(Article changed on November 2, 2012 at 01:24)
Part I of this series suggested that there may well have been massive vote flipping for candidate Mitt Romney in the Republican primaries (Rigged Elections for Romney (10/22/12) The article and the initial research analysis were received broadly. In addition, highly motivated citizens across the country and a team of high school students contacted the authors for help replicating the research in their states. The researchers, Francois et al., point out that this can be done with their open source techniques.
The basic argument is straightforward. If you look at precinct level voting data arranged from the smallest to the largest precincts, you will see Romney's gains increasing substantially as the cumulative vote increases. For example, Ohio and Wisconsin show this clearly as do eleven other states presented here. This extraordinary vote gain from smallest to largest precincts is so out of line, that the probability that this would happen by chance alone is often less than 1 out of a number represented by 1 preceded by 100 zeros and a decimal point, a value beneath the statistical package's lower limits. As a result, the researchers termed the suspected vote flipping for Romney the "amazing anomaly." (The Amazing Statistical Anomaly)
The research team's observation of Romney gains based on precinct size is not unique. The anomaly was raised previously concerning the Republican presidential primaries by a commenter on a political discussion forum.
Richard Charnin, posting as TruthIsAll, first noted the pattern with an analysis of the 2005 special election for a vacated seat for Ohio's 2nd district, in the House of Representatives. The candidates were the liberal-populist Democrat Paul Hackett versus a right-wing Republican, Jean Schmidt. Charnin noticed that Schmidt's votes and percentages increased substantially from the smallest to largest precincts in that district. This was a patently absurd pattern of vote accumulation since the liberal Hackett wins were in highly conservative counties that rarely voted for any Democrat. (See Precincts with the most votes favored Schmidt at nearly 100%)
Vote flipping is a form of election fraud that occurs "when votes are changed [without the voter's knowledge or consent] from one candidate to another or several others during electronic voting and vote tabulation." County election officials conduct computerized vote tabulation as precincts submit their voting results, but cannot detect the fraud because only the total number of votes is checked and vote flipping does not affect the total votes. The activity is suspected in many of the critical Republican primaries in 2012.
In each case, Mitt Romney was the beneficiary. For example, without vote flipping, Romney would have lost the Wisconsin, and Ohio, and Illinois primaries as well as primaries in other states. A comprehensive review will appear in Part III of this series.
Critics of the analysis presented in the first article claim that there is a perfectly logical expectation: that Romney would be more popular in suburbs. Hence the votes increase as precinct size increases, indicating a move to supposedly Romney favorable urban areas. Part I of this series mentioned that the research group anticipated that criticism and had factored out urban density from the analysis. The article linked to a fairly complex research analysis from the team. Here is a much simpler explanation, via example.
Mitt Romney was a candidate in the 2008 Maryland Republican primary. The race shows no statistical anomalies in vote accumulation from smallest to largest precincts for Romney.
In particular, you can see that Romney's flat lined in the 2008 Maryland primary. There is no indication that precinct size played any role in his accumulation of votes. The demographic criticism fails on the basis of extensive statistical analysis presented by the research team and linked in the first article illustrated by the graphic representation above.
What is the Likely Explanation for the Amazing Anomaly?
Which step in the vote counting process best explains suspected vote flipping?
The researchers, Francois et al., maintain that the likely culprits are central tabulators used by county election departments. Typically, election precincts submit their individual results to a county-run central tabulator. The central tabulator combines the precinct totals for a county total for every candidate on the ballot. That information is then transmitted to state election officials. The candidate vote count produced by the central tabulators is the critical element in the election process.
The central tabulators are computing devices owned or serviced by private companies such as Dominion, ES&S, and Hart Intercivic. The operational details and software engineering are almost always the exclusive knowledge and intellectual property of the private companies. As a result, public officials and citizens lack the type of access necessary to monitor the vote process.