(click to increase size)
An updated and more refined version of this graph.
Left to right, the ratio of visits on Hardcore for Nerds received from each state (August-November 2008) to its share of US population (adjusted for broadband penetration). And on the vertical axis, the positive or negative vote-margin in each state for Obama-Biden (Delaware, Hawaii and Illinois have been re-adjusted to 2004 levels). The trend (regression) line has an r-squared value of just under 0.55, which is an expression of how well the date points conform to a linear trend (answer: moderately well).
For example: Pennsylvania (PA) had 1.54 times the number of visits that would be expected from its population size (against the number of readers from the US overall) and broadband penetration, and went for Obama by a margin of 10.37%. It's also quite close to the trend line, meaning it's not an outlier...
Alaska, what's up?
D.C., rock top!
Mississippi, you need more punk...
Addendum: same graph with the readership expressed as visits per 100,000 population (adjusted for broadband penetration) here.
10 comments:
I'm from Alaska but I read via Washington state. That's got to count for something... maybe
yes and no... Alaska has considerably more visits than expected (even adjusting for its high broadband penetration), Washington somewhat less. I guess spiritually Alaskan readers are just a boon.
I really love that one, you should get a PhD or something for that..
Not sure, but could it be Alaska is a bit off because being homestate of Mrs. Palin resulted in some sort of deviant voting behaviour?
ha! thanks, but I'm some way away from that yet.
I adjusted (not in this graph) the Republican home states to 2004 as well, but it didn't make a significant difference (In fact Alaska only went slightly up, and Arizona went down).
For whatever reasons, Alaska and South Carolina are the only red states with above-average readership.
So I guess it would be interesting to cross-check the visits of conservativepunk.com. It must be shooting through the roof in Alaska...
I'm not good with stats, so I won't pretend to understand. But I'm a somewhat regular Alaskan reader--directed here first via the prisionship--so perhaps I skewed your numbers?
would you need more than one reader to see this impact?
kevin
btw: altho' she (theoretically) lives just down the street from me, i won't take credit for Ms. Palin (or the voting behavior of most of the rest of this blasted state for that matter).
kevin - that's quite likely. there were about 30-40 visits between August and November '08, so if you've been coming here that long...
obviously, I'm not working with huge sample sizes here, especially for the smaller states. I'm pretty sure the guy from Geek Down is skewing my results for Connecticut, too.
Looks like DC is really leveraging your R squared here.
that's true... it's 0.39 without it. but the slope of the trendline hardly changes (which was what I understood the main issue of leveraging is), in fact it shifts up a bit.
also, Maine is as far off the trendline as DC is. taking it away ups the r-squared very slightly, basically no change.
Post a Comment