Only age is currently available, none of the other data is yet. We have controlled for every variable we can (age included), and are obviously aware there are effectively limitless demographics to control for. Feel free to pitch in and collect some data or run some analyses rather than pointing out the obvious.
It would be foolish for a would-be election hacker to change the votes by more than 90% variance - you want to change it only enough to win. Thus the magnitude of the effect is irrelevant - the p value, however, is (which as you note is below.05, but as you fail to note, is also below.001).
To address your concern about correlated predictors I've run a two-step regression: one with all the covariates to predict Hillary's votes, and a second to predict the unstandardized residuals from vote method. The diebold effect on clinton is still sig (p.001) as is the effect on Obama (p.001).
RTFA. We controlled for % holding bachelor's degrees, median household income, and population density - that's why this is newsworthy. The diebold effect is still significant.
Slashdot Mirror
Only age is currently available, none of the other data is yet. We have controlled for every variable we can (age included), and are obviously aware there are effectively limitless demographics to control for. Feel free to pitch in and collect some data or run some analyses rather than pointing out the obvious.
It would be foolish for a would-be election hacker to change the votes by more than 90% variance - you want to change it only enough to win. Thus the magnitude of the effect is irrelevant - the p value, however, is (which as you note is below .05, but as you fail to note, is also below .001).
To address your concern about correlated predictors I've run a two-step regression: one with all the covariates to predict Hillary's votes, and a second to predict the unstandardized residuals from vote method. The diebold effect on clinton is still sig (p.001) as is the effect on Obama (p.001).
RTFA. We controlled for % holding bachelor's degrees, median household income, and population density - that's why this is newsworthy. The diebold effect is still significant.