New MI-4 poll shows a close race, and our forecast does not move

A GBAO survey of 500 likely voters in Michigan's 4th District, fielded September 13-16, put Democrat Sean McCann at 47 percent and Republican incumbent Bill Huizenga at 46 percent. A one-point separation in a sample that size is not a lead in any meaningful sense. The poll describes a close race and little more.

Our forecast reads it the same way. The model gives McCann a 46 in 100 chance of winning the seat and Huizenga 54 percent, with a mean margin of R+0.8. The 5th-to-95th-percentile range runs from R+15.5 to D+13.8, which is another way of saying the district could plausibly land well on either side of even. The run before this one, at 12:32 PM ET today, had the same win probabilities and the same mean margin, with a range of R+15.4 to D+13.8. The new number confirmed what was already in the model rather than changing it.

The national backdrop is more lopsided than this district. Our generic-ballot average stands at D+8.3 across 30 polls over the past 45 days. In the House, where 218 seats are needed for control, our model gives Democrats a 98 percent chance of winning a majority and a mean of 245.8 seats. MI-4 is one of the seats that our simulations split closest to evenly, and the outcome here is not what decides the chamber.

One caveat on that House figure is ours to own. Our simulation correlates error nationally, by census division, by state and within each district, but it has no term tying together demographically similar districts in different places. We lack the cluster data to build one. The effect is that our House control probability is marginally firmer than a fuller error model would leave it. We have not measured how much, so we are not putting a number on it.