I am in the midst of explaining how the FiveThirtyEight forecast model evaluates state polls along with national polls in an effort to determine where the overall race stands. I suppose I think that the national polls sometimes receive a bit too much attention.
But we, at FiveThirtyEight, are not dogmatic about this. The model does use national polls as well — including the various tracking polls that are released on a daily or weekly basis, and sometimes they can have a discernible influence on the forecast.
Mitt Romney has seen some improvement in these tracking polls over the last week. The Rasmussen Reports tracking poll had him four points ahead of Barack Obama as of Thursday. Galllup’s tracking poll showed him moving back into a one-point lead after Mr. Obama had held the advantage for most of the past few weeks.
Both the Rasmussen Reports and Gallup polls have been slightly Republican-leaning relative to the consensus of surveys. But that is not true of two polling firms that release weekly tracking polls, Public Policy Polling and YouGov, and both of those polls showed Mr. Romney gaining as well.
These tracking polls are subject to the same random noise that any other surveys are — and sometimes the movement they appear to show can be spurious. But the one nice thing about them is that they do provide a steady basis for comparison. Our hope, of course, is that the forecast model can weigh their pluses in minuses in a sensible way.
Add it up, and Mr. Romney has gained some ground in the FiveThirtyEight forecast over the past few days. His chances of winning the Electoral College were listed at 35.0 percent by the model as of Wednesday night’s forecast, his best figure since the June 26 release of our numbers.