Looking at recent polling trends and their relationship to political news, I had a thought that I explained in some detail at New York:

For a while there in January and early February, American politics revolved around phenomena other than the volcanic personality of our president: the effect of the December 2017 GOP tax cuts; positive economic news; the fate of Dreamers; an appropriations battle in which Trump was mostly on the sidelines. Coincidentally or not, both the president and his party had some of their sunniest polling numbers in a long time last month. Trump’s approval rating moved north of 40 percent and stayed there in most assessments, and the double-digit Democratic advantages in the generic congressional ballot that grew common in December were more than halved overall. One highly publicized poll from Morning Consult in early February had Republicans actually moving ahead.

But as Eric Levitz recently observed, the Democratic “polling panic” has now subsided, with the generic ballot numbers moving back toward the robust advantages for the Donkey Party we saw in December. And now it’s becoming clear Trump’s own approval ratings are following the same pattern: In FiveThirtyEight’s averages (which adjust the numbers for established partisan “leans” and also give slightly greater weight to more accurate pollsters), he’s back below 40 percent, where he spent most of 2017.

It’s possible these trends simply reflect a reversion to the mean after a short, atypical moment. But it may be less than coincidental that the end of that moment occurred during a period when Trump was very large if not necessarily in charge.

During the last three weeks, political news has been dominated by fresh evidence of turmoil in the White House (punctuated by the Porter scandal that represented a combo platter of incompetence and insensitivity about domestic violence), new developments in the Mueller investigation, and erratic Trump behavior on Twitter and elsewhere. A particularly dangerous juncture for Trump may have been the Presidents’ Day weekend when he went on an extended Twitter rampage, mostly about the FBI and the Mueller investigation, even as media focused on the Parkland massacre. The jury’s still out on the effect of the president’s personal involvement in the post-massacre debate on gun control, though his steadily increasing investment in the loopy idea of arming teachers doesn’t bode well for him.

If this theory is right, or even half right, we should expect some more short-term deterioration in the president’s approval ratings and the GOP’s standing in the generic ballot. More importantly, it underscores a persistent dilemma for the president’s team. Without question, Trump being Trump is important to the maintenance of maximum excitement within his electoral base, and that is an asset of great importance in relatively low-turnout midterm elections. But if Republicans need the simmering anti-liberal resentments of the MAGA crowd to remain at a near-boiling-point as November approaches, the presidential behavior that most reliably keeps the heat on also appears to repel voters who might be otherwise persuaded to stick with Trump’s party on policy grounds.

How to balance base mobilization and swing-voter persuasion is a perennial puzzle for any political party. But it’s especially complicated when the base glories in the very characteristics of a leader that actually frighten others. If Republicans become convinced that revving up the base is the only thing they care about in this midterm election cycle, they won’t have to do a whole lot to encourage Trump to go absolutely wild for weeks on end.