New Quinnipiac polling in four states contains pretty good news for John McCain.

In Colorado, Quinnipiac has McCain ahead by 2 points, 46-44. This is the only Colorado poll in which McCain has led all year, save for an oddball results from the GOP-affiliated firm TargetPoint Consulting back in early April. Obama had led by 5 points in Quinnipiac's prior poll of Colorado, taken at the height of Obama's post-primary bounce last month.

Obama maintains his lead in the other three states in this box set, but it is smaller than before in each instance. In Michigan, Obama now leads by 4 points after having been 6 points ahead in June. In Wisconsin, his lead is down from 13 points to a still-healthy 11-point margin. But in Minnesota, the tightening is far more substantial, with Obama's lead going from 17 points to just 2.

Rasmussen also has numbers out today from another swing state, New Hampshire, where Obama holds a 4-point lead -- broadly in line with the recent UNH and ARG surveys -- after having led by 11 in June.

I hope that there is no longer any question that this is more than just statistical noise. Yes, there are individual results we can critique. It's hard to imagine Obama running 9 points stronger in Wisconsin than he does in Minnesota, for instance. And Quinnipiac's results from Colorado are a little odd, as Obama leads among independent voters and does as well as McCain does amongst his party, but trails slightly overall (Quinnipiac does not weight its results by party ID). Our model is designed to account for this noise in a variety of different ways, and for the moment, it doesn't take the possibility of a McCain win in Minnesota seriously, and still regards Obama as a very narrow favorite in Colorado.

But our model is also designed to evaluate trends, and there is an increasingly large body of evidence that Obama is now polling somewhere between 3-4 points off his peak numbers. In the grand scheme of things, that doesn't mean all that much -- it means that perhaps 1 in every 60 strangers you encounter on the street has switched from Obama to McCain within the last month. The more relevant question is where the downtrend dates from. If you look at our tracking graph, it seems to have started -- or at least steepened -- coming out of the July 4 holiday, when some of the Obama is a flip-flopper narrative began to take root. I am less convinced that Obama is getting an anti-bounce out of his trip abroad, and would remind you that there is a lagged effect before certain stories take hold, particularly in the dog days of the summer when the public's attention span for campaign coverage is limited.

The alternate hypothesis is that this is simply a reflection of McCain's greater investments in advertising in the early campaign, something we'll explore at greater length soon.

--Nate Silver