Obama Surges in "T-Shirt Primary" - Obama 47%, Clinton 18%

Folks, I've mentioned a few times in diaries how I'd become intrigued by watching T-shirt sales on CafePress.com as an indicator of how "hot" or "not " a candidate was. (full details why below the fold).

And I mentioned how, CafePress has set up a CafePress "Election Meter where you can track sales and product creations by candidate, over time. Well, here's a fascinating nugget from this past week:

After being nearly tied two weeks ago in weekly candidate sales (Obama at 28% of all sales, Hillary at 26%), Obama surged last week to 48%, while Hillary fell to 19%.

This is the biggest single change we've seen since we started geekily monitoring these numbers (my company has been doing some work for CafePress, which is how we first came across this data).

Is it scientific? Hell no. But... seems to me it's a hell of an interesting tea leaf to read when gauging grassroots enthusiasm or trying to figure if a candidate is waxing or waning.

So here's the question -- might there be something more here than just a fun coincidence? Sure, it stands to reason that more popular candidates will sell more gear than less popular ones. I mean, that's a duh.

But here's a theory I will pose -- sales of gear will both rise and drop in advance of rises and drops in a candidates poll ratings. Why? A few reasons:

1. The hardcore political types who pay close attention are the ones most likely to buy the gear. They are also the ones most likely to be the "mavens" in their communities -- the ones who will ultimately sway friends, family, etc., to support their candidates. So if you see them start to buy shirts, gear, etc., the odds are they are also evangalizing to others. The first part --- buying gear -- takes place right away. But the wooing of others takes a bit more time, both for direct wooing, and the ripples out.
   
2. On the flip side, I would argue that if a candidate's support starts to weaken, people will stop taking strong actions (buying gear) before they switch preference. If I'm strongly supportive of a candidate, I'll want the world to know. But if I'm supportive but questioning, I might hedge my bets. Sure, I'll still tell the pollster I'm supportive, since I haven't changed my mind yet, but I'll start to be less vocal and overt about it.
   
3. Getting really, really, really granular... most polls take at least a day or two to get results after the data is collected, meaning they are snapshots of a day or two ago, not today. Sales, on the other hand, happen in real time. So you could actually see trends a day or two ahead (Again, this is only for the hyper-geeks, but clearly in the case of a debate, for example, it'd be interesting to see the immediate impact). Again, just some thoughts here from a data/political junkie. And I will try to get more info to validate this as the weeks and months go on.

But... I'd be curious if people think there's a "there there" to this T-shirt primary theory