A week or two ago, I thought the early voting data in Texas was less than favorable for Biden. This was based upon TargetSmart partisan preference modeling that showed the GOP with an early advantage (51-43).
In states that don’t register by party, like Texas, that outfit models party preference based upon demographics and consumer data.
In Wisconsin, there is significant disagreement between polls of early voters and TargetSmart projections. For example, one poll by the NYT shows early voters in WI 82-18 for Biden, while the TargetSmart projection is 45D-37R-18I.
With that kind of divergence, one of them is simply wrong. There is no need to try to reconcile the two. There is no possible reality in which 50% of Republicans vote for Joe Biden.
Meanwhile, Texas leads the nation in voting so far at 7.2 million, with several suburban counties already exceeding their total 2016 turnout with five days of early voting left. District-level polls of those same suburban counties show an absolute massacre for Donald Trump, with Biden running 12-14 points ahead of Clinton. TargetSmart data suggests this is totally wrong, that Trump is winning the early vote, and will only extend his lead on Election Day. There are a couple of national-level polls that follow the latter narrative, and include things like 25-30% of the black vote to Trump.
So, if applied to Texas, either Biden is going to do the unthinkable and turn Texas blue on a wave of enormous turnout and newly-blue suburbs, or Trump not only has held onto his base, but has won over a couple million new acolytes and will win by double digits.
There’s not really an argument to be had. These are two incongruous sets of facts and data. One of these is reality, and the other is a partisan wishcasting delusion.
I guess we’ll find out which is which next week.