A close SC-1 poll nudges our forecast, and little else

A Public Policy Polling survey of 536 registered voters in South Carolina's 1st District, sponsored by The Bench and fielded September 9-10, put Republican Jenny Costa Honeycutt at 41 percent and Democrat Nancy Lacore at 40 percent. That is a close race by any reading, and a large share of respondents named neither candidate.

Our model now gives Lacore a 23 in 100 chance of winning the district, with a mean margin of R+6.9 and a 5th-to-95th percentile range running from R+22.1 to D+8.2. Before this run, at 8:32 AM ET, those figures were 22 percent, R+7.1, and R+22.3 to D+8.1. The new poll moved the numbers, but barely: a single point of win probability and two-tenths of a point of margin. It confirmed what was already in the average rather than redirecting it.

The gap between a near-tied poll and a forecast of a seven-point Republican margin is the model doing what it does. A 536-person sample of registered voters carries real sampling error, and the district's prior partisan lean and the national environment both pull the estimate away from the single survey. Our national generic-ballot average stands at D+7.2 across 25 polls over the past 45 days.

Nationally, our model gives Democrats a 96 percent chance of winning a House majority, with a mean of 242.3 seats against the 218 needed. One caveat on that figure is ours to own: our simulation draws correlated error at the national, census-division, state and district levels, but it has no term linking demographically similar districts in different places, because we have no cluster data for it. That omission makes the House control probability marginally firmer than a fuller decomposition would leave it. In the Senate, our model gives Democrats a 51 percent chance of control, with a mean of 50.3 seats against the 51 they need; a 50-50 tie goes to the Republicans, since the vice president is a Republican.