New Michigan poll shows a one-point race and leaves our forecast unchanged

A Fabrizio Ward/Impact Research survey of 877 likely voters in Michigan, fielded August 9-11, put Democrat Abdul El-Sayed at 48 percent and Republican Mike Rogers at 47 percent. A one-point separation in a sample that size is not a lead. The race is close.

Our forecast did not move. The model gives El-Sayed a 58 in 100 chance of winning the open Michigan seat and Rogers 42 percent, the same figures as our previous run at 10:30 a.m. ET on August 20. The mean margin is unchanged at D+1.3. The only difference is at the edge of the distribution: our 5th-to-95th percentile range now runs from R+9.6 to D+12.4, against R+9.9 to D+12.4 before. That is a rounding-scale change, not a new reading of the race.

The new number therefore confirms rather than alters. Michigan is the sort of contest our model treats as genuinely unresolved 75 days out, with outcomes on either side of a nine-point Republican win and a twelve-point Democratic one inside the plausible range.

Nationally, our Senate picture is just as tight. We give Democrats a 49 percent chance of winning Senate control and Republicans 51 percent, with a mean of 50.2 Democratic seats. Democrats need 51 seats for control; Republicans need 50, because the vice president is a Republican and a 50-50 tie goes to them. Our generic-ballot average stands at D+6.7 across 25 polls in a 45-day window.

In the House, our model gives Democrats a 94 percent chance of winning a majority, with a mean of 242.4 seats against the 218 needed. One caveat is ours to own: our simulation draws correlated error nationally, by census division, by state and by district, but we have no cluster data to correlate demographically similar districts across state lines. Without that term, the House control number is marginally firmer than it should be.