A three-point poll in NM-2 trims our margin but not the odds

The one new survey in New Mexico's 2nd District shows a close race. A SurveyUSA poll for KOB-TV in Albuquerque, of 554 likely voters fielded August 28 to September 2, has Democratic incumbent Gabe Vasquez at 46 percent and Republican Greg Cunningham at 43 percent — three points apart, with the rest unaccounted for in the numbers we have.

That is well short of what our model expects on Election Day, and it moved our numbers a little. Our forecast now gives Vasquez a 94 in 100 chance, down from 95 percent in the previous run earlier today, and puts the mean margin at D+13.8, narrowed from D+15.1. The range widened at the low end: our 5th-to-95th percentile band now runs from R+0.8 to D+28.5, where before it ran from D+0.4 to D+29.7. In other words, the model now treats a narrow Republican win as inside its plausible range, if only just.

The practical effect is small. A single poll of 554 voters, taken more than three weeks before today, does not overturn a district-level estimate built on national conditions as well as local surveys. Our generic-ballot average stands at D+8.3 across 30 polls over the last 45 days, and NM-2 remains a race our model expects the Democrat to win comfortably more often than not, while conceding that the polling suggests a closer contest than the mean margin implies.

Nationally, our model gives Democrats a 98 percent chance of winning a House majority, with a mean of 245.7 seats against the 218 needed for control. One caution on that figure is ours to own: our simulation correlates error nationally, by census division, by state and by district, but it has no term linking demographically similar districts in different places. Without that, the House control probability is marginally firmer than it should be.