This is the first of two analyses that will attempt to prognosticate Tuesday’s general election. Here we focus on the overall political climate. Our next, shorter, analysis will look at individual races.
2018’s Place in History
It would be easy to say that we are entering uncharted territory. After all, 2.4 times as many voters cast ballots early this year than in 2014, the last gubernatorial election, and early voting turnout nearly exceeds that of 2006, 2010 and 2014 combined. More people voted early this year than in all of 2014, including Election Day. Voters with no recent primary history cast 42% of early votes through the 11th day of early voting, up from 27% in 2014, and it’s likely to rise slightly when the final early votes are included. Nearly one out of every nine early voters had no recent voting history.
So, in many ways, turnout is record-breaking, extraordinary and unprecedented. In other ways, it’s not record-breaking, nor particularly extraordinary and certain not unprecedented. We aren’t truly in uncharted territory. We’ve been here before. Just not when the governor is on the ballot.
Early voting turnout nearly mirrored 2016, the record-setting presidential election in which 73% of votes were cast early, and resembled 2008 and 2012.
Two years ago, voters with no recent primary history cast 46% of early votes, on their way to casting 47% overall, according to analyses of voter rolls conducted by Republican political consultant Derek Ryan. Voters with no recent voting history – primary or general – cast nearly 16% of early ballots, on their way to casting 18% of all ballots. While not up as dramatically as 2018 compared to 2014, 32% more people cast ballots early in 2016 than in 2012, the most recent presidential election year.
Importantly, a record number of straight-party votes – more than 5.6M – were cast in 2016, up 13% from 2012 and nearly double the number cast in 2014. Statewide, Republicans cast 545K more straight-party votes than Democrats, which was below 2014 (647K), 2012 (629K) and 2010 (602K), but high enough to prove insurmountable for the statewide Democratic candidates.
Nearly 1M more votes were cast in 2016 than in 2012. Donald Trump received just 115K more votes than Mitt Romney, so most of these new voters broke toward Democratic nominee Hillary Clinton (570K) or minor party and independent candidates (290K). If we assume the existing 8M voters stayed in their partisan lanes and voted as they did in 2012, then 34% of “new” voters cast straight Democratic ballots, 26% voted straight Republican, 5% voted straight Libertarian or Green, and the remaining third voted the full ballot. Of those latter individuals, more voted for Libertarian, Green or independent candidates than for Hillary Clinton.
This finding is important. Despite a swath of voters’ reluctance to vote for Trump, a plurality did not want to vote for the Democrat, either, which is indicative of the damage sustained by the Democratic brand.
Comparing 2016 to 2014, 4.25M more people voted. Of that, 1.3M more people voted straight Republican, 1.4M voted straight Democratic and around 75K more voted straight Libertarian or Green. Again, two-thirds of those who did not vote in 2014 cast straight-party ballots with a slight edge going to the Democrats, but not enough to overcome the built-in Republican advantage from 2014 voters.
All of this is a backdrop to evaluating what to make of the huge surge in turnout from 2014.
Keep in mind that early voting turnout is up not only in the biggest counties but also in many of the smaller ones. In 2016, Clinton received 186K more votes than Trump in counties where at least 50K votes were cast. Trump received 993K more votes than Clinton in counties where fewer than 50K votes were cast, including 276K more votes than Clinton in counties where between 5K and 10K people voted. Those voters more than covered Clinton’s margin in the big counties, before accounting for Trump’s margin in counties with between 20K and 50K voters (+429K) and counties with fewer than 10K voters (+288K).
People who voted in the 2016 general election but not in 2014 or in the primaries likely comprise the vast majority of this surge, and we know how that election turned out. Trump defeated Clinton by 9 points – a relatively close race by recent Texas standards – and the rest of the Republican statewide ticket won by an average of 15 points.
Absent significant changes in heart on the part of Trump’s voters, one would expect them to stay in their partisan lanes and continue to cast a high volume of straight-party votes. Republicans have had a statewide advantage in straight-party voting since 1996.
Looking for a Blue Coalition
O’Rourke, and every other statewide Democratic nominee, is depending on some combination of dissatisfied Republicans staying home, suburban Republican women crossing the aisle, angry Democrats showing up who normally vote only in presidential elections, and new voters breaking hard for Democrats. All of that is within the realm of possibility. Let’s look at each in turn.
Are dissatisfied Republicans staying home? According to Ryan’s research, most Republican primary voters have already voted. More than 85% of people who have voted in the last four Republican primaries have already voted, as have more than 77% of people who voted in three of the last four Republican primaries (and no Democratic ones). Nearly two-thirds of voters with any primary history have already voted, and the Republicans (64%) are almost even with the Democrats (65%). So, it looks like the partisans are voting, so any dissatisfied Republicans would have to be weak Republicans who vote less frequently and never in primaries.
Are suburban women having a change of heart? Recent polls have shown a significant gender gap, but women voters historically have gone into general elections fairly evenly divided. According to the October 2014 Univ. of Texas/Texas Tribune poll, women narrowly favored Abbott, 48%-46%, while men favored Abbott, 61%-32%. In the October 2016 UT/TT poll, women were split almost evenly between Trump and Clinton, 44%-44%, while men favored Trump, 45%-39%. In the October 2018 UT/TT poll, women narrowly favor O’Rourke, 49%-47%, while men favor Cruz, 56%-42%. So, like in 2014 and 2016, women do not appear to be breaking strongly for O’Rourke this year. However, according to the pollsters, Cruz is ahead among suburban voters, 55%-43%. Assuming suburban men favor Cruz in proportion to all men, then suburban women would be fairly evenly divided. The same reasoning, applied to the 2014 poll, suggests the same theory. A Democrat being evenly divided among suburban women is better than losing, but it’s not winning.
Are angry Democrats who only vote in presidential elections showing up? According to Ryan’s research, 45% of early voters with no recent primary history voted in 2010 and/or 2014, which means 55% of them didn’t. Based off that data, we calculate that about 12% of all early voters cast ballots in 2012 and/or 2016 but not in a gubernatorial general election AND not in a recent primary election. What are the chances those 12% of voters are heavily Democratic? Well, recent history suggests that Democratic turnout drops more in a gubernatorial election than Republican turnout. Compared to 2008, the average Democratic statewide candidate received 48% fewer votes in 2010, while the average Republican statewide candidate received 28% fewer votes. Four years later, the average Democratic statewide candidate received 46% fewer votes and the average Republican statewide candidate received 37% fewer voters than in 2012. It would require a hard break from recent history for this 12% of voters to be mostly angry Democrats.
Are new voters breaking hard for Democrats? They tell pollsters they are. The October UT/TT poll showed “likely voters” under age 30 favoring O’Rourke, 71%-23%, and an October CBS News/YouGov poll showed the same cohort favoring O’Rourke, 57%-35%. According to Ryan, about one out of every nine early votes was cast by someone under age 30, which is about double the rate from 2014. However, it’s slightly less than in 2016, which saw 13% of all votes, including Election Day, cast by voters under age 30 (In the October 2016 UT/TT poll, “likely voters” under 30 favored Clinton, 50%-26%.). In 2016, 12% of early voters were under age 30, so that cohort represented almost the same proportion of Election Day voters. What remains to be seen this year is whether younger voters will defy history and be a more significant component of Election Day votes.
Implied in the suburban woman question but with similar implications across other Republican voters is the question of whether a sufficient number of voters will split their tickets. Every recent poll, regardless of the credibility of its sample or soundness of its methodology, has shown Cruz with a smaller lead than any other Republican statewide candidate polled:
- Quinnipiac (Oct.): Cruz +5, Abbott +14
- UT/TT (Oct.): Cruz +6, Abbott +19, Dan Patrick +18, Ken Paxton +12
- CNN (Oct.): Cruz +7, Abbott +14
- Quinnipiac (Oct.): Cruz +9, Abbott +20
- Emerson Coll. (Oct.): Cruz +5, Abbott +20
- Ipsos/Reuters (Sept.): Cruz +2, Abbott +9
- Dixie Strategies (Sept.): Cruz +4, Abbott +19
- NBC/Marist Univ. (Aug.): Cruz +4, Abbott +19
These numbers indicate somewhere between 3% and 8% of voters intend to vote for Abbott and O’Rourke. We have previously mused that Cruz is demonstrably less popular than Abbott, which could account for part of ticket-splitting behavior. We also know from polling data that former Dallas Co. Sheriff Lupe Valdez (D) is not as well-known as O’Rourke, who has raised and spent tens of millions more than Valdez.
We previously went into great detail about what a hypothetical 20-point Abbott win would mean for O’Rourke. Only once since 2002 has a Democratic statewide candidate received 10% more of the vote than the lowest performing statewide candidate, and that was the governor’s race in 2006, which featured two strong independent candidates. In an era of increasing use of straight-party voting, the ability of a single Democratic candidate to break from the rest is severely challenged.
Regardless of turnout, the larger the margin Abbott has over Valdez, the harder it will be for O’Rourke to close any gap on Cruz. If Abbott wins by 20 points, O’Rourke would need roughly four out of every nine people who voted for Abbott without doing so by the straight-party option, or an equivalent combination of those people and straight-party voters who broke ranks and voted for O’Rourke.
So, after all of this, what is the state of the state?
As we discussed a few days ago, voters generally are more optimistic entering this year’s election than recent general elections, gubernatorial and presidential. For nine years, the Univ. of Texas/Texas Tribune polls have consistently shown Texans’ net assessment has been and continues to be that the Nation is on the wrong track.
Notice the significant upswing since October 2016. In the October 2018 poll, that net rating was minus-7, the second highest it has been in the poll’s history, just a point below its all-time high of minus-6 in June. Heading into recent general elections, that net rating has been minus-39 (2010), minus-27 (2012), minus-40 (2014) and minus-46 (2016). Relatedly, 32% of the October 2018 poll respondents believed the “national economy is a lot better off” than a year ago compared to just 7% who believe it to be “a lot worse off,” the highest such result in the poll’s history. Heading into recent general elections, the “a lot better off”/“a lot worse off” rating has been 3/33 (2010), 7/23 (2012), 6/14 (2014) and 8/15 (2016).
As one might expect, partisans’ assessment of the nation’s direction depended on who was in office, and they flipped immediately after the 2016 general election. Independents, on the other hand, more closely track to the average assessment.
Forty percent of independents, mostly those who lean Republican, believe the country is on the right track, and half, mostly those who lean Democratic, believe it’s on the wrong track. That minus-10 net rating is the second best in the poll’s history, trailing only June’s minus-5. Looking at presidential approval ratings, independents with strong opinions rate Trump at 37/39 (minus-2). Heading into gubernatorial elections, their strong views of Obama were 14/59 in 2010 and 13/51 in 2014.
Let’s look once more at the October UT/TT and October CBS News/YouGov polls. Those polls show Cruz leading, 51%-45% and 50%-44%, respectively. The UT/TT poll shows Abbott leading, 56%-37% (The CBS/YouGov poll did not test the governor’s race.).
Both polls’ samples had identical age group splits:
- 17% under age 30
- 25% aged 30 to 44
- 36% aged 45 to 64, and
- 22% aged 65 and older.
Both polls showed O’Rourke leading by a lot among voters under 30 and Cruz leading by a lot among voters 65 and older.
Thanks to Ryan’s analyses, we have a good idea of how the early vote split among those age groups. Based on early turnout, voters under 45 and over-represented in the polls’ samples, and overs 45 and older are under-represented. Based on our interpolation of Ryan’s most recent analysis of early turnout, early voters’ ages broke down as follows:
- 5% are under the age of 30, 5.5% fewer than the polls’ sample
- 5% are aged between 30 and 44, 4.5% fewer than the sample
- 5% are aged between 45 and 64, 2.5% more than the sample; and
- 5% are aged 65 and older, 7.5% more than the sample.
If we readjust the poll results for the observed distribution of ages, Cruz’s lead increases to 53%-44% and 51%-43%, respectively. Abbott’s lead increases to 58%-36%.
Given Abbott’s relative popularity and voters’ generally optimistic views of the state and given Valdez’s significant resource disadvantage in comparison to former Sen. Wendy Davis (D-Fort Worth), it is difficult to imagine Valdez doing much better than Davis absent a large shift in straight-party voting toward the Democrats or some systematic polling error(s) that make(s) Abbott’s lead appear larger than it is.
One potential source for such an error is Hispanic/Latino turnout. Ryan’s analysis indicates 19% of early voters had Hispanic surnames, a rough proxy for Hispanic/Latino turnout. This is four points higher than in 2014 but no higher than in 2016, a year during which Hispanic/Latino voters comprised roughly one out of every four “new” voters. Just looking at early voting turnout, measured as percent of registered voters casting ballots, Hispanic/Latino-majority counties are among the lowest ranking of the 30 counties with the most registered voters.
Republicans have won the last 123 statewide races, 129 if presidential races are included, since 1994. Since 2002, Democratic statewide candidates fared best, measured head-to-head against Republicans, in 2008. That year, a significant group of new voters, inspired by the chance to elect a young, charismatic African-American to the White House, lifted Democrats all over the state.
But not the statewide candidates. They still all lost by an average of 8.6% of the vote. All of the attention went to the presidential race, and the rest of the largely anonymous statewide candidates – The highest vote share went to Sam Houston. – were largely lifted by the tide of straight-party voting. We estimate 400K more straight-party Democratic votes were cast in 2008 than in 2004 while the number of straight-party Republican votes remained flat. It was enough to flip four Republican-held legislative seats, but Republicans flipped a couple of Democrat-held seats in the process.
We believe there may be enough energy from the Democratic base, in reaction to President Trump’s first two years in office, and among young voters excited about O’Rourke’s candidacy to deliver a similar statewide average to 2008. Mathematically, that allows for Cruz to win by 5 percentage points and Abbott to win by 17 points. The margins of the remaining statewide races would fall between 8 points and 14 points, most within 10-12.
Accordingly, we project the average statewide Democratic candidate to receive 44.4% head-to-head against the average statewide Republican candidate’s 55.6%. For races with a Libertarian candidate, this translates into 54.3% Republican, 43.3% Democrat and 2.3% Libertarian, on average.
The 44.4% head-to-head average is nearly 2% higher than 2016 (42.5%) and a 6% improvement over 2014 (38.4%). It would be the highest head-to-head average performance since 1994 (47.8%), the last year that a Democrat won a statewide race.
Our district-by-district models begin with this estimate, which is why we have taken so long, and gone into such detail, to arrive at it.
©2018 Texas Election Source LLC