co/efficient poll has New Hampshire Senate tied, and our forecast moves a point

A co/efficient survey of 958 likely voters in New Hampshire, fielded September 9-11, found the Senate race even: Chris Pappas at 46 percent and the Republican line at 46 percent.

Our forecast did not follow the poll. The model now gives Pappas, the Democrat, an 84 in 100 chance of winning the open New Hampshire seat, against 16 percent for Republican Scott Brown. That is up from 83 percent in our previous run, early on September 15. The mean margin is unchanged at D+10.4, and the 5th-to-95th percentile range still spans R+7.2 to D+28.0. In other words, a single tied poll left the distribution where it was.

That gap between one survey and our estimate is worth stating plainly. The model draws on more than the latest number, and a D+10.4 mean with a range that reaches into Republican territory already allows for results like this one. A tied poll of likely voters is consistent with a race our model treats as favoring the Democrat without treating it as settled.

Nationally, our average of 23 generic-ballot polls over the past 45 days stands at D+6.9. In the Senate, we give Democrats a 49 percent chance of winning control and Republicans 50 percent, with a mean of 50.2 Democratic seats. Democrats need 51 seats for control; Republicans need 50, because the vice president is a Republican and a 50-50 tie goes their way. In the House, we give Democrats a 96 percent chance of a majority, with a mean of 241.3 seats against the 218 needed. Our House simulation correlates error nationally, by census division, by state and by district, but it has no term linking demographically similar districts in different places, so that 96 is marginally firmer than a fuller error model would leave it.