Guest Author: Richard Hayes Phillips, Ph.D.: [Updated July 20 to include full detail report] It has been widely noted that the unadjusted CNN exit polls for the 2016 Democratic presidential primaries, in many states, are at variance with the official results. Some believe the exit polls are accurate, and the official results are fraudulent. Others believe the exit polls are inaccurate, and the official results are true and correct. Some begin with one assumption or the other, and present their evidence selectively, It is as if, when navigating a maze, they start with the cheese, and work their way back to the mouse.
No one seems to have compared the early and final exit poll data for black voters, and the exceptional variances found when comparing election results for voters at the polls with the numbers for early and absentee voters beg for closer review. No one has arranged for a public inspection of ballots in locations with the most compelling disparities, or even identified the precincts that need to be audited.
I have two working hypotheses. Either the official results are true and correct, or they are not. Exit poll analysis alone cannot prove fraud. Only by examining the actual ballots can the official results be verified or disproved.
It is my understanding that exit polls conducted in this country, including those posted on CNN, are adjusted to match the official results. The idea is to present an accurate demographic breakdown of the electorate that produced those official results. They are not specifically designed to detect fraud.
What the unadjusted exit polls should closely reflect is the vote count at the polls, on Election Day. Some states have early voting, and all states have absentee voting, and these votes are counted in the official results, but early voters and absentee voters do not vote on Election Day, and would not be available for interviews upon exiting the polls. Therefore, we might sometimes see a disparity when compared to the total vote count.
In 8 of 12 states with early voting (Arkansas, Florida, Georgia, Michigan, North Carolina, Ohio, Tennessee, Texas), the CNN “exit polls” did include telephone surveys of early voters, thus introducing a sampling bias that would not exist in a pure exit poll. Not included would be voters still out of town, or without land lines, or with unlisted numbers, or who use caller ID to screen their calls, or with no telephone at all.
A disparity between the unadjusted exit polls and the vote count at the polls is a red flag, a signal that further investigation is warranted, but it does not prove that fraud has occurred. A disparity between the ballots and the official results does prove that fraud or egregious error has occurred (unless the discrepancy is small enough to be attributed to minor error in counting the votes).
Generally speaking, election fraud does not occur across the board, by shifting votes from one candidate to another in every ward and precinct. Such alterations would be discovered by any random audit. Rather, alterations to the vote count vary from precinct to precinct, and are most apparent at the precinct level. The most anomalous precincts are the ones to be audited. If the ballots match the official count here, in a targeted audit, in suspect precincts specifically chosen for a fraud investigation, it is more convincing than a random audit. But if the ballots do not match the official count, it is prima facie evidence of fraud or egregious error.
I have experience in this field. After the 2004 presidential election, I examined precinct canvass results for the State of Ohio, and was retained as an expert witness in the Moss v. Bush lawsuit, for which I submitted 21 research papers to the Ohio Supreme Court. In the King Lincoln v. Blackwell lawsuit, I filed in Federal District Court an 18-page Declaration which was instrumental in protecting the ballots from destruction.
I subsequently authored and published a book entitled “Witness to a Crime: A Citizens’ Audit of an American Election” based upon photographs of 126,000 ballots, 127 poll books, 141 voter signature books, ballot accounting charts, and other election records from precincts of my choosing in 18 of 88 counties in Ohio, all of which I personally examined and analyzed.
In May 2016 I was hired by Cliff Arnebeck to analyze and compare official precinct canvass results for the 2008 and 2016 Democratic presidential primaries. It quickly became apparent that Hillary Clinton, against Bernie Sanders, had made tremendous gains among voters in predominantly black wards and precincts that she had lost overwhelmingly in 2008 to Barack Obama. It also became apparent, where the data are available, that Hillary Clinton’s percentages among early and absentee voters were almost always higher than her percentages among voters at the polls. I examined the exit polls and the precinct results from these perspectives.
See detailed report: Declaration of Richard Hayes Phillips
The final exit polls, adjusted to match the official results, have proved useful. These numbers display the demographic breakdown of the electorate by which the pollsters explain the official results.
My use of the word “unadjusted” comes with a caveat: these are exit polls posted on the CNN website at or near to the time the polls closed, and they may already have been adjusted prior to that.
In 21 states, the adjusted exit polls report the presidential preferences of the black portion of the Democratic electorate. In 10 of these 21 states, all in the South, Hillary Clinton reportedly received 80% to 91% of the black vote, and 67% to 75% elsewhere.
Between February 27 and March 8, primaries were held in nine southern states (Alabama, Arkansas, Georgia, Louisiana, Mississippi, South Carolina, Tennessee, Texas, and Virginia). In these nine primaries, Clinton won 499 pledged delegates to 212 for Sanders, a lead that has proven insurmountable.
The question arises as to whether Bernie Sanders really did receive less than half the support among black voters in the South as he did elsewhere in the country, or whether these vote counts were altered.
I am in possession of screen shots of unadjusted and final CNN exit polls in 22 states. The Democratic electorate is broken down by gender, and both genders are broken down by choice of candidate. It is an elementary mathematical procedure to calculate the combined percentages for both Clinton and Sanders.
In nineteen states, the margin between the candidates favors Clinton in the official results when compared to the unadjusted CNN exit polls. Either her official margin of victory was greater, or her official margin of defeat was smaller, than the unadjusted exit poll had indicated. Here are the disparities: Alabama (14.0%), Georgia (12.1%), New York (11.7%), Mississippi (9.4%), Texas (9.3%), Ohio (9.2%), Tennessee (8.2%), Massachusetts (8.0%), Indiana (5.8%), Michigan (4.8%), Arkansas (4.7%), Virginia (4.4%), Illinois (4.3%), Missouri (3.85%), Florida (3.4%), Pennsylvania (2.7%), Connecticut (2.2%), North Carolina (1.5%), and Vermont (1.1%). In three of these states (Massachusetts, Illinois, and Missouri), the winner and loser were reversed: the unadjusted CNN exit poll showed Sanders winning the primary, but the official results showed Clinton winning the primary. In three states, the official margin favors Sanders when compared to the unadjusted CNN exit polls: Oklahoma (6.2%), Wisconsin (2.0%), and Maryland (1.7%).
What no previous researcher seems to have done is to compare the exit poll data for black voters. Both the unadjusted and final exit polls report the percentage of the Democratic electorate comprised of black voters, and the percentage of black voters won by Clinton over Sanders. In 15 of 19 states, one or both numbers were adjusted upward in the final exit polls. If both numbers were adjusted upward, the effect is like compound interest. For example, if the percentage comprised of black voters is adjusted upward from 40% to 50%, and Clinton’s share of the black vote is adjusted upward from 80% to 90%, the percentage of the electorate comprised of black voters for Clinton is thus adjusted upward from 32% to 45%. Here are the numbers:
The percentage of the Democratic electorate comprised of black voters was adjusted upward in Alabama (from 47% to 54%), Georgia (from 49% to 51%), Illinois (from 27% to 28%), Indiana (from 17% to 19%), Mississippi (from 68% to 71%), Missouri (from 19% to 21%), North Carolina (from 29% to 32%), Pennsylvania (from 17% to 19%), Texas (from 16% to 19%), and Virginia (from 24% to 26%). In Michigan, the percentage was adjusted downward (from 23% to 21%).
The percentage of black voters won by Clinton over Sanders was adjusted upward in Alabama (from 87%-10% to 91%-6%), Arkansas (from (88%-12% to 91%-9%), Florida (from 78%-22% to 81%-18%), Georgia (from 83%-16% to 85%-14%), Illinois (from 69%-30% to 70%-30%), Michigan (from 64%-32% to 68%-28%), New York (from 71%-29% to 75%-25%), Ohio (from 67%-31% to 71%-28%), Tennessee (from 82%-12% to 89%-10%), Texas (from 80%-18% to 83%-15%), and Virginia (from 82%-18% to 84%-16%). In Oklahoma, her percentage was adjusted downward (from 75%-24% to 71%-27%).
The question is why the exit pollsters needed to do this. Were the numbers of black voters, or their support for Clinton, consistently underestimated in so many states, or were the official vote counts in the black communities altered, thus forcing the adjustments to the exit polls?
This question can only be answered by examining the ballots, where there are any. For example:
Sanders got less than 10% of the vote in 58 of 173 precincts in Jefferson County, Alabama, 47 of 121 in Montgomery County, Alabama, and 57 of 80 in Hinds County, Mississippi, and less than 20% of the vote in 48 of 270 precincts in Prince George’s County, Maryland, 25 of 83 in Brooklyn District 58, 23 of 93 in Manhattan District 73, 23 of 85 in Bronx District 83, and 55 of 336 precincts in Cleveland, Ohio — all with paper ballots marked by voters and tabulated by optical scanners. These precincts need to be audited.
Sanders got less than 10% of the vote in 68 of 315 precincts in East Baton Rouge Parish, Louisiana, and 42 of 114 in Memphis, Tennessee, and less than 20% of the vote in 90 of 347 precincts in Fulton County, Georgia, 27 of 182 in Charleston County, South Carolina, 41 of 106 in Gary, Indiana, and in two entire wards in Philadelphia, Pennsylvania — all with black box voting, that is, electronic voting machines, with no “voter verified paper trail,” and no way to verify the vote count, except for absentee ballots.
In Massachusetts, Clinton owes her victory to 13 of 22 wards in Boston, especially Ward 12 (Roxbury), Ward 14 (Mattapan), Ward 17 (Ashmont), and part of Ward 18 (Precincts 1-6), where Clinton lost every precinct to Obama in 2008, and won every precinct against Sanders in 2016. There are paper ballots and optical scanners everywhere in Boston, and these precincts need to be audited.
The question also arises as to whether Clinton’s impressive numbers among early and absentee voters were entirely earned, or whether these official vote counts, given below, were altered.
In Detroit, Michigan, Clinton won among voters at the polls by 69.1% to 30.0%, and among absentee voters by 86.9% to 11.2%. In Oakland County, Michigan, Clinton lost among voters at the polls by 51.9% to 47.4%, but won among absentee voters by 67.9% to 25.5%. In Detroit, 21.05% of the ballots cast were absentee. In Oakland County, 19.18% of the ballots cast were absentee. The absentee ballots need to be audited.
In St. Louis, Missouri, Clinton won among voters at the polls by 54.2% to 45.2%, and among absentee voters by 71.4% to 25.2%. In Kansas City, Missouri, Clinton won among voters at the polls by 55.6% to 43.9%, and among absentee voters by 70.6% to 27.7%. In St. Louis, only 5.85% of the ballots cast were absentee. In Kansas City, only 3.57% of the ballots cast were absentee. But they did account for Clinton’s statewide margin of victory. The absentee ballots need to be audited.
In Hamilton County, Ohio, Clinton won among voters at the polls by 57.6% to 41.6%, and among absentee voters by 68.2% to 29.7%. In Cuyahoga County, Ohio, Clinton won among voters at the polls by 59.8% to 39.8%, and among absentee voters by 73.0% to 26.2%. Sanders got less than 10% of the absentee ballots in 52 of 336 precincts in Cleveland, and less than 20% of the absentee ballots in seven entire wards. As in Michigan and Missouri, these are very large disparities. The absentee ballots need to be audited.
All absentee ballots are cast on paper ballots marked by voters, even in states that otherwise have black box voting with no paper trail. Wherever a separate breakdown of absentee ballots at the precinct level is available, the absentee ballots can be audited. Wherever there are paper ballots at the polls, everything can be audited.
And that is the point of this study. As long as questions about the validity of Clinton’s primary victories are based almost entirely upon exit polls, we remain in the realm of speculation. Only by examining the forensic evidence — ballots, poll books, voter signature books, ballot accounting charts — can the official vote counts be proven true or false. That is what we did in Ohio. That is what public citizens need to do right now.
Instructions are available here: http://blackboxvoting.org/how-to-audit-an-election
Copies of “Witness to a Crime” can be obtained from the author: richardhayesphillips at gmail dot com