Ed noticed a few days ago that Team Red’s disadvantage on the generic ballot had dipped to five points after swelling to more than eight in July. Just a blip or part of a trend towards improvement for the GOP?

Four days later, it’s looking better for the “trend” theory. Today the Democratic lead has sagged to 3.9 points, less than half of what it was a month ago.

It’s been a bouncy summer. After a brutal winter, Republicans made it all the way back to within 3.2 points of the Democrats at the start of June, creating real hope for holding the House this fall. Within six weeks the Democratic lead was eight points again. Four weeks after that it’s back down to four. What gives? It’s tempting to say “good economy” but the economy’s been solid for many months. That doesn’t explain the bounciness. It’s always tempting to say “Trump!” but Trump’s numbers are even more stable than the economy’s. Take a peek at his job approval. Apart from a brief blip in June, he’s been steady at 43 percent since May.

One possibility is that the GOP’s benefiting from the August recess. The middle of this month is among the slowest cycles for politics all year and feels even slower in the Trump era as you-know-who gets less camera time. Ed pointed out in his post that the dwindling Democratic advantage on the generic ballot is less a matter of Republicans climbing than Democrats dropping. Only once in the past year has the GOP touched as high as 40.8 percent on the GB — and that was for just one day towards the end of May. Ignore that and their high since early 2017 is just 40.5 percent. It’s the recent plunge by Team Blue that accounts for most of the tightening gap, as you can see on the graph above. Theory, then: There’s a group of anti-Trump independents that turns more resolute in wanting to support the Dems the more they’re exposed to POTUS in their daily media diet. When Trump goes on vacation, their annoyance at him subsides and thus so does the Democratic lead.

There’s another theory circulating, not necessarily inconsistent with mine but with real data to back it up. This one comes from a journalist at The Economist. Look back at previous midterm cycles, notes Elliott Morris, and you’ll find that the president’s party usually gains a bit of ground at this point in the election cycle:

See the upswing in the top line of the first graph at around 100 days out from the midterms? We’re seeing it again in Republican numbers this year. It may be, in other words, that public disgruntlement about the ruling party “naturally” softens a bit during the August recess, when Congress is out of town and most Americans are enjoying the summer rather than paying attention to politics. If that’s true then you’d expect antipathy to the ruling party to harden again in the fall, once Congress is back and candidates are campaigning in earnest — and that is indeed what you see, per Morris’s graph. In fact, ominously for the GOP, Morris’s historical data suggests that the ruling party bottoms out right before Election Day, reaching its lowest point of the cycle as late deciders break against it. I doubt that’ll happen this year considering how low the dregs of Republican polling got last winter, when Trump’s job approval was significantly worse. But it suggests that the final pre-election margin for Dems will be closer to eight points than it will be to four.

Here’s Pelosi encouraging all of the Democratic candidates who are falsely claiming that they’ll oppose her as Speaker to just say whatever they need to say to win.