I did a post a while back on the power of field offices to affect elections. Specifically, I showed that the presence of an Obama field office was associated with a modest but detectable bump in the Democratic vote share in the 2008 general presidential election. (The results were published here.) I also have a chapter coming out (hopefully) that shows the power of field offices to affect delegate selection during a party nomination campaign. I'm now finding some evidence that such field offices have a persistent effect in later contests.
I've been looking at the results of the March 2010 Democratic caucuses in Colorado's U.S. Senate race. Using a multi-variate analysis of the caucus-night vote, I find that Michael Bennet did, on average, twenty-one percentage points better than Andrew Romanoff in counties where Obama had a pre-caucus campaign office two years ago. This is statistically significant (p ≤ .05), even after controlling for virtually every major demographic characteristic of the counties.
I don't fully understand the mechanism here. It may be that the early Obama office counties just built strong organizations, which translated into strong OFA groups, which are active in the current Romanoff/Bennet contest. Or it may be that the early Obama offices developed more lasting connections with active Democrats in those counties, so that the Obama endorsement of Bennet this year was more influential in those counties.
Another hypothesis might be that there is some peculiar facet of the Obama field office counties that causes them to prefer certain candidates in Democratic nominations contests but has nothing, in fact, to do with any effort exerted by those offices. Obama put his offices in those counties, but the offices didn't change the vote -- they would have preferred him, and Bennet, anyway.
One way to test this hypothesis is to look at a Democratic statewide primary contest that occurred prior to the establishment of these field offices. Probably the best one is the 2004 Senate contest between Mike Miles and Ken Salazar -- this was the last time there was a major Democratic nomination contest for statewide office in Colorado. In fact, there was no relationship between that primary vote and the location of Obama's field offices. Now, this isn't an overwhelming piece of evidence for my case -- it was six years ago, and people do move in and out of counties. But, in theory, if there was something peculiar about the Obama field office counties in '08, we might have seen some evidence of it in '04, and we don't. This is at least suggestive that the offices Obama built in '08 did cause the vote boost for him and are affecting the current Democratic Senate nomination contest.