Trump won 75 of 83 counties.
Clinton won Wayne County (Detroit) by 290,451 votes (69.4-30.1%)
Trump won the other 82 counties by 301,155 votes (54-46%)
NY Post Dec. 14, 2016:
“The Detroit News found voting scanning machines at 248 of the city’s 662 precincts — 37 percent — tabulated more ballots than the number of actual voters counted in the poll books.
“There’s always going to be small problems to some degree, but we didn’t expect the degree of problem we saw in Detroit. This isn’t normal,” Krista Haroutunian, chairwoman of the Wayne County Board of Canvassers, told the paper.”
Based on the following data, it is difficult to believe that Hillary won the state by 4.27 million votes. She won the national recorded vote by 2.8 million, so Trump won the other states by at least 1.5 million (and that is conservative).
– Clinton did 7.0% better in CA than Obama in 2012. Clinton won by 61.7-31.6%, a 30.1% margin (4.27 million votes). Obama won by 60.2-37.1%, a 23.1% margin (3.01 million votes). If Clinton’s margin was 23.1%, she would have won by 3.3 million votes.In the CNN final CA exit poll (matched to the reported vote), the Party-ID is 47D-23R-30I . Was Clinton more popular than Obama? Not plausible.
– When the CA final exit poll Party-ID is adjusted from 47D-23R-30I to 34.2D-22.3R-43.5I based on the change in national Party ID from 2014, Clinton wins CA by 56.1-36.5% (2.78 million votes).
– Humboldt County, CA is the only county in the U.S. which uses Open Source software to count and audit votes. In the CA primary, Sanders had his highest vote share (71%) in Humboldt. Jill Stein also had her highest share (6.2%) in Humboldt.
Is it just a coincidence that Bernie and Jill both had their highest vote shares in Humboldt? Or was it due to the foolproof Open Source voting system?
– Hillary had 56% in Humboldt, nearly 6% lower than her total CA share. It is a fact that Bernie was cheated in CA by massive fraud. Who is to say that Hillary did not also cheat in CA to pad her popular vote margin?
There were 1307 NY Exit Poll respondents at 9 pm and 1391 at the final – an increase of just 84 respondents. Adjustments made to force the final 1391 exit poll to match the recorded vote in all exit poll categories are mathematically impossible. Therefore, the recorded vote was also mathematically impossible. The impossible adjustments are irrefutable proof of election fraud.
Let’s review the adjustments as a quiz.
1. At 9pm, Clinton had a) 51%, b) 52%, c) 53%
2. Clinton won the recorded vote with a) 57.3%, b) 57.7%, c) 57.9%
3. She had 28% of 18-29 year-olds. In the final she had a) 33%, b) 35%, c) 37%
4. She had 45% of males. In the final she had a) 49%, b) 50%, c) 51%
5. She had 71% of blacks. In the final she had a) 74%, b) 75%, c) 76%
6. She had 57% of Democrats. In the final she had a) 60%, b) 61%, c) 62%
7. She had 55% of Urban voters. In the final she had a) 59%, b) 60%, c) 62%
8. At 9pm, Urban voters comprised 55% of the total vote.
At the final, they comprised a) 62%, b) 64%, c) 66%
9. At 9pm, Clinton had 680 (52%) of 1307 respondents. She had 802 (57.9%) at the final (1391), an increase of 122 among the 84 final respondents.
This was a) a polling error, b) of no consequence, c)an absolute indicator of fraud.
10. At 9pm, Sanders had 622 (48%) and 589(42.1%) at the final, a 33 vote decline.
This was a) a polling error, b) of no consequence, c)an absolute indicator of fraud.
11.The probability of the 11.8% exit poll discrepancy from the recorded vote is
a) 1 in 91,000, b) 1 in 94,000, c) 1 in 102,000
12. The probability that Sanders exit poll share would be greater than his recorded share in 21 of 23 primaries is a) 1 in 13,000, b) 1 in 30,000, c) 1 in 40,000
13. In the NY Cumulative Vote Share analysis, Sanders and Clinton were tied after
a) 400,000, b) 500,000, c) 600,000 of 1.79 million total votes
Michigan Primary: Sanders did better than his recorded vote indicates
Richard Charnin (with John Brakey) Updated: March 13, 2016
This analysis indicates that Sanders did much better than his recorded vote in the Michigan primary. Sanders had 590,386 votes (49.8%) and Clinton 570,948 (48.3%). Sanders won in 73 of 83 MI counties with 56% of the vote. He won the preliminary exit poll by 52.1-45.9%, a 97% win probability. Clinton won urban counties Wayne and Oakland with approximately 55% of the vote.
Clinton won the Massachusetts primary by just 1.4%, but she did well in urban areas. Sanders won hand-counted precincts by 57-40% in 68 Towns (32,360 votes, 2.7% of votes cast). Sanders also had 52.1% in the preliminary exit poll which he won by 52.1-45.7%. His win probability was also 97%.
Once again, we have multiple confirmation indicating fraud: Cumulative vote shares, preliminary exit poll, absentee vote anomalies and other anecdotal information.
Will we see the same fraud indicators in FL, OH, IL, MO and NC on March 15?
It should be conventional wisdom by now: in state elections, fraud abounds in heavily populated urban and suburban locations. Of course, the media never talks about it. They report the recorded numbers as if there was not a fraud factor.
Election Fraud Indicators
Sanders had 1) 56% at the 600,000 Cumulative vote share mark, 2) 54% of approximately 500,000 votes cast on AccuVote and Sequoia voting machines and 3) led 52.1-45.9% in the unadjusted exit poll.
Clinton had 1) 75% of approximately 240,000 absentee votes and 2) 51.2% of approximately 700,000 votes cast on ES&S Mod 100 machines. The percentages are highly suspect.
Voting Machines (optical scanners)
Sanders’ county vote shares were negatively correlated to machine types. The ES&S Model 100 was a highly negative -0.68. The bigger the county the lower Sanders’ vote share.
Wayne and Oakland counties used ES&S Model 100 optical scanners. Macomb used both ES&S and Premier/Diebold/Dominion AccuVote optical scanners.
Adjusted Final Exit Poll – 1601 respondents (forced to match the recorded vote)
Absentee Votes (AV) differed substantially from the overall county vote results. Generally absentee voting is a close match to the precincts.
The Democrats had an estimated 237,000 AV. Approximately 177,750 (76%) voted for Clinton and 59,250 for Sanders. How did Clinton win AV by 76-24%? One would expect that Sanders and Clinton would nearly split AV.
Some have suggested that the reason Clinton won absentees by 50% is that they are typically older voters who supported Hillary. But Clinton won 60% of 45+ voters in the adjusted final exit poll. Since all exit polls are forced to match the recorded vote (see AAPOR ), Clinton probably had less than 60%. So much for the age issue.
This Sensitivity Analysis shows the effect of Clinton’s share of absentees from her estimated 177,750 recorded absentee votes to Sanders 59,250.
In the CVS analysis, Sanders had approximately 56% at the 600,000 mark. Notice the abrupt change to straight lines at the 600,000 vote mark. They represent the largest counties (Wayne and Oakland which used ES&S optical scanners exclusively.
Dec. 28, 2015
Updated: Jan.18, 2016
This post reviews the following statistical modelswhich strongly indicate fraud in the 2014 Governor elections:
1) True Vote Model, 2) Cumulative Vote Shares, 3) Voter Registration vs. Exit Poll Party-ID, 4) Uncounted Votes Cast (Census).
The CVS method uses actual precinct votes in each county. The data is sorted by precinct size. The votes and shares are accumulated and displayed graphically. Typically, in the biggest counties, Democratic shares peak at the 10% CVS mark and decline at the final 100% (recorded vote). This is counter-intuitive because a) the most populous counties are in urban locations which are strongly Democratic and b) as the number of votes are accumulated, the Law of Large Numbers (LLN) should result in a Steady state of equibrium in which Democratic and Republican vote shares are nearly constant.
Click these links to view the summary 2014 CVS analysis (each has a link to the precinct votes for each county): ILFLWIMDMAKY
The True Vote Model (TVM)
The 2012 presidential election is used as a basis for returning 2014 voters. There are two options for estimating returning voters: the Recorded Vote and estimated True Vote.
The TVM closely matched the CVS in all governor elections except for Maryland. Hogan(R) won the recorded vote by 51.0- 47.2%, a 66,000 vote margin. Brown(D) won the True Vote by 56.4-41.9%, a 251,000 margin. The CVS analysis understates Brown’s vote since precinct votes were provided only for Election Day; early, provisional and absentee precinct voting were not included. This omission dramatically reduced Brown’s CVS since he had 54% of the excluded votes. Click these links to view the 2014 Governor True Vote Model: MDILFLWIKYMAMEOHKSMIGACO
In each election cycle, the Census Bureau interviews 60,000 households nationwide to estimate how many were registered and voted in each state.The national margin of error (MoE) is 0.3% for 60,000 respondents at the 90% confidence level. The MoE is approximately 2% for each state.
In 2014, 92.2 million votes were cast but just 78.8 million recorded. The 13.4 million discrepancy (14.5%) was greater than in any presidential election. What is going on here?
In every one of the 1968-2012 presidential elections, votes cast exceeded the recorded vote. The percentage of uncounted votes has declined steadily since 1988, from 10.4% to 2.9% in 2012. Uncounted votes peaked at 10.6 million in 1988 and declined to near zero in 2008. Approximately 75% of uncounted votes were Democratic (50% in minority locations).
The recorded vote was adjusted to total votes cast by adding the uncounted votes. The majority (60-75%) of uncounted votes were assumed to be Democratic, based on the historical fact that approximately 50% of uncounted votes are in minority locations.
Uncounted Votes = Census Total Votes Cast – Votes Recorded
True Vote (est.) = Recorded Vote + Uncounted vote
The simplest measure of party strength in a state’s voting population is the breakdown-by-party totals from its voter registration statistics from the websites of the Secretaries of State or the Boards of Elections. As of 2014, 28 states and the District of Columbia allow registered voters to indicate a party preference when registering to vote.
In 2014, the party voter preference/registration split was 40.5D-35.3R-24.2I. The 2014 National Exit Poll indicated a 35D-36R-28I Party-ID split in forcing a match to the recorded vote (Dem 46.2-Rep 52.9%). Assuming the voter registration split and the Party-ID vote shares, the Democrats and Republicans were essentially tied.
The registered voter split for the 12 Governor elections in this analysis was 40.6D-34.4R-24.4I, a very close match to the national split.
The partisan “demographics” were obtained from the state’s party registration statistics (in late 2014 whenever possible). For the 22 states that don’t allow registration by party, Gallup’s annual polling of voter party identification is the next best metric of party strength.
Matching the Recorded and True Vote using the Party Registration split
To match the recorded vote, an implausibly low percentage of Democrats had to have voted for the Democratic candidate. Note the difference between the percentage of Democrats required to match the recorded vote and True vote shares. Democratic and Republican candidates usually win approximately 90-92% of registered Democrats and Republicans, respectively.
Percentage Share of Registered Democrats Required to Match
Conway (D) lost the recorded vote by 52.5-43.8% despite the fact that the Democrats led 53.4-38.8% in voter registration. Bevin needed an implausible 24.6% of Democrats to match the recorded vote. Assuming just 81.7% of Democrats voted for Conway, he won by 48.8-47.5%, closely matching the CVS and True Vote.
According to the 2014 Census, 1.525 million total votes were cast in 2014. In 2015, Conway won by 49.3-47.0% – assuming he had 60% of an estimated 50,000 uncounted votes.
Of 2,298,000 registered voters, 974,000 (42.4%) voted. If 39% of Democrats voted and Conway had 88%, he won by 49.0-47.3%. If 43% of Democrats voted, he won by 54-42.3%.
Hogan (R) won the recorded vote by 51.0-47.3%. Brown(D) had 53.7% of early votes, 45.3% on Election Day and 54.5% of absentee and provisional ballots. Precinct votes on touchscreens and optical scanners were provided for Election Day only.
When 295,000 uncounted votes are added to the recorded vote, Brown is the winner by 51.2-47.0% . When Election Day CVS at the 10% mark is added to the 390,000 early, absentee and provisional votes, Brown is a 52.9-45.5% winner.
– Kathy Dopp is a mathematician and an expert on election auditing. She has written a comprehensive analysis of the 2014 elections: Were the 2014 United States Senatorial and Gubernatorial Elections Manipulated? Dopp wrote: Is it possible electronic vote-count manipulation determines who controls government in the United States? The probability that the disparities between predicted and reported 2014 election vote margins were caused by random sampling error is virtually zero. A method for extending and simplifying fuzzy set qualitative comparative analysis (FsQCA)’s measure for necessity reveals that lack of effective post-election audits is a necessary condition for the occurrence of high levels of disparity between statewide polls and election results. Maryland’s 2014 gubernatorial contest is consistent with an explanation of vote miscount having altered its outcome. An analysis of Maryland’s partisan voter registration, turnout, and vote data by ballot type statistically confirms vote miscount as an explanation for its unexpected outcome.
Maryland, Illinois, Florida, and Kansas gubernatorial contests exhibited sufficient disparities between polls and election results (PED) to alter election outcomes; all used inauditable voting systems or failed to conduct post-election audits (PEA)s. Vermont’s PED was within one percent of sufficient to alter its outcome. In Nevada, Tennessee, New York, Ohio, and South Dakota PED were large but smaller than winning margins.
Kansas and North Carolina senatorial contests exhibited sufficient PED to alter election outcomes and no audits were conducted. Virginia’s PED was within one percent of sufficient to alter its outcome. In Arkansas, Wyoming, Tennessee, Kentucky, and Nebraska PED magnitude were large but smaller than winning margins.
A case study of Maryland’s unexpected 2014 gubernatorial outcome affirms there is, as yet, only an explanation of vote manipulation consistent with the statistical disparity patterns in Maryland’s pre-election poll predictions, and its partisan voter registration, turnout and vote data by ballot type.
Precinct data for 109 of 120 counties were downloaded. Eleven counties did not have detailed precinct data in Excel format (county votes included). The details are in two spreadsheets: KY2015Gov1 and KY2015Gov2
Cumulative vote shares Conway’s cumulative vote share declined from the 25% mark to the final in 75 of the 109 counties for which there is precinct data. The probability of this occurrence is 1 in 26,000. His vote share declined from the 10% mark to the final in each of the 17 largest counties. The probability is 1 in 262,000. We would normally expect a nearly even split of gains and losses.
Assuming the True Vote was at the – 10% CVS mark, Conway won by 49.5-46.4%. – 25% CVS mark, Bevin won by 48.3-47.7%
County vote data was categorized into groups from the largest 15 counties to the 120 total. Conway had 56.4% of the Top 15 at the 10% CVS mark and 53.9% at the 25% mark.
Vote share correlation
KY has 3,201,852 registered voters and 3,663 precincts (874 registered voters/precinct). An average of 268 votes/precinct (31.5% turnout)
Conway’s vote share was positively correlated to
– registered voters: 0.32 correlation at 25% CVS and 0.25 at final.
– ballots cast: 0.32 at 25% CVS and 0.27 at final.
– voter turnout: 0.20 at 25% CVS and 0.22 at final.
Six county audit
In the six small counties chosen for an audit (54,000 votes), Conway had 42.1% at the 25% mark and 39.3% at the final. A better choice would have been six large counties (352,000 votes) i which Conway had 60.2% at 25% and 52.9% at the final.
True Vote Model
A Sensitivity Analysis over a range of vote shares and returning voters from the 2012 presidential election indicated that Conway won the base case scenario by 49.9-46.8% with a 98% win probability, matching the pre-election polls. He won 64 of 75 scenarios.
In 17 counties for which there is absentee voter data, Conway had 45.4% in absentee paper ballots (984) vs. 32.6% in total ballots (65,915). His vote share in absentee paper ballots exceeded his total vote share in all 17 counties. He had 37.3% in machine absentees.
In the KY 2015 Governor election, Matt BEVIN (R) defeated Jack CONWAY (D) 52.5-43.8%, an 84,000 vote margin. Conway was leading in virtually all pre-election polls by 3 to 5 points yet he had fewer votes than other Democrats on the ticket.
In Kentucky, registered Democrats led Republicans by 53.4-38.8%.
The objective of CVS analysis is to view the effects of county/precinct size on the cumulative vote share trend. Since the largest counties are usually heavily Democratic, the consistent pattern of Republican Governor candidates gaining share from small to large precincts is counter-intuitive. On the other hand, there is virtually no change in vote shares in smaller, heavily GOP counties. This defies political reality and the Law of Large Numbers.
The analysis indicates that the pre-election polls were likely correct and that Conway probably won the election. CVS analysis of the 2014 FL, IL, WI, MA and MD Governor elections exhibited the same anomaly: a counter-intuitive rise in GOP cumulative vote shares in larger counties. But the KY analysis indicates that the anomaly occurred in many smaller counties.
Precinct votes for 120 Kentucky counties were downloaded to two spreadsheets: KY2015Gov1 and KY2015Gov2
TheSummary sheet displays recorded final vote shares and cumulative shares for 120 counties at the 10% and 25% CVS mark.
For example, in a county with 20,000 votes, the 25% CVS mark represents the first 5000 cumulative votes, starting from the smallest precinct.
Assuming that the True Vote was at the – 10% CVS mark, Conway won by 49.5-46.2%. – 25% CVS mark, Bevin won by 48.3-47.7%.
As in previous CVS analysis, it appears that votes were stolen in Democratic strongholds. In the 25 counties in which he was leading at the 25% mark, Conway’s 11.6% margin declined to 1.5%. In the other 95 counties, Conway’s -23.3% deficit declined just 1.9% to -25.2%.
Conway’s cumulative vote share declined from the 25% mark to the final in 53 of the TOP 60 counties, a 1 in 3 million probability! In a random process, we would expect a nearly even split. The probability is overwhelming evidence that there is an external fraud factor in the vote tabulation.
Another statistic of interest: a strong 0.33Correlation between county vote size and Conway’s vote share margin.
The 120 counties were sorted from largest to smallest to calculate the change in cumulative vote shares from the 25% mark. Conway’s share fell 5.6% from 53.9 to 48.3 in the TOP 15 counties; declined by 2.0% from 40.8 to 38.8 in the other 105 counties.This is a clear indication that votes were flipped in the largest counties.
The CVS method consists of the following steps:
1) Download the county precinct data
2) sort by precinct total vote
3) calculate a rolling sum of votes for each candidate
4) calculate corresponding cumulative vote shares
5) create the cumulative vote share line graph
6) check for divergence in shares from small to large precincts
7) calculate shares and changes from the 25% mark to the final
For example, view the Bath County spreadsheet. Conway’s vote share increased from 47.1% at the 25% mark to 49.9% at the final.
Down ballot Anomalies
Bev Harris of BlackBoxVoting.org noted that the higher Democratic vote totals in the down ballot races were a “significant anomaly”.
Bevin (R) won the total recorded vote by 52.5-43.8%.
Lundergan (D) won the Secretary of State by 51.16-48.84%.
Beshear (D) won Attorney General by 50.12-49.88%
Bevin won the Top 25 counties by 49.3-47.0%.
Dem SOS candidate Lundergan won the Top 25 by 53.3-46.7%.
Confirming CVS at 10%, Conway wins by 50.1-46.2% if we add the 6.3% difference in Lundergan’s Top 25 share to Conway’s total 43.8% share.
This sheet shows two-party vote shares for Governor, Attorney General and Secretary of State in the largest 25 counties.
TheTrue Vote Model displays various scenarios of returning Obama and Romney voters along with corresponding vote shares. The model uses two returning voter estimates based on Obama’s 2012 recorded KY vote and his True Vote. The model matched the 2015 recorded vote assuming Obama’s recorded KY 2012 vote. It matches the CVS vote shares at the 10% mark assuming Obama’s True Vote. The Sensitivity analysis tables display Conway’s total vote shares and margins over a range of returning voter turnout and vote share scenarios of returning and new voters.
The cumulative vote share at the 25% mark is the baseline for estimating the True Vote. But we may want to view changes in vote share in the 0-25% range if we assume that votes were also flipped in the smallest precincts.
Let V25 = the cumulative vote share at the 25% mark and VF = the final recorded share. The adjusted vote share (TV) is given by the formula:
TV = (V25 – VF) * 1.33 + VF
For example in Jefferson County, Conway had V25= 66.0% and VF= 58.8%, a 7.2% decline. The adjusted True Vote estimate is: TV = 68.4% = (66.0-58.8) * 1.33 + 58.8 = 9.6% + 58.8%
The 10% mark is also used as an estimate of the True Vote.
TOP 80 counties (895,000 votes – 92% of the total)
Conway led by 50.6-45.2% at the 10% mark.
Conway led by 48.6-47.3% at the 25% mark.
Bevin won the recorded vote in the TOP 80 by 51.9-44.4%.
TOP 60 counties (829,000 votes – 85% of the total)
Conway led by 51.4-44.3% at the 10% mark.
Conway led by 49.5-46.4% at the 25% mark.
Bevin won the recorded vote in the TOP 60 by 51.2-45.1%.
TOP 40 counties (735,000 votes – 75% of the total)
Conway led by 50.9-44.9% at the 25% mark. At the final, Bevin led 50.2-46.1%, a 10.1% reduction in Conway’s margin. Assuming Conway had 113,000 (47%) of the remaining 239,000 votes in the other 80 counties. he would win by 50-46%.
TOP 25 counties (625,000 votes – 64% of the total)
Conway led by 51.9-43.9% at the 25% mark. At the final, Bevin led 49.3-47.0%, a 10.3% reduction in Conway’s margin.
TOP 15 counties (518,000 votes – 53% of the total)
Conway led by 53.9-41.9% at the 25% mark. At the final, Conway led the Top 15 by 48.3-47.9%, a 11.7% reduction in Conway’s margin.
Given Conway’s 53.9% vote share in the Top 15 counties at the 25% mark, Bevin needed 305,000 (67%) of 455,000 votes in the other 105 counties to match his recorded vote margin (52.5-43.8%). He had 263,000 votes in the 105 counties. Given the anomalies in the TOP 15 counties, it is safe to assume that he had less than 263,000 votes.
Sensitivity Analysis (15 scenarios)
Conway had 40% of the recorded vote in the Bottom 60 counties, but we do not have the CVS for the group. Since we have CVS estimates for the TOP 60, we calculate estimates of Conway’s total vote using a range of vote share estimates for the Bottom 60.
1) formula, 2) 10% CVS, 3) 25% CVS. He won 12 of the 15 scenarios.
Conway ties Bevin with 40% (the break-even share) in the smallest 60 counties.
Assuming Conway had…
41% of the Bottom 60 and
– 51.5% of the TOP 60, he wins by 49.9-46.1% (36,800 votes).
– 49.5% of the TOP 60, he wins by 48.2-47.8% (4,700 votes).
43% of the Bottom 60 and
– 51.5% of the TOP 60, he wins by 50.2-45.8% (42,600 votes).
– 49.5% of the TOP 60, he wins by 48.5-47.5% (10,500 votes).
Compare the recorded vote shares in the following county subgroups to the cumulative shares at the 25% mark. Conway leads Bevin in all groups from the Top 15 (53% of the total vote) to the Top 80 (92% of the total).
A Democratic stronghold, Jefferson is the largest county in KY with 192,391 recorded votes. At the 25% mark (48,000) votes, Conway led in Jefferson by 66-30%. He won the county by 58-39%. The 17% change in margin lowered his vote margin from 69,000 to 31,000. But there may have been vote flipping from zero to 48,000. Conway led by 70-27% after the first 11,500 votes.
There were 69,953 recorded votes. At the 25% mark, Conway led by 60-34%. He won the county by 55-40%. The 11% change lowered his margin from 18,000 to 10,000 votes.
There were 31,453 recorded votes. At the 25% mark, they were tied at 47%. Bevin won the county by 57-39%. The 18% change increased Bevins’ margin from 80 to 5700 votes. Conway led by 53-41% after the first 2,200 votes (7% mark).
Cumulative Vote Shares were calculated for the following 2014 elections. All exhibited the same counter-intuitive rise in GOP cumulative vote shares.
2014 KY Senate Election
McConnell (R) defeated Grimes (D) by 56.2-40.7%, a 222,000 vote margin. In Jefferson County, Grimes had 56.1% and won by 35,000 votes Grimes’ vote share was 67.5% at the 10% CVS mark and 64.9% at 25%. But Grimes had just 53.3% from 25% to the final – an 11.6% decline. Assuming Grimes had a 64.9% True Vote, she would have won Jefferson County by 80,000 votes. Conway had 66.0% at the 25% mark and ended with 58.2%, a 7.8% decline. Grimes had 64.9% and ended with 56.1%, an 8.8% decline.
A statistical study by G.F. Webb of Vanderbilt University, Precinct Size Matters: The Large Precinct Bias in US Presidential Elections, reveals a correlation of large precincts and increased fraction of Republican votes.
Francois Choquette and James Johnson exposed anomalies in the 2012 primaries. 2008/2012 Election Anomalies, Results, Analysis and Concerns
Mathematician Kathy Dopp has written a comprehensive essay on the 2014 elections. She is an expert on election audits.
My interview with Thom Hartmann on the KY election
Compare the 2014 Senate race to the 2015 Governor contest.
Before discussing the CVS (below), a quick review: Kloppenburg (Independent) apparently won the election by 200 votes. But two days later, 14,000 votes were “found” in Waukesha County. Prosser (Republican) was declared the unofficial winner by 7,000 votes. The subsequent recount was a travesty. Scores of slit ballot bags and poll tapes dated a week before the election were uncovered. A stack of 50 consecutive Prosser ballots were found in Verona where Kloppenburg won 67% of the recorded vote – a zero probability.
The 2011 WI Supreme Court True Vote Model was enhanced to calculate the True Vote in all counties. It indicated that Kloppenburg won the election. Assuming a 50% turnout of both Obama and McCain voters, the recorded margin required implausibly low 81% Kloppenburg share of returning Obama voters while Prosser had 93% of returning McCain voters.
As previously shown in the 2014 WI Governor and in the 2012 Recall election, CVS anomalies occurred in the largest counties where the average ward vote is higher than in smaller, rural (heavily GOP) counties. Overall, there was a 2.47% decline (37,000 votes) in Kloppenburg’s vote share from the 25% mark. But there was a 5.0% decline (42,000 votes) in the Top 10 counties in which 56% of the votes were cast. Kloppenburg gained nearly 9,000 votes in the 52 smallest counties, a confirmation that they were effectively ignored by the GOP.
In Milwaukee County, Kloppenburg had 74% at the 26,000 vote mark but ended up with 57% at the final 228,000. The 17% decline meant that 38,000 votes were flipped to Prosser – a 76,000 decline in margin! She “lost” by 7,000. Click for the Milwaukee County CVS chart.
In Waukesha County, Kloppenburg’s vote share dropped 3,400 votes from 28.9% at the 25% mark to 25.2% at the final – a nearly 6,800 decrease in margin and close to the magical Waukesha vote adjustment which gave the election to Prosser.
Note that the declines (discrepancies) may actually be greater than above as they reflect changes from the 25% CVS mark – not from the start to the 25% count.
The results confirm previous counter-intuitive findings that Republicans consistently gain share in the most populated counties where precincts/wards are usually heavily Democratic. There were virtually no vote share changes in small, heavily Republican rural counties. In fact, Democrats and Independents often gain vote share from the 25% mark in these counties.
Kloppenburg lost 46,000 votes from the 25% mark in the largest 20 counties. She lost share in 15 of the largest 18 counties, but gained share in 37 of the smallest 54. She actually gained 7,000 votes in the smallest 52 counties. She lost 40,000 votes in Democratic counties in which she led at the 25% mark, but gained 1,000 votes in (Republican) counties in which she had less than 50%. Kloppenburg actually gained share in the smallest 52 counties.
As in recent WI Governor elections, vote share declines were highest in Milwaukee (11%, 25000 vote loss), Racine (15%, 7600), Waukesha (2.7%, 3300), Kenosha (10.3%, 3100) and Winnebago (5.6%, 2100).
Kloppenburg’s vote shares were higher in the smallest (0-50%) precincts compared to the largest (50-100%).
Milwaukee 64%> 50%
Brown 47 > 43
Kenosha 58 > 48
Racine 52 > 36
St. Croix 50 > 48
Waukesha 27 > 25
Winnebago 51 > 45
Once again, the evidence shows that Republicans steal elections in big urban counties that are strongly Democratic and ignore small rural counties where they are dominant.
To appreciate the vote changes, think of the starting 10,000 votes as a poll with a 1% margin of error. Move the cursor over the CVS trend line to view the exact vote count and share. Milwaukee County Steady 17% decline from 74% at 25,000 to 57% at the final 228,000. Brown After leading at 10,000 votes, Kloppenburg’s share declines to 45% at the final 61,000. Kenosha Steady, massive decline from 65% at 3,000 votes to 53% at the final 31,000. Racine Strange decline from 60% at 10,000 votes to 45% at the final 51,000. St. Croix Coincident shares all the way to the final 16,000. Was St. Croix legit? Waukesha The biggest GOP stronghold, but is it this strong? Kloppenburg gained shares in smaller GOP counties, but not in Waukesha where her share declined from 32% at 10,000 votes to 26% at the final 125,000. Winnebago Decline from 55% at 5,000 votes to 48% at the final 40,000.
Questions have been raised as to whether the number of elections analyzed is sufficient to draw conclusions. Given that approximately 20 million votes in 13 elections have been analyzed, the results are statistically significant. The analysis is confirmed by other forensic methods (True Vote Model, exit polls) for competitive and non-competitive races.
The analysis of cumulative vote shares (CVS) has revealed a consistent pattern. It is a well-known fact that Democrats are the majority in highly populated urban locations; the largest precincts are usually Democratic. Republicans are heavily represented in rural areas. But in scores of state elections there has been an increase in cumulative Republican vote shares in larger precincts. This anomaly has been noted by PhDs in Kansas and Vanderbilt University.
The basic premise is that Republican increase in cumulative precinct vote shares is counter-intuitive since the Democrats do much better in urban and suburban counties than in rural areas where the GOP is dominant. Precincts in Urban areas contain more voters than rural areas.
Since the GOP gains share in Democratic locations in virtually all of the competitive elections analyzed, it is highly suggestive evidence that Democratic precincts are where the majority of votes are stolen. In competitive elections, the correlation between county/precinct vote-size and the change in Democratic vote share is negative; Democrats lose share as county/precinct size increase. On the other hand, in non-competitive races, the Statistical correlation is close to zero; there is virtually no relationship.
The numerical evidence in each election is clear.
1- In the 15 largest counties, Republican vote share increases from the 25% mark to the final.
2- In the other smaller counties, there is virtually no change in vote share from the 25% mark.
3- In counties where the Democrats led at the 25% mark, their vote share declined significantly.
On Nov. 5, 2015 I posted this CVS analysis for the KY Governor race. We see the same counter-intuitive vote shares in the largest counties.
Consider the following changes from the 25% cumulative vote share to the final recorded share in six Governor elections: Table of CVS changes. Compare the CVS anomalies to the non-competitive races:
SD Gov: Daugaard (R) won by 70.5-29.4%. Note the slight 0.49% change in vote shares.
KS senate: the Independent lost by 53.2-42.2% but nearly tied the Republican in the Top 15 counties (618,000 votes). In the other counties (264,000 votes), he lost by 64.3-31.3%.
In the Maryland election, Hogan (R) defeated Brown (D) by 65,000 votes (51.7-47.2%). But Brown won the 301,000 early and 83,000 late votes (absentee and provisional paper ballots) by 53.9-44.5%. Hogan led on Election Day voting machines (1,319,000 votes) by 52.9-45.3%.
This anomaly also occurred in the 2000-2012 presidential elections. The Democrats did much better in early and late voting. In 2012 Obama led in early voting (40 million) by 55-43%; he led in late voting (11.7 million) by 58-38%. He lost to Romney on Election Day (77 million) by 50.4-47.9%.
– A Vanderbilt Univ. statistical study of precinct level data in US presidential elections reveals a correlation of large precincts and increased fraction of Republican votes.
– Wichita State University engineering professor and statistician Beth Clarkson has accused three states — Wisconsin, Ohio, & Kansas — of voting irregularities that indicate a tampering of electronic voting machines.
Cumulative Vote Share (CVS) anomalies were noticed in a 2005 Ohio special election election favoring Schmidt (R) and in the 2012 GOP primaries favoring Romney.
Michael Collins wrote this article on the 2005 race: “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.
Precincts with the most votes favored Schmidt at nearly 100%, with Hackett winning in only those with less than 200 votes counted. A review of precinct level results by Charnin on Democratic Underground reveals this interesting trend. This data is preliminary and more detail needs to be obtained from the Clermont Board of Elections. However,the data observed for Clermont makes little sense on the face of it.
Hackett won 38 of 191 Clermont precincts with fewer than 187 votes, but lost ALL of the largest 54 precincts (those with more than 187 votes each). This is reflected in a graph produced by Democratic Underground poster TruthIsAll, one of the first election fraud analysts to notice anomalies in Clermont County. Hackett’s percentage by precinct group size: 46.9% in precincts under 100 votes 43.5% in precincts of 100-200 votes 39.6% in precincts of 200-300 votes 34.6% in precincts of 300 + votes
These results raise interesting questions. Why does Hackett do much better in the smaller precincts? Are they more rural than the larger precincts? If so, does this not present a counter-intuitive pattern, with the Democrat taking some of the conservative, less populated areas and the Republican winning all of the precincts in the most populated areas?
A question can be raised about the difference between turnout (the votes cast) and the actual size of the precinct, which may or may not be a reflection of votes cast. The following graph, also produced by Charnin, answers the question.
As he said while commenting on this data on 8/5/05: “The regression line has zero slope. Voters turned out at a fairly constant rate across precincts. So turnout wasn’t a factor in explaining why the Schmidt vote percentage increased as precinct size increased.”
Collins also wrote a two-part article on the 2012 GOP primaries in which a CVS analysis showed a consistent pattern of votes being flipped to Romney. “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 Choquet 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 on a political discussion forum.”
This review of Jonathan Simon’s book Code Red mentions CVS analysis:
Simon discusses another kind of evidence pointing to suspicious election outcomes. “The evidence is based on a method called “cumulative vote share” (CVS) analysis in which a graph is made for county results that shows the cumulative vote percentage by adding in precincts according to the precinct size, with the smallest precincts included first. According to statistical theory, the resulting graph should look something like the following result for DuPage County in the 2014 Illinois Governor race. Note: these graphs were produced by Richard Charnin and presented at his web site.
These curves are fairly flat on the right side as the larger precincts are included in the vote totals, which means that vote percentages did not change a great deal with precinct size. This is what is expected. However, some counties showed a result that is unexpected from a statistical viewpoint. Here is the same plot for Peoria County in the same election:
This kind of graph, in which the Republican’s result shows an increasing percentage, while the Democrat’s percentage is decreasing is repeated in many races in many states throughout the country. Again, these patterns are not proof of election fraud, but they raise questions and underscore the need for a transparent, auditable system that can be trusted by the voters. Why the public is not more alarmed about the distressing state of our election system is a question that perplexes and frustrates Simon. Perhaps his book, along with the work of other election integrity activists will eventually rouse the public to demand reform. He expresses these sentiments with eloquence in the book:
“I’d like to think this story will have a happy ending, that history will review in appreciative terms the struggle of a few activists—Cassandras really—to prod leaders and public alike to scale the towering Never-Happen-Here Wall Of Denial so that they can then act together to restore the essential element of observable vote counting to our nation. Most truths eventually come out. All we can do is keep trying in every way possible to help this one find its way into the light.” (CODERED, p. 14)
CVS and TVM analysis is confirmed by the following studies:
– A statistical study of precinct level data in US presidential elections reveals a correlation of large precincts and increased fraction of Republican votes.
– Wichita State University engineering professor and statistician Beth Clarkson has accused three states — Wisconsin, Ohio, & Kansas — of voting irregularities that indicate a tampering of electronic voting machines. In her recently published journal article, she reviews the statistical anomalies in the three states — including laying out her entire mathematical methodology, inviting others to replicate the study. Clarkson has filed suit trying to gain full access to the ballots for an independent audit of the paper ‘hard copies.’
Rauner (R) defeated the incumbent Quinn (D) by 170,000 votes (50.7-45.9%). The following analysis indicates that Quinn may have actually won re-election. The 2014 Illinois Governor spreadsheet contains precinct votes by county, the True Vote Model and adjusted exit poll.
– Pre-election Polls: Quinn led the final LV pre-election polls: 45.6-44.8%. LV polls are a subset of Registered Voter (RV) polls. Respondents deemed unlikely to vote (most of them Democratic) are eliminated from the full RV sample. RV polls usually match the Unadjusted exit polls and the True Vote Model. LV polls have closely matched the recorded vote.
– Cumulative Vote shares: The largest counties all showed Rauner vote shares increasing with cumulative precinct vote totals. This is a major red flag. The results confirm the CVS trend: GOP cumulative shares rise from the smallest to the largest counties, as shown in the graphs. At the 10% CVS mark, Quinn had 54.4%, compared to his final 45.7%. At the 25% mark, Quinn had 52.7%.
– Exit poll anomalies: The Governor exit poll matched the recorded vote to within 0.4%. It is standard procedure to adjust the poll to match the recorded vote.In the Party-ID category, Democrats led Republicans by 43-30%. But only 85% of Democrats voted for Quinn while 64% of Independents voted for Rauner.
– The True Vote Model Quinn wins by 54.2-42.4%, a 428,000 vote margin.
1) Obama’s 2012 recorded 57.6% Illinois share
2) 60% turnout of Obama and Romney voters
3) Quinn has 87% of returning Obama voters
4) Quinn has 7% of returning Romney voters
The built-in sensitivity analysis shows the effects of a range of voter turnout and vote shares assumptions. The basis is the 2012 presidential election. His share of returning Obama voters ranges from 83-89%; returning Romney voters from 5-9%; new voters from 44-52%. Quinn wins 72 of 75 scenarios. A win probability matrix is displayed for 25 combinations of Quinn’s share of returning Obama and Romney voters. Quinn wins all scenarios.
Cumulative Vote Shares
The famous bank robber Willie Sutton was asked why he robbed banks. He replied: “That’s where the money is”. The Republicans know where to go to steal votes: the large urban counties where Democrats live.
The largest counties showed increasing Rauner vote shares as the cumulative precinct vote totals increased – a major red flag. It is clear that the largest counties were most likely fraudulent. There is little to gain in small counties which are strongly Republican to begin with.
Quinn had 54.4% at the 10% mark which declined to 46.5% (344,000 votes).
His 61.5% share of the 15 largest counties dropped to 51.2%.
His 33.5% share in the other 87 counties dropped to 30.4%.
CVS percentage and vote changes (in thousands)
(25% mark to the final recorded vote)
Cook.......... 8% 108
Will......... 15% 29.4
St. Clair.... 15% 11.4
Peoria....... 15% 7.8
Rock Island.. 11% 7.2
Sangamon..... 10% 7.15
Macon......... 9% 3
Winnebago..... 9% 7.1
Lake.......... 6% 12.1
DuPage........ 3% 8.6
There is a -0.37 correlation between county size and Quinn’s vote share change from 25% to the final 100%. Quinn’s share goes down as county size increases.
In Democratic leaning counties (1.7 million votes), Quinn’s share declined from 69.6% to 60.2% (162,000 thousand votes).
Heavily Democratic Cook county had 1.3 million of the 3.6 million state voters.
Quinn had 75% of the first 100 thousand votes in the smallest Cook precincts,
72% of the first 500 thousand,
69% of the first 1 million,
64.8% of the total 1.3 million who voted in Cook county.
Exit Poll Anomalies
The exit poll matched the recorded vote to within 0.4%. But it is standard procedure to adjust the poll to match the vote.The following crosstabs reflect the recorded vote – not the True Vote:
Gender: Quinn led the female vote by 51-44%, an increase from his 49-44% share in the 2010 election.
Race: Minority voters were 9% of the vote, but the vote shares are missing.
Philosophy: Liberals comprised just 25% of the electorate. Quinn’s 80% share declined from 84% in 2010. His share of moderates declined from a winning 7% margin in 2010 to a 12% loss.
Party-ID: Self-identified Democrats led Republicans by 43% – 30%. But only 85% voted for Quinn? Independents voted for Rauner by 64-29%?
Education: Quinn won Post graduates by 55-43% (20% of the vote). But Rauner won College grads by 60-36% (31% of the vote). This is an implausible discrepancy.
Labor: Quinn had just 58%?
Senate Election: Durbin (D) easily won re-election by 55-43%.
But just 82% of Durbin voters voted for Quinn?
– A statistical study of precinct level data in US presidential elections reveals a correlation of large precincts and increased fraction of Republican votes.
– Wichita State University engineering professor and statistician Beth Clarkson has accused three states — Wisconsin, Ohio, & Kansas — of voting irregularities that indicate a tampering of electronic voting machines. In her recently published journal article, she reviews the statistical anomalies in the three states — including laying out her entire mathematical methodology, inviting others to replicate the study. Clarkson has filed suit trying to gain full access to the ballots for an independent audit of the paper ‘hard copies.’ http://ivn.us/2015/07/20/report-2014-voting-machine-tampering-likely-wisconsin-ohio-kansas/