In Depth Data Analysis

Where we get the data

NY Citizens Audit has acquired data from multiple sources. These include official government websites, such as the census bureau, Secretary of State, and online voting regulation guides. NY Citizens Audit has also made numerous Freedom of Information Law (FOIL) requests to all 62 NY counties and the state Board of Elections (BOE). Some of those requests were ignored, others were partially satisfied, and others were complied with in full. In addition, citizens and candidates for public office have come forward with election-related information. Some of these persons have signed affidavits attesting to suspicious or illegal activity that they witnessed. All of the information acquired by NYCA is publicly available and may be disclosed publicly.

Discrepancies Between Official Sources

Down-ballot and Unopposed Races

Because of the massive discrepancies between official sources regarding NY 2020GE voting, New Yorkers are facing a crisis in governance. When the discrepant counts outnumber the margin of victory in a particular race, the outcome is no longer known.

A majority of our state legislators cannot prove they won, due to this real conundrum. When the discrepancies are added to illegal votes cast within the district in question, more and more races are affected. If the margin of victory was small, the discrepancies needn’t be very large to impact the legitimacy of the contest results.

In order to publish final results about which politicians are impacted, NY Citizens Audit is working to analyze outcomes by district. Until that work is completed, we can only say with confidence that this affects over 50% of our state legislature.

Adding to the mess, the NY GOP is implicated in a huge betrayal of the people it claims to represent: NY voters registered to the Republican party. In just 6 counties–Bronx, Kings, New York, Queens, Richmond and Westchester–the NY GOP failed to mount an opposing candidate in FORTY-SIX (46) state legislative races, effectively ceding control of 25% of law-making power to a radical progressive agenda that does not represent the majority viewpoint.

Republican candidates in other state legislative races, who were leading significantly on election night, received no support from the NY GOP when they were denied their right to contest the avalanche of mail-in ballots that swept away their victories. One received death threats until she finally conceded. All told, another THIRTY (30) state legislative races in Kings and Queens counties alone went uncontested to radical progressives.

When the discrepancies and the violation of public trust (not to mention the donations gleefully solicited by the NY GOP) are combined, over 70% of our state legislature is impacted. NY Citizens Audit looks forward to publishing the specific races as soon as we can be confident in the accuracy of that report.

Dr. Frank Replication Analysis

Learn about voter analysis using the Dr. Frank model.

The first analysis made by the Statistics team was a replication of the Dr. Frank analysis (described separately in NY Citizens Audit report #001). The purpose of the analysis was to determine whether voter turnout by age could be predicted in every county with a statistically improbable degree of accuracy. To do this, the team created two polynomial equations based on voter registrations, one for New York City counties (n=5), the other for all remaining counties (n=57). Queens County was the “base” county for NYC counties. Albany County was the base county for all other counties in NY State. The equations were used to generate “prediction curves” or “keys”. The keys were applied to county registrations to predict voter turnout by age in 60 counties other than the two used to generate the keys (Fig. 1 and Fig. 2).

Figure 1 Queens county prediction “key” used to predict voter turnout by age in all remaining NYC counties
Figure 2 Albany county prediction “key” used to predict voter turnout by age in all counties outside of NYC

The product of the keys was a table of predictions that was applied to all other counties. The table consisted of a list of registered voter ages between 18-100, in one year increments, and a percentage of those registered voters predicted to vote by the prediction key (Fig. 3).

Age Queens predicted turnout by age Albany Predicted turnout by age Difference Albany to Queens
30 51.87% 55.74% 3.87%
31 51.29% 56.46% 5.17%
32 50.80% 57.34% 6.54%
33 50.41% 58.35% 7.94%
34 50.11% 59.46% 9.35%
35 49.90% 60.63% 10.73%
36 49.77% 61.84% 12.07%
37 49.72% 63.06% 13.35%
38 49.74% 64.29% 14.55%
39 49.84% 65.51% 15.67%
40 50.00% 66.70% 16.70%

Figure 3 predicted voter turnout by age extracted from two base keys for Albany and Queens

To determine correlation between predicted turnout and actual turnout, charts were generated with curves representing predicted and actual turnout (Fig. 4). The curves were compared for similarity, expressed as an “r-value”. An r-value of 1.00 (100%) denotes full agreement between the curves. A value of 0.00 (0%) denotes no similarity between the curves. An r-value of -1.00 (-100%) is an exact reverse match. Across all counties, the average correlation is .993 (99.3%). The range of values for all 60 non-key counties is between r=.934 to r=.999. These are very high r-values for a prediction involving human behavior

Cayuga County chart
Figure 4 Correlation graph for Cayuga County; yellow is predicted turnout, green is actual turnout

This analysis showed that voter turnout by age could be predicted with a statistically high degree of correlation in all 60 counties compared. This finding is true for every age from 18-100. The agreement of predicted and actual turnout for each of the 83 ages compared amounts to an 83 number combination lock, where every county has the same combination.

Critics have claimed that the Dr. Frank analysis is flawed because there is a natural explanation for his findings. That is, just as office workers may be expected to visit delicatessens more often at the lunch hour, voter behavior is also governed by such a constraint(s). The most visible and likely of such constraints is voter registration by age. If registration can be used to estimate the enthusiasm of voters, actual turnout should be similar. To an extent, this is true. However, NY Citizens Audit found that registrations did not follow the population curve for each specific county. This is due to registration numbers that deviate significantly from county specific population counts. Because actual turnout follows percentages derived from false population numbers, the natural explanation argument fails.

Figure 5 Hamilton county registrations compared to national and county census population shows 121% registration rate

The Dr. Frank replication is significant because the results are inconsistent with natural human behavior. The alternative is that the high level of correlation is artificial. There is considerable evidence in the voter rolls and elsewhere to indicate that the rolls were artificially inflated with false registrations, and those registrations were used to cast votes illegally.

Excess Registration

Learn about the number of voter registrations that are illegal.

Excess registrations are registrations that exceed the population of their respective counties. The numbers provided for this category are estimates rather than fixed values for two reasons. First, population figures published by the census are estimates themselves. Second, the “actual” registration rate is unknown. “Actual registration” is the number of genuine legal voters who have registered in a county, as opposed to recorded registrations that may not reflect the true number. We know that the recorded number cannot be true when registrations exceed population, thus casting doubt on remaining registrations.

In a county where overall registration exceeds 100% of the population, all registrants over 100% are counted as “excess”. Similarly, even if overall county registration does not exceed total population, in many counties, registration by age does exceed population figures for that age. The amount by which registrations exceed 100% in this category are also counted as “excess” (Fig. 7).

It is unlikely that a registration rate of 100% represents a true picture of legal registrations in any county for any age. However, the actual value cannot be determined without a canvass. For that reason, while high registration rates at or slightly below 100% of population are suspicious, they are not counted as irregular until further investigation has taken place.

Age NatCensus Population Population Registered Votes Registered percent population Excess registrations
39 12181.007 10954.1 10,954 17,282 6,800 157.77% 6,328
40 12398.207 11077.53 11,078 17,131 6,783 154.65% 6,053
41 11614.438 10005.05 10,005 16,753 6,531 167.45% 6,748
42 11373.795 10044.17 10,044 16,675 6,201 166.02% 6,631
43 11245.033 9360.117 9,360 16,879 6,419 180.33% 7,519
44 10903.555 9382.183 9,382 15,765 5,945 168.03% 6,383
45 11223.911 9812.475 9,812 16,182 6,174 164.91% 6,370
46 10857.919 9011.464 9,011 16,020 6,248 177.77% 7,009
47 11003.514 9682.375 9,682 15,738 6,296 162.54% 6,056
48 11480.691 10326.96 10,327 16,298 6,657 157.82% 5,971
99656.424 99,655 164,723 64,054 165.29% 65,068

Figure 7 Example of excess registrations by age from Erie county, NY

Multiple SBOEIDs & CIDs

A State Board of Elections voter ID (SBOEID) is distinct from a county-specific voter ID (CID). “Excess” or cloned SBOEIDs are SBOEID numbers assigned to a person in excess of the one they are allowed to have. For instance, if one person, identified by first name, last name, and date of birth, has two unique SBOEIDs, that person has one legitimate SBOEID and one cloned SBOEID. Voter records that contain excess SBOEIDs tend to have the same registration date (Fig. 1).

Figure 1 Example of excess SBOEIDs
Figure 1 Example of excess SBOEIDs

County-provided voter IDs (CIDs) are assigned by some counties in addition to the SBOEID assigned by the state. Records for voters who have excess CIDs will display one SBOEID and multiple CIDs (Fig. 2). Unlike cloned SBOEID registrations, it is possible that an innocent explanation for excess CIDs exists. In the example below, the voter record shows the voter moving back and forth between two counties 6 times in one 12 month period. Every time he switches counties, he picks up another CID. When he returns to his original county, he is not reassigned to his previous CID, even when he is moving to the same address. There are approximately 1,300,000 records with excess CID numbers. NYCA considers these innocent for now, though it is troubling that the same voter can get different CID numbers when moving back and forth between the same two counties and even to the same 2 addresses, within a limited span of time.

Figure 2 Excess CIDs, Seneca and Steuben counties
Figure 2 Excess CIDs, Seneca and Steuben counties

Purged with no Purge Date

When a voter is “purged” from the voter rolls, their record is not deleted. Instead, the voter is marked as “inactive” and “purged”. An inactive or purged voter may not legally vote. The purge date distinguishes when votes cast by the registrant are legal from those that are not. Without the date, inactive or purged status has no meaning. A purge date of 11/02/2020 would render any votes after that date, including for the 2020 GE, invalid. A purge date of 11/04/2020 would allow a vote in the same election.

Most but not all records in the NYBOE voter rolls have purge dates for purged voters. On closer examination, it appears that some counties, such as Westchester (Table 1) and Wyoming, did not provide purge dates to the NYBOE. Other counties, like Onondaga and New York (Table 2) provided purge dates. The implication is that absent purge dates reflect either a decision made at the county level to deny the information to the state, or that the information is not collected or retained. Whatever the answer is, the importance of this data and the number of records without purge dates is too substantial to ignore. Without a date, purged status has no meaning in a retrospective investigation of past elections. Like other suspicious records, alteration or deletion of purge dates can allow for the creation or use of phantom voters. If this happened, it violated federal document retention law.

Table 1 Example purged voters without a purge date, Westchester County
Table 1 Example purged voters without a purge date, Westchester County
Table 2 Example purged voters with purge date, New York County
Table 2 Example purged voters with purge date, New York County

Age Discrepancies and Deceased Voters

Voters that are too young to legally vote or too old to be alive are “age discrepancies”. An underage voter is defined as under the age of 18, based on birth date. Some “underage” voters cannot possibly have been born yet because their birth dates are as far in the future as the years 9182 and 4950. The majority of age discrepancies are “overage” voters, defined as people whose age exceeds the oldest living man and woman known to be alive in the United States. The majority of these records exceed the oldest living person in the world by decades, with ages over 170 years old.

The explanation given for some of these records, those that share the birth date of January 1st, 1850, is that they are “default entries”. The idea is that if a birth date is unknown at the time of registration, it can be confirmed later. In the meantime, a default value is used to complete the registration. By using a birth date that cannot be genuine, they can be easily segregated from other records. However, if this practice is real, it purposely introduces false information into the voter rolls and creates (or should create) confusion if any of these voters turn up to vote before the information is corrected. If voters are allowed to vote on the basis of a false record of their birth date, and we know they have, one significant prong of the state’s test of valid identity is corrupted and meaningless.

Other overage discrepancies exist, with a variety of birth dates that make them older than any known living human but not as old as those that share the January 1st, 1850 birth date. The “default value” explanation does not apply to those records.

Table 2 Example overage voters from Monroe county
Table 2 Example overage voters from Monroe county

Blank Address

New York Election Law

§5-204 Local registration; General Provisions for the Conduct of

  • If the applicant’s name does not appear on the list of registered voters and if the applicant is not challenged, and he is found by the inspectors of election to be otherwise qualified, they shall complete his registration as provided herein.
  • If the person’s name appears on the list of registered voters and he is residing at the same address as set forth therein, his registration shall be refused as unnecessary.

The NY state BOE voter roll contains a large number of voter records that do not include a valid address. These are called “blank address” registrations because one or more essential address fields are blank, or empty, in the record. The current count of these is based on missing street names but it is known that some addresses lack other essential information like street numbers and apartment numbers (Fig. 8).

Figure 8 Example blank address registrations, from Onondaga
Figure 8 Example blank address registrations, from Onondaga

Secretary of State Revisions

Video transcript

Usually when the bad guy posts on their website, “I robbed a bank and this is how much money I stole”, you might want to think, “Either this guy is really dumb or maybe he’s kidding or something like that but this is a screen capture of a page from something that is published on the Secretary of State’s website. So because its published there I’m a little cautious of saying that they are admitting openly that they did a bad thing here. So maybe there’s an innocent explanation for it. But what this is, this is a revision history for the certified vote count. So what it tells us is they changed the vote count 26 times after they certified it.

Okay so to me what that that means is we [they] can’t count. And that means also I can’t trust them. Which means to me, “Why did you certify this and how did I did I let you certify this”? Okay for the most part of the revision dates, notice the first one starts like immediately after certification but it goes all the way to June 18th of the following year. They were monkeying with the counts for quite a while after the election. If you look at the numbers you see “Blanks +6409”. Go to row 27 you see “Blanks 305,904”. These blanks and voids that we see recorded in this document are highly suspicious for a number of reasons, however this is something where we need some expert advice on so we fully understand this. But we do understand that it is possible to utilize both blanks and voids to manipulate the vote count. Okay so what do you think you would see if they were manipulating the vote count and they needed to cover their tracks? They would convert ballots into blanks and voids. Does that make sense to you guys? And then they would have to add them to their revision history so they can account for all this stuff. So in total in Westchester if you add up all the blanks they added, it actually amounts to more blanks in Westchester County, for that election, than they have people in that county. Okay it was something like 544,000 blanks that they have opposed to, I think, 486,000 people who voted, something like that.

Now what is a blank? A blank is a ballot that they print prior to going to the pole station so that if somebody comes in there and they register on the spot they have to be given a ballot to vote with right? So if you’re going to do this, do you print more ballots than people you have registered in your county? Oh wait a minute, this is Westchester, they’re one of the counties that went above the 100% line. So maybe they were thinking the fake people where going to show up. In any event it’s highly suspicious.

Threshold of Materiality

The number of irregular votes discovered to date is important to determining whether they would have had a “material” effect on the election outcome. The standard of materiality promoted in the media and in some courts is that if the number of irregular votes in question is below the margin of victory for the winning candidate, the irregular votes discovered are “not material”. For instance, a judge in Georgia’s Supreme Court upheld a lower court’s ruling that fraudulent votes discovered in an election for probate court judge were not material to the outcome of the race because the margin of victory was greater than the number of known fraudulent votes. Stated simply, the standard for materiality expressed in that case was a number of votes that exceeded the margin of victory.

The voter rolls provided to NY Citizens Audit by the state and some counties postdate the 11/03/2020 election by many months and likely differ from the original rolls. Despite this, they contain enough data to demonstrate that large numbers of irregular votes were cast in that election. The numbers involved are sufficient to affect the outcomes of dozens of down ballot races that have been investigated to date. On that basis, NY Citizens Audit believes that the “exceeds the margin of victory” standard can be met in many down ballot elections.

However, this would be an unfair standard in the context of a citizen-led effort that has to date had no access to the ballots themselves. The voter rolls contain many hundreds of thousands of unexplained irregularities distributed throughout New York’s 62 counties. These are compelling on their own but suggest that an audit would uncover much more.

Albany County’s CIS officer Michael Chen stated in a written communication to NYCA that Albany County has not retained or preserved voter rolls as they were on Election Day. This violates Title 52 § 20701 requirements that election records be preserved and raises the possibility that other counties, perhaps those who refused to comply with lawful FOIL requests, have also violated Title 52 § 20701. For this reason, NY Citizens Audit intends to demand proof from every NY County that they have followed the law by preserving 2020 election records. There is no point in asking for an audit if the records no longer exist in their original state. Given the amount and type of irregularities discovered so far, there is no reasonable expectation that state or county officials are behaving in good faith.

If it can be shown that any county or the state cannot provide proof of records retention on request, the affected elections must be decertified to be in compliance with the law. The standard of materiality in accounting and finance is called the “materiality threshold”. Broadly stated, it is a quantitative and qualitative measure designed to capture misstatements on official documents that would likely affect decisions made by interested parties who rely on those statements. If a misstatement concerns 5% or less of a company’s net income, it is considered not material. If it concerns 10% or more, it is considered material. For amounts between 5%-10%, it is a judgment call for the auditor. However, according to the Securities and Exchange Commission’s guidelines, any misstatement is material if “the misstatement involves concealment of an unlawful transaction.”

The reason an intentional misstatement designed to conceal or commit a crime is material is that any such statement must be reported to the SEC or other governing authorities. Intentional misstatements by officials responsible for a company’s financial records call into question the honesty and accuracy of those records. This situation is materially different from the case decided in Georgia. In that case, seven individuals decided on their own to vote twice in the election. Their criminal behavior was not inspired, encouraged, directed, condoned, or concealed by election officials. For that reason, their actions did not reflect on the overall honesty of election officials, apart from exposing weaknesses in fraud prevention.

The type of irregularities found in New York State do not appear to be the product of individual instances of voter fraud. Initial investigation of voter rolls, and lack of compliance from some county officials, suggests that election officials have created systemic fraud within NY elections for the purpose of altering election outcomes. For that reason, it is the position of NY Citizens Audit that the stricter accounting industry threshold of materiality should be used instead of the “exceeds the margin of victory” threshold.