Ted writes:

> :-).  Another moral might be to pick the most dramatic way to present your
>  point.
>
>  Let's take a brief look at the actual percentages which Buchanan received
in
>  various counties.  I've included the first 10, alphabetically, plus Palm
>  Beach and a couple of others I calculated during my exploration.
>
>  Alachua:         262 of  85,559 = 0.306%
>  Baker:            73 of   8,151 = 0.895%
>  Bay:             248 of  58,770 = 0.422%
>  Bradford:         65 of   8,666 = 0.750%
>  Brevard:         570 of 218,297 = 0.261%
>  Broward:         789 of 573,099 = 0.138%
>  Calhoun:          90 of   5,170 = 1.741%
>  Charlotte:       182 of  66,864 = 0.272%
>  Citrus:          270 of  57,121 = 0.473%
>  Clay:            186 of  57,340 = 0.324%
>  Ockaloosa:       267 of  70,654 = 0.378%
>  Palm Beach:     3407 of 432,695 = 0.787%
>  Volusia:         396 of 118,280 = 0.335%
>  Walton:          120 of  18,295 = 0.656%

Just as a note, Ted's absolutely right. The graph that CNN put up implicitly
presumes that all of the bins (in this case, Florida counties) are
equiprobable. Clearly, that isn't true because the populations of the
different Florida counties varies so dramatically, thus the data must to be
normalized in some manner in order to present a fair and accurate picture.
The CNN is just one way out of thousand to "lie with figures." Ted's method
of correcting the data is about as good as any. Division by county total
population would be another, but if the ratio of voter turnout was rather
uniform in all counties, then Ted's results above would be essentially
identical for the second normalization, too.

The only other condition that has to be added to a true pictorial
representation are confidence bars. The smaller the population of voters, the
less the calculated ratio can be trusted (the greater the uncertainty). One
way around that might be to aggregate adjoining counties that have very
similar demographics (rural, urban, suburban, etc.), although that too is
always arguable. Nonetheless, it's this kind of analysis that tends to
suppress the outliers that always tend to pop up in the smaller bins, such as
Baker and Calhoun county data points above.

In fact, analyzing the data from a number of different perspectives is the
only way to get a true sense of its robustness, and how much faith you can
put in your analyses.

Wirt Atmar