Five Wisconsin Elections: A Pattern of County Unit/Ward Vote Share Anomalies
Dec. 23, 2012
Updated: Aug.2, 2015
The purpose of this analysis is to determine if there were repetitive patterns in the cumulative county vote shares in five recent Wisconsin elections. The patterns are obvious; the county graphs are virtual duplicates.
This post is a work-in-process, but since the data tables and graphs are completed, I wanted to make them available while the analysis is ongoing.
The following counties appear most anomalous: Brown, Dane, Jefferson, Kenosha, La Croix, Milwaukee, Oneida, Ozaukee, Racine, Richland, Sheboygan, Trempealeau, Walworth, Washington, Waukesha and Winnebago.
Republican vote shares are increasing (lines slope upward) while Democratic shares decrease (slope downward) at the same rate. This is an indicator of likely vote switching.
Summary of Key Walker Recall Results
Walker won the recall by 171,000 votes (53.1-46.8%).
In 15 large counties, Barrett’s vote shares at 25%, 50% and 100% of the cumulative total were 54.2%, 52.1% and 48.1%, respectively. The counties had 1.51 million of the total 2.52 million recorded votes.
Milwaukee County is the largest and most anomalous. In the recall, Barrett had 63.3% of the total 396,000 votes. But he had 74.4% at the 25% mark, 70.4% at 50% and 66.5% at 75%. Looking at Barrett’s shares in terms of remaining votes, he had 59.4% of the final 75%, 55.9% of the final 50% and 53.0% of the final 25%. In other words there was a 21.4% decline in Barrett’s 74.4% vote share of the first 100,000 votes to 53.0% in the final 100,000 votes.
Barrett’s True Vote Model 54.4% share is within 0.2% of his 15 county cumulative share at the 25% mark. His total Wisconsin share (assuming an equal level of fraud in the other 57 counties) was 52.4%.
In the 15 counties, there was a 6.0% difference between Barrett’s 54.2% at the 25% mark and his final 48.1%. Adding 6.1% to Barrett’s official 46.3% total share, he had an estimated 52.4% Wisconsin True Vote share.
In the 15 counties, there was a 4.0% difference between Barrett’s 52.1% at the 50% mark and his final 48.1%. Adding 4.0% to Barrett’s official 46.3% total share, he had an estimated 50.3% Wisconsin True Vote share.
2008 Presidential Election
The cumulative vote analysis essentially confirmed the unadjusted exit poll. Obama won the WI recorded vote by 56.2-42.7%. He won the unadjusted exit poll 63.3-35.7%, a 7.1% increase over the recorded vote share.
In 15 of the largest counties, Obama’s vote shares at the 25%, 50% and 100% of the cumulative total were 62.4%, 60.6% and 57.1%, respectively. The counties had 1.85 million (62%) of the 2.98 million total recorded votes.
Democratic votes shares declined by an average of 6.0% from the 25% mark to the final recorded vote:
15 Wisconsin Counties
Democratic Vote Share Trend
15 Wisconsin Counties
(Votes in thousands)
.................... Percent of total vote
15 Counties Votes 25% 50% 100% Change
2008 Obama 1853 62.38% 60.59% 57.07% 5.31%
2010 Feingold 1375 54.70% 52.38% 48.69% 6.02%
2010 Barrett 1372 55.04% 51.86% 48.23% 6.81%
2012 Barrett 1551 54.24% 52.11% 48.14% 6.10%
2014 Burke 1511 53.96% 52.22% 48.50% 5.46%
The following spreadsheets use data provided by GAB. Note that Milwaukee County is displayed at the top of the screen in each spreadsheet to illustrate the similar cumulative vote pattern in each of the four elections.
2012 Walker recall (contains voting machine types for each county and municipality).
In the process of working on analysis of Wisconsin elections, I have developed a number of models and databases which are available online as Google Doc spreadsheets. They can be linked to from the following posts: