I remain officially Not Worried™. Too many other polls lately have Cruz at 50 percent or better to make me believe that Beto! can actually pull this off.

But I confess to being a little less Not Worried™ than I was last week.

The new data comes from Change Research, a left-leaning polling outfit that was formed last year. If you want to scoff at their numbers, there are reasons to do so. A last-minute poll showing O’Rourke neck and neck with Cruz is conveniently powerful turnout motivation for Democrats, the target audience of Change Research. And Nate Silver’s site gives them just a C+ rating in accuracy to date.

Cruz is doing quite well among pure Republicans, leading 95-4 among 2016 Trump voters. But Beto has consolidated support among everyone else: over 99% of Clinton 2016 voters, over 90% among Jill Stein voters, and double digit leads with 2016 non-voters and Gary Johnson voters.

— Change Research (@ChangePolls) November 3, 2018

Texas is a state that Trump won 52-43. But, in a high turnout election, this will be a very close race if Beto can win virtually all Clinton voters, a few Trump voters, and the vast majority of non-voters and Stein/Johnson voters.

— Change Research (@ChangePolls) November 3, 2018

There are reasons to take them seriously, though. A sample of 1,211 likely voters is larger than the samples of most polls, which should in theory produce a more accurate result. Change Research also had a notable success in polling just within the last few months, becoming the only pollster to predict Andrew Gillum’s surprise win in the Democratic gubernatorial poll in Florida. And there’s other evidence that the Cruz/O’Rourke race has tightened somewhat lately.

Those are all of the polls since October 21. Cruz had a run earlier last month, post-Kavanaugh, where he was leading fairly steadily by 6-9 points. But as you can see, more than one survey lately had his lead under four before the new one from Change Research showing the race tied. FiveThirtyEight now gives O’Rourke a better than 20 percent chance of pulling the upset. I expect Cruz will hold on, but a 52/47 win in Texas will be a shockingly tight result against a Democrat who made no bones about his liberalism.

Since I’ve become obsessed with the Arizona Senate race, let me toss a new poll of that state into this post as well. Not great:

Add in the leaners and it’s Sinema 49.8, McSally 47.1. That’s from Trafalgar, the same pollster that nailed Trump’s shocking wins in the Rust Belt on the eve of the 2016 election. And it’s based on a big-big-big sample of 2,166 Arizonans. The good news for Republicans is that the Harris/Rasmussen poll continues to have McSally ahead by a safe-ish 5.6 points. The bad news for Republicans is that that’s the only poll lately that has her up: One recent survey had the race tied and four others had Sinema winning by anywhere from 3-6 points. And don’t forget that the Green Party candidate quit the race in the last few days and endorsed the Democrat. That may be worth only one percent, but in this race one percent may be decisive.