good in eastern and southern Ohio).

    In 2008, OH and PA were 5.66 points apart.

  • effthis

    What the hell is good about Trump winning. U assume that all Americans are for Trump u are wrong. This American is for an intelligent leader. Trump is a fascist dictator there is nothing good about Trump leading. What is good is Clinton leading in these. We should fear Trump leading because America will get into World War 3. Tuck Frump.

    • fatty


      Hillary will soon be wearing handcuffs.

    • Ryan Faulk

      Regarding WW3, Hillary has a much more belligerent attitude toward Russia, wanting to shoot down Russian planes over Syria, and was the secretary of state (i.e. head of foreign policy just below Obama) and engineered the invasions of Syria and Libya, while at the same time allowing Iran to nuke up.

      Trump meanwhile is talking about working with Russia, stopping Iran from nuking up, and keeping Assad in power in Syria, and has said many times that the middle east would be better off with Saddam, Mubarak and Gaddafi still in power.

      It’s amazing how totally backwards you have it on WW3.

      As for “fascist”, Trump’s not the one who probably assassinated Seth Rich and Scalia. Trump’s not the one who looked into drone strikes, cutting off the internet and sending an assassin to kill Assange. Trump’s not the one organizing professional agitators to start fights at his opponents’ rallies. Trump’s not the one carpooling in voters from other states to swing elections. And Trump’s not trying to replace the middle class / small business economy with an open-borders / free trade economy that is destroying the ability of workers to negotiate. All that is stuff Hillary’s campaign is doing.

      It’s just shocking how backwards you have it.

    • lol nigga u mad

  • The Keystone State, indeed.

  • By the way, what happened to the “Pseudo White Pill” article? Taken down?

  • nodroz

    Hi Ryan, I thought New Europe Radio was the best show on TRS. I heard about this site on the latest TDS, and it is nice to see your stuff again.


    It is difficult for me to buy-in that the republicans who voted for Romney in 2012 will vote for Trump in 2016. Early ballots with no party affiliation requested from voters in traditionally republican counties, cannot be extrapolated as automatic votes for Trump, any more than ballots requested from traditionally democratic counties are votes for Clinton.

    Even if Trump is winning independents/and or no party affiliated voters (NPA) like Romney did in 2012, massaging their numbers as Trump’s vote buffer makes no sense.

    There is a likelihood that the Romney republicans will not vote for Clinton but will for down ballot republican races. Still, if a sufficient number of combined Romney and #NeverTrump voters do vote for Clinton all of these NPA state ballot voting results will never accurately reflect who is voting for whom in the general.

    How many Hispanic and Latins are newly registered which would not be captured in a likely voter poll? Also aren’t likely voter polls based on prior election participation, giving the impression that having voted 2008-2016, that they will in the 2016 general election.