The week before an election is the most stressful of the year for political junkies. The time for messaging and debate has largely passed, and the focus switches to voter turnout. Traffic to forecasting sites like FiveThirtyEight and RealClearPolitics shoots up as partisans try to get a sense of what might happen when the votes are tallied. Observers dig into last-minute polling and attempt to parse the turnout data from early voters to predict who will win key races. 

Polling in Texas—and, to some extent, nationally—is bouncing around much more than usual. The makeup of the electorate is harder to predict than ever: Does the absence of Donald Trump from the ballot make some percentage of his supporters less likely to turn out? Will the Supreme Court’s June decision to allow states to restrict or outlaw abortion motivate a pool of untapped voters or persuade moderate Republicans to switch parties? Assumptions about turnout are at the heart of pollsters’ predictive models, and even without the complicating factors, our confidence in election forecasters to accurately predict results has taken a big hit since the days when the internet pondered whether Nate Silver was a witch. 

In Texas, nobody has lost money over the past three decades by betting on Republicans to win statewide elections. But even as most polls suggest a GOP rout, there are questions heading into the final week whose answers will only become clear once the votes are tallied. 

Will GOP efforts in South Texas pay off?

One of the biggest stories to come out of the 2020 election was the massive improvement for Republicans in the traditionally solid blue stronghold of South Texas. Since then, the party has invested in building a bench of GOP candidates running up and down the ballot in the region. In addition to the surprise results in 2020, the party also notched a win over the summer, when Republican Mayra Flores claimed a special election in Texas’s Thirty-fourth Congressional District. That result wasn’t necessarily predictive; the special election was an unusual occurrence owed to an early retirement, and Flores now faces a reelection campaign against a popular longtime incumbent Democrat, Vicente Gonzalez, who moved over from a neighboring district. But the data points are strong enough to bring a rush of GOP interest—and money, and voter-turnout foot soldiers—to the area. 

A trio of well-funded Latina candidates are running in three South Texas congressional districts, all of whom (Flores included) are likely to be rising stars in the party should they win their races. At the same time, lower-profile races for positions such as constable, county judge, and justice of the peace, some of which haven’t seen a GOP candidate in decades, are being contested. The organization Project Red TX has recruited and supported Republican candidates in the down-ballot races throughout South Texas and the Rio Grande Valley. In a story about the organization, the New York Times ran the headline “Democrats Should Be Freaking Out a Little About South Texas.” 

There’s virtually no recent public polling specific to South Texas, so it’s hard to know exactly how much Democrats should be panicking. There have only been public polls in two of the three contested congressional districts, both showing close races. It’s possible that having competitive local races in South Texas could be motivating for both Republicans and Democrats in the region, all of whom previously had little reason to turn out to vote in races that were foregone conclusions. But any new motivation Democrats feel will be pushing up against a well-funded ground game for South Texas Republicans that puts the region’s offices up for grabs for the first time in most voters’ lifetimes.  

Does Anyone Know How to Poll Latino Texans?

As South Texas has become a politically competitive part of the state, the long-standing assumption that the state’s booming Latino population would lead to a bluer Texas has been challenged. What if Latino voters aren’t a monolithic voting bloc that will reliably cast ballots for Democrats? 

If you’ve been looking for evidence that the 2020 results in South Texas augur a wholesale change in the political makeup of the state’s fastest-growing demographic group, you can find it: just look at this poll from the University of Texas that has the gubernatorial race as a toss-up among Hispanic voters, with both Greg Abbott and Beto O’Rourke getting support from 48 percent of Latinos statewide. Alternatively, if you want evidence that Latino voters are still solidly a Democratic constituency, you can find that, too: here’s a poll from Univision in which Latino voters say they prefer O’Rourke to Abbott by a thirty-point margin. And what if you feel like the large swing from UT to Univision is unrealistic, but you still think Latino voters in Texas probably lean Democratic? Here’s a Siena College poll that has O’Rourke up by nine points with Latino voters. 

Polls of Texas Latino voters are like the weather—if you don’t like today’s, wait till tomorrow, and it’ll be completely different. None of these polls are outliers, either; a UT-Tyler poll found a 23-point preference for O’Rourke among Latino voters, while an Emerson College poll found O’Rourke up by only 7 points. Polling discrepancies this large are unusual, and the pollsters can’t all be right, which means that most of them have missed big (they could even all be wrong!). Experts who are attempting to predict who will win the governor’s race, then, have some wildly inaccurate data in their forecasting models—but in which direction?

The answer to that question is significant. If Latino voters are roughly split between Abbott and O’Rourke, then Abbott will win the election handily. If those same voters prefer O’Rourke by thirty points, it’s anybody’s race. The same likely holds true for the rest of the Democratic statewide ticket. 

What Will Abortion Mean for Democrats? 

The challenges of using polls to forecast this election were compounded when one of the long-standing issues in American politics was upended by the Supreme Court’s decision in Dobbs v. Jackson Women’s Health Organization, which overturned abortion rights established in the court’s 1973 Roe v. Wade decision. The politics of abortion, which had for fifty years motivated evangelical voters to vote for Republican candidates, appear to have flipped. Polling since the decision suggests that nationally, at least, those who support abortion rights are now likely to be single-issue voters for the Democrats, a finding that polling of the issue in Texas seems to support as well. 

Pollsters can attempt to capture this enthusiasm in their research, but polls struggle with seismic realignments in the political landscape (see the defection of white working-class voters, many of whom had supported Barack Obama but voted for Trump in 2016). Polls try to model what the electorate will look like and then ask questions of the voters pollsters expect will turn out. The model, broadly speaking, is based on who voted in previous elections—which can pose problems if something has happened to change the political landscape in the interim period. 

Accordingly, even pollsters are uneasy about what they expect will happen in this election. In a story last week about the anxieties pollsters are experiencing, the New York Times quoted Democratic-campaign pollster Anna Greenberg, who said, “Nobody knows who’s going to vote. In fact, right now, every poll could be wrong, because there are some people . . . who think that we’re going to have higher turnout than 2018. No model is prepared for that. Most of the models are prepared for higher Republican turnout.” In a rare display of bipartisan agreement, GOP-campaign pollster Jon McHenry echoed Greenberg’s assessment: “You don’t know exactly who you’re missing. And that’s always been the kind of thing that will keep a pollster up at night.”

It is entirely possible that the Dobbs effect, if it exists, will exert a minimal force on the election. Texas has added half a million new voters to the rolls since the March primaries, which is a large number, but also one consistent with the growth in voter registration in prior election years. That’s a data point that suggests that abortion rights had little to no impact on getting new voters to register. But it doesn’t mean that there won’t be newer or younger Texans who are less-reliable voters who are drawn to the polls this month in order to cast ballots opposing abortion restrictions—and those are exactly the kinds of voters that polls tend to miss the first time they show up. 

Does Early-Voting Data Mean Anything?

One potential set of tea leaves we can attempt to read are the turnout numbers we’ve seen from the first week (and change) of early voting. Raw numbers of early votes are down from 2018; in the thirty largest counties in Texas, about 2.3 million ballots had been cast after seven days of early voting, compared to almost 3 million in 2018. Which voters are staying home? Trying to figure that out, as Republican political consultant Derek Ryan does in his (excellent) newsletter tracking the early vote, is fraught with additional uncertainty: We know that younger voters have made up a smaller percentage of the electorate so far than they did in 2018. We know that rural voters make up a slightly larger percentage of the electorate up to this point than they did in 2018. We know that voters who have a history of voting in Republican primaries make up 45 percent of the electorate so far, while Democratic primary voters account for 32 percent (the rest have no primary history). 

But those things aren’t necessarily predictive. Younger voters might turn out on election day, while rural voters may have been motivated to turn out early and could end up outpaced by urban and suburban voters as voting draws to a close. Plus, folks who previously voted in one party’s primary may vote for the other party’s candidates in this election. 

The election will be a consequential one, and it’s understandable why so many of us would like an early glimpse of what the result might be. Accordingly, pundits, predictors, pollsters, and partisans are doing their damnedest to project the outcome on November 8. But if they really want to know who’ll claim victory, their best bet will be to do it the old-fashioned way and stay up late on election night.