South Carolina Senate probability halves as our model's range narrows

Our model now gives Democrat Annie Andrews a 9 in 100 chance in the South Carolina Senate race, down from 18% in our run of August 24. That is a drop of 8.2 points of Democratic win probability over seven days, all of it toward Republican incumbent Darline Graham, who is now at 91%.

The projected margin moved the other way. Our mean margin is R+8.4, against R+11.0 a week ago. What changed more is the width of the distribution: our 5th-to-95th percentile range ran from R+30.5 to D+7.9 in the earlier run and now runs from R+18.8 to D+2.1. A tighter range with the center still on the Republican side leaves fewer simulations in which the Democrat finishes ahead, which is the arithmetic behind a lower probability alongside a smaller mean lead.

Two surveys with field dates inside that window came from Abacus Data, both fielded August 26-28. Its likely-voter sample of 306 had the Republican at 51% and Andrews at 38%. Its registered-voter sample of 500 had both at 34%. An Impact Research likely-voter poll of 700, fielded August 18-24 and posted by the Andrews campaign, had the race at 41-41 with 5% for another candidate. Our earlier note on the Abacus likely-voter numbers is superseded on the probability, if not on the margin.

We are not attributing the full week's move to those polls. The model reweights the whole polling record and the national environment each run, and the narrowing of the range is doing most of the work here. Our national Senate numbers give Democrats a 49% chance of a majority and Republicans 50%, with Democrats needing 51 seats and Republicans 50, since a 50-50 tie goes to the Republicans under a Republican vice president.