Lone GA-11 poll shows a close race; our model still sees a wide Republican edge

One new poll of Georgia's 11th District has Democrat Chris Harden at 42 percent and Republican John Cowan at 40 percent. That is a Miller Consulting Services survey of 402 likely voters, fielded August 31 to September 11. With a sample that size, a two-point gap is not a lead; the poll describes a close race, and about 18 percent of respondents are not accounted for by either candidate.

Our forecast did not follow it. The model now puts the mean margin at R+24.5, with a 5th-to-95th percentile range running from R+39.3 to R+9.7. Before this run, at 2:32 PM ET, the mean margin was R+24.6 and the range was R+39.4 to R+9.8. Cowan's win probability stands at 100 percent, rounded, and Harden's chance is under 1 in 100. A single small district poll that sits roughly 26 points from our average does not move a district estimate that rests mostly on other information; it moved this one by a tenth of a point.

That gap is the story here, and it is worth stating plainly: either the poll is an outlier, or our estimate of GA-11 is badly wrong, and one survey of 402 voters is not enough to tell us which. Nothing else in the payload for this district speaks to the question.

Nationally, our average of 29 generic ballot polls over the past 45 days stands at D+7.5. Our model gives Democrats a 97 percent chance of winning House control, with a mean of 243.2 seats against the 218 needed. One caveat on that figure is ours to own: our simulation correlates error nationally, by census division, by state and within each 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.