Substack cometh, and lo it is good. (Pricing)

"Americans" don't like Barack Hussein Obama

I post some data analysis over at my other weblog. For example, today I looked at the relationship between food stamp usage and unemployment. The Census makes a lot of county-level data available, though it’s often slapdash and disorganized. But using R I’ve constructed many data sets including most American counties. I don’t post here much because I concentrate more on science in this space, and the 500 pixel width means that integrating scatter plots into a post seamlessly is pretty much impossible.
But since readers of this weblog are much more liberal than over at GNXP Classic, I thought you’d be interested in what demographic variables predict voting for Barack Obama on the county-level. So the dependent variable is the 2008 results for Barack Obama by county.
The independent variables are:
% of non-Hispanic whites who identify German
% of non-Hispanic whites who identify as “American”
% black
% Median household income
% Median home value
% with a college degree
% on food stamps
% obese
Below are the coefficients with errors and p-values if not significant).


% of non-Hispanic whites who identify German 0.13 (0.032)
% of non-Hispanic whites who identify as “American” -0.89 (0.046)
% black 0.20 (0.022)
% Median household income ~ 0
% Median home value ~ 0
% with a college degree 0.33 (.037)
% on food stamps 0.008 (0.0006)
% obese -0.0012 (0.001, p-value 0.23)
r-squared = 0.52
I also calculated earlier the white vote for Obama using a combination of county-level data and state-level exit polls. So here are the coefficients for prediction of the white vote for Obama:
% of non-Hispanic whites who identify German 0.27 (0.026)
% of non-Hispanic whites who identify as “American” -0.68 (0.045)
% black -0.25 (0.023)
% Median household income ~ 0
% Median home value ~ 0
% with a college degree -0.089 (.036)
% of whites on food stamps 0.008 (0.0006)
% obese -0.0037 (0.0092)
r-squared = 0.56
“Americans” are generally those of British or Irish ancestry who live in the South. The correlation between counties in their voting for Obama and % with a college degree = 0.38, and -0.42 for % who are “American” in ancestry. The sort of model above tries to control for other variables which are correlated together. When you control the other variables college degree % becomes less powerful as an independent effect in relation to the % who are “American.” Median house value (at least in 2005-2007) was correlated with the Obama vote at 0.33, but its effect goes to nothing when you control for other factors.
Note: The r-squared represents the proportion of the variance in the dependent variable which can be explained by the independents.

Posted in Uncategorized

Comments are closed.