BERKELEY, CA -- November 18 -- Today the University of California's Berkeley Quantitative Methods Research Team released a statistical study - the sole method available to monitor the accuracy of e- voting - reporting irregularities associated with electronic voting machines may have awarded 130,000-260,000 or more excess votes to President George W. Bush in Florida in the 2004 presidential election. The study shows an unexplained discrepancy between votes for President Bush in counties where electronic voting machines were used versus counties using traditional voting methods - what the team says can be deemed a "smoke alarm." Discrepancies this large or larger rarely arise by chance - the probability is less than 0.1 percent. The research team formally disclosed results of the study at a press conference today at the UC Berkeley Survey Research Center, where they called on Florida voting officials to investigate.
The three counties where the voting anomalies were most prevalent were also the most heavily Democratic: Broward, Palm Beach and Miami-Dade, respectively. Statistical patterns in counties that did not have e-touch voting machines predict a 28,000 vote decrease in President Bush's support in Broward County; machines tallied an increase of 51,000 votes - a net gain of 81,000 for the incumbent. President Bush should have lost 8,900 votes in Palm Beach County, but instead gained 41,000 - a difference of 49,900. He should have gained only 18,400 votes in Miami-Dade County but saw a gain of 37,000 - a difference of 19,300 votes.
"For the sake of all future elections involving electronic voting - someone must investigate and explain the statistical anomalies in Florida," says Professor Michael Hout. "We're calling on voting officials in Florida to take action."
The research team is comprised of doctoral students and faculty in the UC Berkeley sociology department, and led by Sociology Professor Michael Hout, a nationally-known expert on statistical methods and a member of the National Academy of Sciences and the UC Berkeley Survey Research Center.
For its research, the team used multiple-regression analysis, a statistical method widely used in the social and physical sciences to distinguish the individual effects of many variables on quantitative outcomes like vote totals. This multiple-regression analysis takes into account of the following variables by county:
- Number of voters
- Median income
- Hispanic/Latino population
- Change in voter turnout between 2000 and 2004
- Support for Senator Dole in the 1996 election
- Support for President Bush in the 2000 election.
- Use of electronic voting or paper ballots
"No matter how many factors and variables we took into consideration, the significant correlation in the votes for President Bush and electronic voting cannot be explained," said Hout. "The study shows, that a county's use of electronic voting resulted in a disproportionate increase in votes for President Bush. There is just a trivial probability of evidence like this appearing in a population where the true difference is zero - less than once in a thousand chances."
The data used in this study came from public sources including CNN.com, the 2000 US Census, and the Verified Voting Foundation.
For a copy of the working paper, raw data and other information used in the study can be found at: http://ucdata.berkeley.edu .