Iowa Senate moves toward Turek as our OH-7 estimate pulls back

Our model gives Democrats a 93 percent chance of winning a House majority, with a mean of 240 seats against the 218 needed for control. The Senate is close to even: a 49 in 100 chance for Democrats and 50 percent for Republicans, on a mean of 50.2 Democratic seats. Democrats need 51 seats for control and Republicans need 50, because the vice president is a Republican and a 50-50 tie goes to that side.

Our generic ballot average stands at D+6.5, drawn from 26 polls fielded in the last 45 days. Five of the most recent are in that window. Morning Consult, surveying 28,429 registered voters August 24-30, had Democrats at 47.1 percent and Republicans at 42.3. Quantus Insights, with 1,200 likely voters August 26-28, had it 49.7 to 42.8. A YouGov poll for The Economist fielded August 28-31 had Democrats up 46-40 among 1,434 registered voters and 36-31 among 1,589 adults. Ipsos, polling for Reuters over the same dates, had 43-38 among 924 registered voters and 38-31 among 1,167 adults, with 3 percent for another party.

The largest single-day change on our board was in OH-7, where the Democratic win probability fell 22.5 points, from 54 percent to 31. Our Iowa Senate estimate moved the other way, up 17.1 points to a 39 percent Democratic chance from 22. Three Iowa polls landed in the past week. Wedgewood Polls, with 600 likely voters August 27-29, had Ashley Hinson at 50 percent and Josh Turek at 48, a close result. Abacus Data, fielding August 26-28, had Turek ahead 48-41 among 323 likely voters and 41-36 among 500 registered voters. Abacus also polled Ohio over those dates, finding Sherrod Brown at 40 percent and Jon Husted at 33 among 500 registered voters.

Our New Hampshire Senate estimate fell 13.9 points, to an 81 percent Democratic chance from 95. Kansas moved up 8.9 points to 15 percent. Michigan is at 69 percent after an 11.5-point rise; EPIC-MRA, polling 600 likely voters August 22-28, had Abdul El-Sayed at 47 percent and Mike Rogers at 43. New Mexico sits at 99 percent, and Research & Polling Inc., surveying 516 likely voters for the Albuquerque Journal August 21-28, had Ben Ray Luján at 53 percent and Larry Marker at 38.

One note on the House figure. Our simulation draws error at four correlated levels: nationally, by census division, by state, and within each district. It does not include a term linking demographically similar districts in different places, because we have no cluster data for one. That missing correlation means our 93 percent is somewhat more confident than a fuller treatment of the same evidence would produce. The size of the gap is not measured in this run.