Democracy Lost: A Report on the Fatally Flawed 2016 Democratic Primaries – V. Evidence for Fraud or Suppression in EACH Primary Phase, State-by-State

V. Evidence for Fraud or Suppression in EACH Primary Phase, State-by-State


In the present section, we provide a full date-by-date, state-by-state breakdown of evidence for election fraud, voter suppression, or other irregularities documented throughout the course of the 2016 Democratic presidential primaries.

Based on this work, Election Justice USA has established an upper estimate of 184 pledged delegates lost by Senator Bernie Sanders as a consequence of specific irregularities and instances of fraud. Adding these delegates to Senator Sanders’ pledged delegate total and subtracting the same number from Hillary Clinton’s total would more than erase the 359 pledged delegate gap between the two candidates. Election Justice USA established the upper estimate through exit polling data, statistical analysis by precinct size, and attention to the details of Democratic proportional awarding of national delegates. Even small changes in vote shares in critical states like Massachusetts and New York could have substantially changed the media narrative surrounding the primaries in ways that would likely have had far reaching consequences for Senator Sanders’ campaign.

In what follows, we summarize the evidence on a state-by-state, date-by-date basis. Each state is introduced followed by a minimum and maximum estimate of the delegate cost to candidate Sanders. Each state name is followed by an acronym signifying the following fraud or suppression types:

  • Direct voter suppression (DVS)
  • Registration tampering (RT)
  • Illegal voter registration purges (RP)
  • Inaccurate voting machine counts (H; hacking)
  • Miscellaneous Forms of Election Fraud (M)

B. Early States

·Iowa (M) (2; 6)

Beginning with the first Democratic Caucus in Iowa on February 1, there were clear signs that the Democratic Party had no intention of playing above board in its quest to anoint insider favorite Hillary Clinton. These are not simply the observations of partisans of Bernie Sanders. The Des Moines Register‘s Editorial board was so disgusted with irregularities, some of which they witnessed with their own eyes, and the Democratic establishment’s refusal to respond to them transparently that it penned an editorial entitled: “Something Smells in the Democratic Party.”1It noted that the whole process produced a “whiff of impropriety,” called into question “the accuracy of the caucus results,” and found that the Democratic Party response “reeks of autocracy.”

Specifically, the Des Moines Register made the case that the results, at just a 2/10ths of percentage point win for Clinton, were “too close not to do a complete audit of results.” Party officials, however, refused to even entertain the possibility of releasing exact figures of reported votes. The details of the Registers complaints are worth quoting at length:

Second, too many questions have been raised. Too many accounts have arisen of inconsistent counts, untrained and overwhelmed volunteers, confused voters, cramped precinct locations, a lack of voter registration forms and other problems. Too many of us, including members of the Register editorial board who were observing caucuses, saw opportunities for error amid Monday night’s chaos.

The Sanders campaign is rechecking results on its own, going precinct by precinct, and is already finding inconsistencies, said Rania Batrice, a Sanders spokeswoman. The campaign seeks the math sheets or other paperwork that precinct chairs filled out and were supposed to return to the state party. They want to compare those documents to the results entered into a Microsoft app and sent to the party.

“Let’s compare notes. Let’s see if they match,” Batrice said Wednesday.

Dr. Andy McGuire, chairwoman of the Iowa Democratic Party, dug in her heels and said no. She said the three campaigns had representatives in a room in the hours after the caucuses and went over the discrepancies.

McGuire knows what’s at stake. Her actions only confirm the suspicions, wild as they might be, of Sanders supporters. Their candidate, after all, is opposed by the party establishment — and wasn’t even a Democrat a few months ago.

The kinds of complaints noted by the Register and other observers fit almost exclusively into the “miscellaneous” category of types of election fraud.In Election Justice USA’s database of 2000 complaints from various states shows just one, from Boone County, that may be indicative of registration tampering. Since Iowa caucuses allowed for same day regsistration, a widespread tampering campaign would have made no sense.

Based upon the number and kinds of complaints about the Iowa caucus along with initial full entrance polling by Edison Research (a three-point edge for Clinton), Election Justice USA fairest estimate is that Sanders may have won an additional one to three national delegates in Iowa if the show had not been run by an “autocratic” Democratic establishment beholden to Clinton. Clinton’s delegate total would have been reduced by a corresponding one to three delegates.

·New Hampshire (H?) (0; 6)

On February 9th, Sanders won a convincing victory (61.1% – 38.9%) over Clinton in the New Hampshire Democratic primary. When including Martin O’Malley, this win represented a 4.2% improvement for Sanders over the initial full exit poll projection. According to this predictor, Hillsborough County, New Hampshire’s largest, ought to show no irregularities in its CVS-CVT or precinct size data. A column chart, however, of Hillsborough shows a smooth and dramatic increase for Clinton, closing the gap between her and Sanders by 29.8% from smallest to largest township/ward.

Examining the income levels and racial demographics of the townships in Hillsborough gives us no insight that would suggest such an unexpected increase for Clinton.

Given smaller size of New Hampshire, though, and Sanders already sizable win, and given the way Dem proportional counting works, Election Justice USA has calculated that Sanders could have won the additional 8,000 or so votes it would take to keep Hillsborough from showing an irregular precinct size vote result without it affecting the overall delegate count.2 If, however, the same irregular results are present statewide and a reasonable explanation could be given for why the exit poll missed by a margin substantially outside the sample-size alone margin of error, Sanders may have picked up as many as three additional national delegates, with Clinton losing the same. Because of the discrepancy between the vote share by precinct size analysis and the exit poll, Election Justice USA has not shifted any national delegates to Sanders in our lower estimate of national delegate count.

Given Clinton’s unexpected big win there in 2008, New Hampshire remains fertile ground for further study of the vote share increase by precinct size increase pattern for particular candidates.

·Nevada (M; RT?) (2; 8)

The Nevada Democratic caucuses saw Senator Harry Reid and casino bosses rig the election in favor of Hillary Clinton over Bernie Sanders. This is neither hyperbole nor sour grapes from sore Sanders losers. The rigging was openly lauded inUSA Today by Jon Ralston – no friend of Sanders or his followers – as an example of Reid’s political prowess.3 Harry Reid would not have been able to help control the processes of caucuses and conventions in Nevada if he had publicly endorsed Hillary Clinton. Instead, Reid feigned neutrality in public while rushing home at the last minute in February to convince casino bosses to get their Culinary Unionworkers to caucus for Clinton, a move they were not planning prior to Reid’s intervention. Ralston calculates that Clinton won six Casino locations by healthy margins, all of which she had been previously expected to lose according to Ralston.

Casino bosses simply turning out the vote would not be problematic if there were not highly credible reports that they also unethically gamed the system in a way that inevitably raises questions of a potential quid pro quo with Reid, a possibility Ralston strongly hints at. Deborah Burger, Co-President of the National Nurses Union (NNU), was present in Clark County. Other NNU nurses were spread out across various casino caucus sites. A NNU press release after the “faux riot” at the Democratic state convention months later relays eye-witness testimony of Casino bosses overseeing a process where they selected who could have time off to caucus, provided selected workers with food, drink, and Hillary Clinton shirts and hats during the caucus along, and directly watched in person to see how those employees voted.

The entire scene amply demonstrates the ‘squash’ after Iowa and New Hampshire predicted by FiveThirtyEight’s Harry Enten or the “rigging” about which so many voters and outside observers have complained in the 2016 Democratic primaries and caucuses.

Election Justice USA read complaints about irregularities ahead of and at the State Convention that saw so much vitriol and false claims of violence by Sanders delegates and supporters. While we do not think it can be established to any degree of certainty that Democratic establishment antics ahead of and during the convention were responsible for lost national delegates for Bernie Sanders, an unconfirmed registration tampering report on social media report fits a pattern seen in several other states. According to the report, at least one of the delegates refused entry to the Nevada Democratic convention was deemed not to be a Democrat after his registration was switched without his knowledge or consent.4

Election Justice USA, primarily on the basis of Ralston’s analysis, estimates that Bernie Sanders would have picked up an additional one to four national delegates if Reid had not put his thumb on the scales in Clark County. Hillary Clinton’s national delegate count would have been reduced by a corresponding one to four delegates.

·South Carolina (4; 8)

Writing in The Nation, Joshua Holland has suggested that it makes no sense that Clinton would cheat in a contest she was clearly winning already.5 Besides massively begging the question and ignoring the historical example of Watergate and Richard Nixon, the argument ignores the contours of the primary calendar and when those arguing for rigging that benefitted Clinton believe it began. Heading into the February 27th vote, the pledged delegate count between Hillary Clinton and Bernie Sanders was an extremely close 51-50. Sanders’ virtual tie in Iowa, large victory in New Hampshire, and close running in Nevada put the Democratic establishment on notice that Sanders could be competitive for the long haul. South Carolina was the first Democratic primary to definitively turn the inaccurate machine count hat trick.

  • The first full exit poll missed by more than 10%

  • Voting machines in South Carolina are extremely hackable

  • Voting results by precinct in Richland, South Carolina’s largest county, are noticeably irregular

Theodore Soares, a co-author of this report, has calculated the sample-size only margin of error for South Carolina at 5.0% while the initial full exit poll missed by 15.3% in Clinton’s favor. Edison Research projected a 37.2% win for Clinton, very close to FiveThirtyEight’s polls plus projection of a 38.3% win for Clinton.6 The final result, however, shows a 47.5% win for Clinton, a 10.2% miss, more than double the expected margin of error. Notably, Edison’s initial projection did not miss outside the margin of error for black voters, as one would expect if misreading black preference for Clinton were responsible for the increased share of the vote by precinct size increase in the scatter plot chart above. Instead, Edison’s initial projection of substantial win with white voters gave way to an adjusted exit poll margin that favored Clinton with white voters.7

Since the exit poll missed by a large margin with white voters, most Democratic voters in Richland County are black, and racially polarized Mecklenburg County in South Carolina’s neighboring state directly to the North does not show this pattern, the burden of proof now lies with Clinton supporters for a non-fraudulent explanation of the irregular increase in Clinton’s share of the vote as precinct size increased. From the smallest twenty precincts in Richland County (130 or fewer total Democratic votes) to the largest twenty-six precincts (455 or more votes), Clinton’s gap over Sanders increased by a stunning 38.8%.

Based on Election Justice USA calculates based on initial exit polling projections, vote percentage in the change in the smallest versus largest precincts, and comparison to Democratic proportional assignment of delegates as viewable at the Green Papers blog, that inaccurate machine counts likely cost Bernie Sanders at least two more national delegates and perhaps as many as four in South Carolina, with Clinton having a corresponding number fewer delegates.

C. Super Tuesday (March 1)

·Alabama (H) (6; 8)

In May 2016, Doug Johnson Hatlem, one of the authors of this report, wrote a series of articles for CounterPunch on election fraud allegations. In pieces on exit polling and hackable voting machines, he zeroed in on Alabama as a Democratic primary state where the exit polling miss of fourteen percentage points simply could not be explained away by available non-fraudulent theories. The articles pointed out that Clinton did especially well in three of the four largest counties in Alabama that used Model 100 vote tabulators as discussed in §II.D.3 as extremely hackable.

In those articles, and before having encountered the irregular precinct increase patter in favor of Clinton, Johnson Hatlem theorized that Clinton did especially well due to inaccurate machine counts in Jefferson, Mobile, and Montgomery counties, while suggesting that, if there was a way of testing the theory, Madison and Tuscaloosa counties would be clean. Madison is the second largest county by population in Alabama and Tuscaloosa was the fifth largest in terms of Democratic primary votes. Jefferson is the largest county in Alabama; Mobile and Montgomery round out the top five.

The irregular increase for Clinton by precinct size increase pattern, in fact, shows up as predicted in Jefferson (the largest county) and in Mobile. The irregular increase does not show up, also as predicted, in Madison and Tuscaloosa. In Montgomery, the pattern likely does not appear as expected, though there are some complicating factors for analysis, and the irregular pattern may show in weaker form.

Both Madison and Tuscaloosa show the recognizable “fishtail” pattern where an otherwise relatively flat, parallel lined pattern (clean) jumps for Clinton in the final precinct grouping. In several otherwise clean states or counties, this fishtail pattern shows rather than a steady increase by precinct grouping for Clinton and can in every circumstance be attributed to a findable cluster of very large precincts with heavy elderly or African American populations or registrations. In Madison, for instance, the second largest precinct by total vote for Clinton plus Sanders is at the predominantly African American Bob Harrison Senior Center.

Jefferson County and Mobile County, with the easily corruptible machines, show the radically different, irregular pattern. Jefferson saw a 53.4% increase in Clinton’s spread between the smallest fourteen precincts and the largest ten. Mobile saw a 48% increase for Clinton from the smallest twenty-nine precincts (0-50 total votes) to the largest (451+ total votes) sixteen.

Montgomery County, the very largest black county by population percentage in Alabama, meanwhile, shows a pattern that is difficult to call either totally clean or definitively irregular.

If placed in a scatter plot chart, it the pattern appears in weak form with about a seven or eight percent increase for Clinton from smallest to largest. The line chart is more revealing, however. It could almost be considered to fit the parabola pattern (clean) where Clinton does somewhat better in richer, whiter small precincts and somewhat better in the very largest, heavily black precincts with the middle precinct groupings showing a regular flat or even structure. There appears, however, to be a clean break at around two hundred votes with improving her gap by 22% from the 150-200 grouping of precincts to the 501+ votes grouping.

Election Justice USA has seen no public indications of voter registration tampering or illegal purges in Alabama for the Democraticprimary cycle and has likewise received no specific voter complaints from our online, fillable complaint form.

While Montgomery, and even Alabama as a whole, requires further investigation, the clear existence of the dramatic, irregular vote share increase by precinct size increase in two of the largest three counties in Alabama with provably hackable vote tabulators combined with the otherwise unexplained fourteen point exit polling miss, leads Election Justice USA to a low end estimated vote loss for Bernie Sanders of three national delegates (in Congressional District 1, the PLEO and At-Large categories if you are following along via Green Papers8) and a higher end estimation of four lost delegates (adding a possible delegate pick-up in Congressional District 2), with Clinton estimated to have gained a corresponding number of delegates.

·Georgia (H) (14; 20)

The entire state of Georgia votes on electronic machines without a voter verified paper trail. Georgia’s vote was held in an open primary and Election Justice USA received no complaints about registration tampering from Georgia. As noted and charted in an earlier section of this report on suspicious precinct size vote share increases, Clinton outperformed exit polling projections in Georgia by 12.2%, and on election day voting swung in Clinton’s favor by 11.4% from smallest to largest precincts and in early voting, which made up 22% of the overall vote in Georgia, by 23.0%. Election Justice USA estimates a seven to ten national delegate loss and an equal pick-up for Clinton based on inaccurate machine counts

·Tennessee (H) (6; 14)

Hillary Clinton outperformed exit polling expectations without explanation in Tennessee by 8.3% or 2.3% outside the margin of error as calculated by sample size. With the difference between Memphis and its somewhat whiter suburbs, Shelby County, Tennessee presents a good example of how Election Justice USA has been able to control for race in analyzing irregular patterns in Clinton’s vote totals.

Nearly all of Tennessee, including Shelby County, votes using highly hackable AccuVote machines without a voter verified paper trail. Originally manufactured by Diebold, the machines are now sold by Premier.

Shelby County as a whole shows a 37.4% growth in gap over Sanders for Clinton from the smallest seventeen precincts (200 total votes or fewer) to the largest 15 precincts (900 votes or more). But it could be counter-argued, reasonably, that just one of the smallest 17 precincts is in Memphis (63.3% Black) while all but three of the largest precincts are within Memphis.

But does racial polarization tell the whole story of the irregular precinct size increase pattern?

The chart on the left shows Shelby County as a whole, while the chart on the right separates out just the Memphis precincts. There is a less dramatic, irregular increase pattern, but the pattern remains and is well above the 10% threshold Election Justice USA is using as a relevant marker. From the smallest thirteen precincts (300 or fewer votes) in Memphis alone to the largest 12 precincts (900 or more votes), Clinton still shows an increase of eighteen percentage points in her gap over Sanders. Controlling for race, as in Louisiana, the Bronx, Mecklenburg County, North Carolina and more does not explain away the irregular increase in Clinton’s vote share by percentage as precinct size increases. Tennessee has extremely hackable voting machines, exit polls missed outside the expected margin of error, and precinct size turns out to be the most outstanding factor in how much Clinton won by.

Election Justice USA did not receive any complaints of registration tampering in Tennesse but estimates that electronic rigging of voting machines moved between three and seven national delegates from Bernie Sanders column to Hillary Clinton’s.

D. March 2 – March 15

·Louisiana (H, RT) (4; 10)

While Edison Research did not conduct exit polls in Louisiana, the state uses all electronic voting machines that have no paper trail and are extremely hackable and, as discussed and graphed in the precinct size §II.D.2, the irregular pattern shows up dramatically in East Baton Rouge, Louisiana and throughout (and statewide) as in other Southern states such as Alabama, Georgia, South Carolina, and Tennessee.

Assuming inaccurate machine counts between 8% and 15% as in those states yields an estimate of between two and five delegates moved from Bernie Sanders to Hillary Clinton.

Louisiana is where Election Justice USA has been able to control most assiduously for race as we were able to obtain registration data by part and race for every precinct in the state and rigorously compare the data to a clean state, North Carolina, where we also obtained precinct level registration data by part and race for Mecklenburg County. Time and again, our modeling shows that while race and precinct size and race and Clinton’s share of the vote are related in both Louisiana and North Carolina, only in Louisiana is precinct size a much more determinative factor on a statistical level.

Regression analysis is available for those with statistical training. Nicholas Bauer, who was the first to suggest and research taking cumulative vote share analysis down to the county and neighborhood level, has created a series of twenty powerful charts for East Baton Rouge and for comparing East Baton Rouge to Mecklenburg County (Charlotte), North Carolina.

All precincts within East Baton Rouge that reported at least one vote, were lumped into 1 of 10 blocks based on %White registered Democrats in that precinct (0-10% White, 10-20% White, etc.). These precinct results contained in these 10 blocks were sorted by Clinton’s vote share (smallest to largest).

When examining the first chart (ORANGE outline) on the next page, the BLUE dots represent Clinton’s reported vote share, and the BLUE dotted line represents a linear parabolic trend line. This model suggests that the output (i.e. vote share) can be almost entirely (or 99.09%) explained by race alone. Although these results were smoothed since each dot representsresults from multiple precincts and in some cases 1000s of voters, a 99% R2 value is a bit concerning whenever human-behavior is introduced.

The ORANGE dots represent Clinton’s vote share without the “fish tails” (marked by orange boxes) that were observed in 8 out of the 10 racial blocks. For instance, in the 90-100% White Registered Democrats block, one precinct with 90 total ballots cast went for Clinton with about 94% of the vote. This one precinct (circled in ORANGE), increased Clinton’s vote share in this block by nearly 5% and remain in line with the parabolic trend line.

The chart outlined in GREEN, represents the Clinton’s best performing precincts in East Baton Rouge Parish. In these 13 precincts, which represented 2,280 voters, Clinton’s vote share ranged from 95% – 97%. Clinton received 2,275 votes to Sanders 105 votes, or about a 20:1 lead in Clinton’s favor.

13 PRECINCTS 2,280 votes total CLINTON 2,275 votes (95.6%) SANDERS 105 votes (4.4%)


When comparing the results from East Baton Rouge Parish, LA and Mecklenburg County, NC, there is a stark difference in racial tendencies. In Mecklenburg, Berniecarried about 25% of the vote but less than 10% in East Baton Rouge when examining only the precincts with less than 10% white Democrats:

But probably the most concerning comparison between North Carolina and Louisiana is examining turnout rates for the same 10 racial blocks used in the above comparisons. In Mecklenburg County, turnout skyrocketed as the%white registered Democrats increased. While on the other hand, Louisiana’s turnout by race is nearly a flat line. The nearly even turnout by race in Louisiana would explain the tight correlation between race and vote share, but this would not explain the extreme difference in relative turnout rates from state to state.

·Mississippi (H) (6; 8)

In Hinds County, Mississippi’s largest by population, Clinton’s share of the vote increased by 5.1%, or 10.2% swing, from smallest to largest precinct groupings. Clinton beat exit polling by Edison by 9.9%, 4.7% above the sample-size calculated margin of error.

Nearly every county in Mississippi uses electronic voting machines that are highly-susceptible to tampering and do not have a voter verified paper trail. Hinds County, however, is an exception. It uses the DS200 model to count paper ballots from Election Systems and Software, the same model used throughout New York City. This report discusses the DS200’s vulnerabilities in more depth in §II.D.3 (hackable machines) and in §III.F (New York).

In Congressional District 2 (CD 2) in Mississippi, Bernie Sanders was awarded no national delegates while Clinton was awarded nine as inaccurate machine counts pushed Sanders below the 15% viability threshold. On the basis of showing all three factors (divergence from exit polling outside the margin of error, demonstrably hackable machines, and an irregular precinct size increase pattern), Election Justice USA estimates that Sanders should have won at least two delegates in CD 2 and either three or four more delegates total in Mississippi, with Clinton’s earned total reduced by the same.

Election Justice USA received just one complaint in our database of 2000 registration complaints. It was from a young voter who believed he had registered through an on-campus registration drive in 2015 but discovered he had not been registered on election day.

·Illinois (H, M) (6; 20)

Cook County (Chicago), Illinois presents several unique features for analysis in terms of inaccuratevoting machine counts. Clinton beat exit polling in Illinois by Edison Research by 4.1% putting the state into a category of states that could be considered clean or dirty based on exit polling data alone. Cook County Suburban and Chicago have separate Boards of Election (BOE), but both use the provably quite dangerously hackable AVC Edge II voting machines for early voting. Furthermore, Illinois requires a 5% audit of election results for each political contest. As noted in the section on lawsuits, citizens’ groups launched a lawsuit on Thursday July 21, 2016 (the same week this report was being concluded) after observing the audit in Chicago and finding it to be wholly inadequate, especially with respect to early balloting. Several of the groups’ members gave sworn testimony, publicly available online, at a meeting of the Chicago BOE shortly after the audit. A spokesperson for the Board admitted many of the points thecitizens’ groups have made about the audit to one of the authors of this report by phone.9

Chicago’s early balloting by precinct may show the irregular pattern somewhat weakly and a bit unusually in terms of the way the pattern usually works. Clinton’s vote share increases from beginning to middle precincts by 7.4% then drops back down a bit below the smallest precincts before rising again by 12.6% gradually from middle to largest precincts. Overall, Clinton’s share of the early vote in Chicago rises by 9.2%. It is unclear whether this represents an irregular pattern at all since it falls below the 10% threshold.

Early balloting in Cook County Suburban (82% white), however, unmistakably shows the irregular pattern.

Sanders is shown in the smaller precincts with a near 50:50 split in Suburban Cook, but gradually tapers off to losing the largest precincts by 17.2%, a 34.4% swing.

On the basis of some uncertainty in the midst of a clear case of the irregular pattern that cannot be dismissed by racial polarization or other factors as well as on the basis of many Congressional districts having been narrowly lost by Sanders, Election Justice USA’s lower and upper estimates for Illinois are quite stark in range, a six national delegate swing in Sanders favor at the low end, a twenty delgate swing at the high end.

·Missouri (H) (8; 12)

St Louis City and St Louis County, MO are interesting cases. Along with Detroit and Milwaukee, St Louis represents a very racially divided city. Page Boulevard divides the city into two distinct areas, white residents to the south of Page Boulevard (with the exception of a small area between Delmar Blvd and Page Blvd) and black residents to the north of this divide.

Although our model above did not indicate an increase in vote share, based the racial division in St Louis, precincts were divided up based on location. Furthermore, this in depth analysis of racial demographics for each section of St Louis City, indicated the same pattern of vote share divergence in Clinton’s favor as precinct size increased in St Louis(regardless of race). The table below represents polling locations north of Page Blvd (racial divide) which housed voting booths for at least one small precinct (less than 165 ballots) and one large precinct. In 6 out of these 7 polling locations, Sanders’ small precinct vote share is significantly greater than his large precinct vote share. On average, Sanders outperforms by 8.3%, or a 16.6% swing.

·Ohio (RT, H) (8; 14)

Charts for Ohio’s early balloting and election day balloting were included in §II.D.2. In short, Cuyahoga (Cleveland), Ohio shows a 12.4% discrepancy from smallest to largest precincts. The discrepancyfavors Clinton on election day only totals with a clean bill of health for early balloting in the same. Clinton outperformed initial full exit polling by 10% (and perhaps more as Theodore Soares argues in the exit poll section of this report).

Election Justice USA’s estimate is that Bernie Sanders lost four to seven national pledged delegates from Ohio with Clinton picking up the same without truly having earned those votes.

E. Arizona10

·Arizona (DVS, RT) (4; 12)

Arizona is the one primary state out of thirty-six for which Election Justice USA could not obtain precinct level results for its largest county, Maricopa. An executive assistant for Maricopa County Recorder Helen Purcell stated that they were no longer keeping track of precinct level votes after dramatically slashing the number of polling stations, as discussed in §II.A.1 of this report. Direct voter suppression in the form of closed polling station, lack of precinct level results, and no exit polling make it more difficult to estimate the impact of election fraud on delegate totals for Hillary Clinton and Bernie Sanders, but it is clear that registration tampering happened on a previously unseen scale.

Election Justice USA reviewed each individual record in a 151person database collected and published by the online collective Anonymous.11The beginnings of Election Justice USA saw one of our current board members independently collect a database of more than fifty registration tampering complaints, very few of which we found to overlap with the Anonymous database. Our analysis of each entry in the Anonymous database and the early Election Justice USA database confirms that the primary target for registrations switched without the knowledge or consent of the intended voter overwhelmingly affected supporters of Bernie Sanders. Anonymous found that registration switching affected the following: 1 Unknown Party, 12 GOP, 139 Dem, 113 Sanders supporters, 24 Unknown Preference, 2 Clinton. These numbers are highly consistent with our analysis of more than 200 records.

While Southwest Voter Education Registration Project’s Antonio Gonzalez thinks the major issue in Arizona was voter suppression “consciously, or unconsciously” by Republican state officials, he is very clear about what the remedy should be:“In my view, the Arizona primary, the Democratic primary, the Republican primary, should be thrown out and a new election should be held.”According to Gonzalez, “the suppression of tens of thousands if not hundreds of thousands of Arizona voters” was a violation of the Voting Rights Act as it had a “disproportionate effect on Latino” voters. “Typically, Latino voters vote more heavily on election day” rather than in early or absentee balloting. Seventy percent of Latino voters in the state are in Maricopa county, according to Gonzalez, and the Democratic (and Republican) party should not seat delegates from Arizona: “Those delegates should be thrown out, the awarded delegates based on a fraudulent vote should not be allowed. Either that delegation should be disqualified in total at the party convention or you have to have a new election.”

While registration switching affected voters all over Arizona, Maricopa County was hardest hit in combination with a reduction in the number of polling places. There were more than 20,000 uncounted provisional ballots in Maricopa alone.12 As the Anonymous report puts it quite well:

The lack of polling stations alone in Maricopa County cannot explain how Phoenix, with a Democratic Mayor, could see Republicans show up at the polls on election day to the tune of around 80,000 voters, while Democrats cast a paltry 33,000 votes in Maricopa County on election day.While Early Voting is increasing in Arizona with each election, we are rather skeptical of the idea that Maricopa had nearly 100,000 fewer Democrats voting in-person over against 2008’s primary between Clinton and Obama.

The argument made by Clinton’s campaign manager on Reddit that these issues affected Clinton’s supporters as much as Sanders supporters does not hold water. Apart from solid analysis of registration tampering complaints primarily affecting Sanders supporters, early voting plays a heavy role in Arizona. Everywhere else that was the case, Clinton won early voting handily with early voters skewing much older.

Election Justice USA agrees with estimates of the likelihood that 50,000 to 250,000 Democratic voters, largely Sanders supporters, were disenfranchised between direct voter suppression and registration tampering in Arizona. On that basis, and in consultation with Green Paper records13 for Democratic delegates in Arizona, estimates that election fraud cost Bernie Sanders between two and six national delegates, with Clinton’s total inflated by the same amount.

F. New York

·NY (RT, RP, H) (16; 28)

G. April 26

·Maryland (RT, H) (0; 6)

H. May Contests

·Indiana (H, RT) (6; 8)

·Kentucky (RT, H) (2; 4)

I. June Contests

·Puerto Rico (DVS) (10; 14)

·California (DVS, RT, H)(40; 70)

On May 31st the Wall Street Journal published a piece entitled “Clinton Might Not Be the Nominee: A Sanders win in California would turbocharge the mounting Democratic unease about her viability.”14 While most commentators disagreed, it was a legitimate argument that appears to have precipitated the same full scale effort to suppress votes from Sanders supporters and electronically manipulate voting on an industrial scale.

A previous section of this report details six different ways California elections officials engaged in voter suppression targeting no-party-preference (NPP) voters. Our online complaint form saw Election Justice USA receive testimony and affidavits from over 700 California voters who experienced voting and registration problems. Of these respondents, 84 were switched to another party without their knowledge or consent. In some cases, these changes were back-dated such that they were listed as made before the voter initially registered.

1) We have testimony from CA voters who were given provisional ballots by pollworkers despite their names being on the Democratic voter rolls; 2) We have testimony from CA Democratic voters who received the wrong ballot type in the mail; 3) We have testimony/video evidence from CA Democratic voters who were given provisional ballots instead of being directed to a recently-changed polling location; 4) We have testimony from poll inspectors about a shortage of ballots: in some cases, fewer than 39% of registered voters would have been covered by the amount of ballots provided for Los Angeles County precincts, despite a CA State Law requirement that 75% coverage be guaranteed. We also have testimony from voters who were forced to use provisional ballots due to ballot shortages; 5) We have testimony from poll workers did not count or keep a roster of provisional ballots in CA, hence no chain of custody is possible.