Analysis Tests for “Fraud” “Tampering” and “Illegals Voting” Claims—with Statistical Methods and Comparison with Recounts and Audit in Two States
CONTACT: Peter Schwartz
For Immediate Release
Results from Expert Team at Anderson Economic Group:
- No evidence of systemic fraud or tampering in MI, WI and PA.
- One county identified as an “outlier” had numerous irregularities, and a small number of non-systemic frauds detected, as confirmed by separate recounts and audits.
- No evidence of “3-5 million illegals” voting for Democratic candidate in these states; in fact, evidence against it.
- Findings inconclusive in CA; insufficient evidence from state’s election-reporting methods.
New York, NY, May 30, 2017—On May 27, Anderson Economic Group Principal and CEO, Patrick L. Anderson, presented the firm's most comprehensive election forensic analysis to date—covering four crucial electoral states—at the 14th Annual National Association of Forensic Economists International Meeting in Milan, Italy.
This analysis, like that which AEG experts released in January, involved a battery of statistical tests, applied to county-by-county election data for the 2012 and 2016 US presidential election, to evaluate allegations made by computer scientists and competing political factions of “fraud” and “tampering” of vote totals, as well as the vaguely-defined assertion of “3 to 5 million illegals” voting for the Democratic candidate in 2016. The analysis also made use of contemporaneous recounts done in Michigan and Wisconsin, and an audit done of over 100 precincts in the City of Detroit.
“The contemporaneous recounts and audits, done by election officials in two states, provide a unique comparison of these statistical methods with those from very expensive, time consuming manual recounts,” said Anderson.
Anderson Economic Group experts focused on the states of Michigan, Wisconsin and Pennsylvania, where recounts had been undertaken, along with California, which has the largest number of voters in the nation and, presumably, would be a center of any hypothetical “3-5 million illegals” vote operation.
In addition to recount data in Wisconsin and Michigan, Anderson Economic Group researchers had access to audit results from over 100 precincts in Detroit, including the referrals for criminal prosecution of a small number of election frauds. The carefully documented results rely on nonparametric and data analytics methods described in the paper, and conservatively assess the evidence both for and against the allegations. The results and recommendations should be of interest to citizens and election officials in all these states, as well as the President’s recently announced Commission on Election Integrity.
No Evidence of Systematic Fraud or Tampering. AEG found no evidence of systematic fraud or tampering in voting results in the states of Michigan, Wisconsin, and Pennsylvania. In particular, the data fulsomely reject the contention that tampering or hacking of voting machines caused the reported results in 2016 to systematically vary from the actual voting.
Small Numbers of Frauds Separately Confirmed in “Outlier” County. One county identified as a statistical “outlier” through forensic methods had numerous irregularities, and a small number of non-systemic frauds were detected, as confirmed by separate recounts and audits conducted by the State of Michigan.
This result suggests that election forensics methods can substantially reduce the costs involved in manual recounts of counties for which no substantial evidence exists of irregularities, and instead allow election officials to focus on the relatively small number of counties where the statistical pattern of voting is quite different than would be expected given other counties in the same state.
No Evidence of “Illegals” Voting En Masse in Three States. AEG experts also performed a separate analysis using demographic data on foreign-born population in these states, to test the vaguely defined assertion by the President of the United States that millions of non-citizen immigrants illegally voted for the Democratic Party candidate in 2016.
In Michigan, Wisconsin, and Pennsylvania, we find no support for this contention, and instead found evidence against it. This latter conclusion is based on many instances of lower proportional voting for the Democratic candidate in counties with high concentrations of immigrant populations in 2016 than in 2012, at least in these states.
California Results Inconclusive. In California, the state’s election data aggregated to the county level does not allow for sufficient variation in voting patterns to be examined by these methods, and therefore no evidence for or against the assertions of “fraud” or “tampering” in that state could be concluded. Similarly, due to the large aggregation in the county data, and the much higher immigrant population there than in the other states, AEG did not find the data assembled on alleged illegal voting to be conclusive.
Mr. Anderson recommended in his presentation to the National Association of Forensic Economists International Conference that a subsequent election forensic analysis be conducted in California that makes use of disaggregated data. He also recommended that election forensic techniques be implemented more frequently to discourage tampering, confirm legitimacy of results, and encourage democratic institutions.
CONTACT: Peter Schwartz
Selected Data Visualizations:
A sampling of comparative analytics from AEG’s election forensic findings includes vote share and distribution examinations across four crucial electoral states, at both the county and precinct levels: