> >CurtAdams@aol.com wrote:
> >
> >> Impressive that he could get this out so fast. I note in
> >> http://elections.fas.harvard.edu/node3.html
> >> that this isn't the only really outrageous Buchanan outlier, just
> the...
> >
> >spike: They calculated the predictor of expected
> >Buchanan votes by the number of Nader votes!...
>
> No; Buchanan and Nader voters are two very distinct groups
> of people and I would expect a strong negative covariance.
> Buchanan's supporters are mostly conservative and rural
> while Nader's are liberal and urban. The Palm Beach outlier
> graph I've seen several times shows that Buchanan's vote
> percentage decreases with the size of the county, presumably
> due to the rural effect.
I acknowledge the perception and I disagree that this is a
legitimate means of comparing candidates. I calculate thus:
if a county were to have some *special reason* to have
strong negative feelings towards one of the non-mainstream
candidates, then the other non-mainstream candidates, to
whom the county had no special antipathy, would appear
to be anomalously favored. Palm Beach had a special reason
to have heartburn with Nader: his green stance doesnt play
well in those parts.
The URL gives the formulas used to determine the 15 sigma
result, and it basically looks at the ratio of non-mainstream
candidate votes. This approach does not take into account the
anti-Nader fervor in the county, which I would expect to be
very strong, with so large a retired population. I expect that
generation will not look favorably on a candidate that
wishes to spend buttloads of money to create habitats for
mosquitos.
Note that I am not arguing that some voters mistakenly
cast the wrong vote, I imagine a few did that. But looking
at the Nader votes is the wrong way to determine it. spike
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:50:21 MDT