Here's the spreadsheet of variables, as promised...
To post this comment click here:
Otherwise click cancel.
PopDensity should be obvious, that's the population density, calculated using 2000 census data (I know it's old, but the 2010 one isn't out yet. I wish we had a Census each 5 years, it would make things for statisticians a lot easier).
'American' is the % of people who identify their ancestry as 'American or United States'. You encounter most of these people in Appalachia, there they're up to 30%- Conservadems and Reagan Democrats. The main reason why Obama lost there so badly in both primary and general election, but they were also significant factors in the Kerry loss.
It's for a reason that the two counties where this percentage is highest (Garfield and Petroleum) are among the most Republican.
Median Household Income (source: Census 2000) is also fairly self-explanatory, as are the variables Evangelical, Catholic, Collegegrads and w/o Health insurance- the latter two are made of Census data, the former two of data by the Association of Religion Data Archives (ARDA).
Population 2000 is also self-explanatory, as is Natives. I don't think population will actually be of much use in the regression, as its effect should be better captured by population density, but it doesn't hurt to have it- we don't have to include it if it's statistically insignificant.
Clinton primary is the two-party share that Hillary Clinton received in the 2008 primary.
Finally, some data that I estimated in a fairly long process in 2009- those estimates don't come from an external data sources, but were calculated by me, so you're sort of the first outside of the Tyler Gernant campaign to view them.
Liberalness measures the... well, Liberalness of a county (or, actually, a standard General-election sized electorate of a county) on a so-called Likert scale, where every Liberal gets a 0, every Moderate a 5 and every Conservative a 10, then dividing by population. In a County with a score of 10, there would be only Conservatives, in a County with a score of 0, there are only Liberals.
To give you a better qualitative handle on that, 4.5 would be roughly like Hawaii or New York State, 5.5 is like California, 6.5 like Missouri or Florida, and 7.5 like Texas or Wyoming. Note that counties will always tend to be a bit more extreme because they're not as diverse as whole states, and also note that that's just estimates, there never has been actual data on it.
PartyID measures again, estimated by me the difference between the percentage of self-identifying Democrats and Republicans. That isn't necessarily the same as party registration- quoting the write-up of a University poll prior to the MA special election between Coakley and Brown:
[It's] important to point out that a high percentage of the registered Independents in MA (they're actually called unenrolled) are really either Democrats (36%) or Republicans (34%) when you look at PARTY ID.
So there. Note that if you're stumbling over the low value for Missoula (it's just at D+9), that's because Missoula's liberalness is driven more by extremely liberal Independents (single-payer supporters, people who think that the two parties are alike, REAL socialists...) than by actual Democrats- unlike for example Butte, where the Liberals are actually all Democrats as well. I'm estimating that there are about 39% Independents in Missoula (again, identifiers, not registered), and that 68% of them voted for Obama.
Independents is the estimated percentage of self-identifying Independents in each county.
And finally, a bit randomly labeled (the variable originally was used a little bit differently and in another context), BaseDem is the distance to Helena, a bit amateurishly measured with an online ruler and an online map of MT. I don't know what this variable is going to do, maybe some East-West divide could constitute itself here. It might have made more sense to use Missoula, but I'm not going to go over the boring procedure of collecting that data once again. I actually used that variable back in 2009 to try to estimate how much effect a base has in elections- and the answer is, a lot within the same county (you get about 10 bonus points support for being a favorite son, and I guess that'd be even more in primaries, I tested for General Elections), but not much in terms of 'County A is just 200 miles away and more likely to support me than County B that's 250 miles away).
