http://www.opednews.com/articles/Whats-wrong-with-Waukesha-by-Eric-Nelson-110419-812.html?show=votes#allcomments(snip)
With all of these anomalies and calls for investigations and considering the initial closeness of the race between Prosser and Kloppenburg we decided to look at some of the county by county election totals. We went under the hypothesis that any major attempt at election fraud by either party might shift variables that presumably would be mostly random into something that might be less random.
We compared the vote totals per ward to the vote spread between the two candidates in counties that were strongly partisan. Because some of these counties had significant differences in vote totals we combined multiple counties in some cases and normalized the data so that the standard deviation was between 60-75% of the median vote for each county.
We noticed a very interesting difference, of about 15 different counties that we looked at that were either strongly Democratic or strongly Republican, only Waukesha county and to a lesser extent Washington and Ozaukee showed positive correlations between vote totals and vote spread (see figure 1 below). What this means is that in Waukesha as the number of votes that were tabulated in a particular ward became larger the magnitude of Prosser's victory tended to become larger.
When we looked at counties where Kloppenburg won by a large margin such as Dane, Ashland and Bayfield counties there was no correlation between the vote total in a particular ward and the vote spread between Kloppenburg and Prosser. This was also true for Douglas and Eau Claire counties.
(end snip)