(CNN) - Texas politics are in the spotlight again this week. A number of House Republicans from the state have announced their retirements, and, after a mass shooting in El Paso this past weekend, Beto O'Rourke tussled with President Donald Trump over differing responses to the shootings.
In the backdrop of these developments is a Texas electorate that seems to be divided much more than it used to be.
It's a state with a voter base that seems to rapidly be shifting toward the center -- and I believe it could go Democratic in next year's presidential election for the first time since 1976.
I was once a skeptic on Texas turning blue, but I've changed my tune because Trump is a uniquely unpopular Republican in Texas who seems to be the driver of an important development: Like other Americans, Texans with a college degree are shifting rapidly from red to blue, and Democrats have a lot of room to grow with them in Texas.
This doesn't mean the state will go Democratic in 2020 (Democrats not named Joe Biden currently are losing to Trump in the Lone Star state in high quality polling), but it is a real possibility.
Trump's net approval rating (approval - disapproval) among registered voters has been slightly negative in Texas throughout his presidency. The latest Quinnipiac University poll put it at -1 points. All other elected Texas Republican officials had at least a +8 point net approval rating. This poll comes on top of the 2018 exit poll giving Trump a +1 net approval rating, and the midterm electorate in Texas is likely more Republican leaning than a 2020 presidential electorate will be. Trump won the state by 9 in 2016.
If the 2020 election were held today and it were solely a referendum on Trump, Texas would be a toss-up.
Traditionally, Texas has been a lot more Republican than the nation as a whole. In 2014, for example, Democratic President Barack Obama's net approval rating was 18 points worse in Texas than nationally. In 2018, Trump, a Republican, was just 10 points higher in Texas. In other words, there was an 8-point shift toward the Democrats, on this measure, compared to the nation as a whole in just four years. This followed the 2016 presidential race being the closest in the state since the 1990s.
Trump's unusually low approval rating in 2018 created the environment in which Republican Sen. Ted Cruz won re-election by less than 3 points. It was the worst Republican performance in a Senate race in the state since 1988. In 2012, Cruz won his first term by 16 points. This 13 point pro-Democratic shift occurred even though Cruz was an incumbent (who usually do better) and the national environment (as measured by the presidential vote in 2012 and House vote in 2018) shifted by less than 5 points toward the Democrats. (As I've pointed out, this shift was driven mostly by opinions toward Trump, not towards the Texas Senate candidates running in 2018.)
Of course, just because the state became more blue doesn't mean it will keep moving blue into 2020. The reason I think it could shift more is because of why the state has been going blue over the last few years: the aforementioned highly-educated voters.
High-quality consistent public polling data of Texas isn't easy to find. And while county-by-county returns are not a perfect way of examining the issue, they are a powerful tool and are sending a very strong signal. Take a look at what occurred between the 2012 and 2018 Texas Senate races.
Traditionally, the thought was growing diversity (i.e. Hispanics becoming a bigger part of the electorate) would largely be responsible for turning Texas blue. That, though, could take years for Democrats and could be counteracted by whites becoming more Republican.
By examining the 254 counties in Texas, we see that Cruz's losses were not strongly correlated with the percentage Latinos made up of the population or Latino population growth. This doesn't mean Latinos didn't move left with the state; they almost certainly did. Rather, it indicates they probably weren't the big reason the state saw more movement left than the nation over the last few years. (Limited high quality non-partisan public polling suggests the same.)
Instead, Cruz lost a disproportionate amount of ground in places with a high percentage of college educated residents. Weighted for the number of voters, Cruz's percentage of the major party vote share dropped by 10 points in the 20 counties where at least 30% of adults 25 years and older have a college degree. These are counties that contain major cities and suburbs like Dallas, Harris (Houston), Tarrant (Fort Worth) and Travis (Austin). Cruz's vote share dropped by less than 2 points in the state's other 234 counties.
We see the same pattern in the Texas congressional districts (7th and 32nd) that flipped from Republicans to Democrats in 2018. Latinos make up a much lower share of the citizen voting age population compared to the state in these districts, while college graduates make up a larger share than average.
Importantly, the aforementioned well-educated cities and suburbs seem to be continuing to shift blue. When weighted for the number of voters, Cruz did 4 points worse in 2018 than Trump in 2016 in the same 20 counties where at least 30% of adults have a college degree. That is, it wasn't just that Trump became president and there was a one-time shift in these places. The counties are becoming more Democratic even under Trump. The state's other 234 counties also shifted toward the Democrats from 2016 to 2018, but by less than half as much as the 20 most well-educated counties.
If you look at these 20 most educated counties, they have a lot more room to move left. Cruz only lost these counties by about a little more than 10 points. This matches with exit poll data indicating that college graduates in Texas are still considerably more GOP-leaning than nationally. Given Trump's ability to instigate college educated voters (as seen in the fact that the college-educated counties in Texas keep moving left), it doesn't take a leap of logic for these voters to turn bluer. Add on any serious "get out the vote" operation to enroll Latinos (who still punch well below their weight in Texas) to vote, and things could get even dicier for Republicans.
Meanwhile, there is less room for the 234 counties with less than 30% of the adult population with a college degree to move right. They are currently about twice as Republican-leaning as the well-educated counties are Democratic leaning. This lines up with exit polls showing non-college voters in Texas are right of where they are nationally.
These county patterns remind me a lot of what happened between 2012 and 2014 in Iowa, but in reverse. As I noted at the time, the counties with a lower percentage of college-degreed adults jumped significantly to the right in only two years time. This foreshadowed the state taking another move to the right in 2016.
Again, this doesn't mean these counties and, therefore the state, will move left. But we have seen big movement across numerous elections before. Beyond my Iowa example, a number of smart folks on Twitter have noted that the Northeast underwent a major shift to the left between 1988 and 1992 and a second time from 1992 to 1996.
Put another way: The story of 2016 was non-college educated voters in the north voted more like college-educated voters in the south than previously. The story of 2020 could be college-educated voters in Texas voting more like college-educated voters in the north than they used to.