A close CO-8 poll leaves our forecast where it was

The only new survey in Colorado's 8th District shows a close race. McLaughlin & Associates, polling 400 likely voters on September 16-17 for Enhancing American Competitiveness, put Republican incumbent Gabe Evans at 48.0 percent and Democrat Manny Rutinel at 46.3 percent. A gap of 1.7 points on a sample that size is not a lead; it is a tie within the survey's expected error.

Our forecast did not move. Before the poll was added, our model gave Rutinel an 87 percent chance and a mean margin of D+9.9, with a 5th-to-95th percentile range of R+4.8 to D+24.6. After, the numbers are an 87 in 100 chance for Rutinel, the same mean margin of D+9.9, and a range of R+4.7 to D+24.6. That is the kind of change that comes from simulation noise rather than from new information.

The reason a single close poll does not move much is that the model is weighing it against everything else, including a national environment our average puts at D+8.1 on the generic ballot across 30 polls over the last 45 days. It is worth being plain that the McLaughlin result sits toward the Republican end of our distribution for this seat: our interval still allows for an Evans win of several points, and 13 percent of our simulations produce one. A sponsored poll of 400 respondents fielded over two days is a thin evidentiary base either way.

Nationally, our model gives Democrats a 98 in 100 chance of winning House control, with a mean of 245.3 seats against the 218 needed. One caveat on that figure belongs to us: 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 wherever they sit. That omission makes our House number marginally firmer than it should be.