Yesterday I mentioned that to statistically identify the impact of vice-presidents on presidential races we'd need data points where the same vice presidential candidate ran with different presidential candidates, which was wrong. Actually, we'd need examples of presidential candidates running with different vice presidential candidates.
Of which we have some examples even in the modern era; Roosevelt and Nixon come immediately to mind. But this is not a lot of data, and you couldn't exactly include a lot of other variables in the model.
My supposition is that (1) at an aggregate level the vice presidential impact on the ticket is minor once you've taken out the effects of demographic and economic conditions , and (2) the impact at the level of the individual voter is low once you've accounted for partisan affiliation, opinions on the issues, and the quality of the presidential candidates (except perhaps in their home states).
Posted by robe0419 at June 25, 2004 7:36 AM