First the Impossible, Now the Improbable, in NY-23
Northern NY News. 27 November 2009. Written by Richard Hayes Phillips, Ph.D.
Editor's Note: Based on additional information provided by the St. Lawrence County Board of Elections, Dr. Phillips revised this article to improve clarity and accuracy.
CANTON, NY – As reported last week, impossible numbers were found in the St. Lawrence County election results for the special election in New York’s 23rd Congressional District. 93 “phantom votes,” more votes counted than the number of ballots cast, were reported in six election districts, and negative numbers reported for the “blank ballots,” or “undervotes.”
These were not the certified results. The author deeply regrets having said that they were. The numbers, which the Board of Elections attributes to data entry errors, have since been corrected. However, scrutiny of the certified election results reveals numerous districts (precincts) where the results, although not always mathematically impossible, are not credible.
On Friday, November 6, three days after the election, one of the involved campaigns obtained from the Board of Elections a spreadsheet of the preliminary (unofficial) election results, precinct by precinct. Absentee ballots had not yet been counted. This serves as an important “snapshot” with which to compare the final (certified) results.
As previously reported, voting machine failures at eight polling places in St. Lawrence County caused the Board of Elections to hand count those ballots. Realistically, there was no other choice but to do so. According to the Board, the locked voting machines were transported to a warehouse in Canton where the ballots were counted by hand. The problem with this procedure is that it is illegal under § 9-100 of New York State Election Law, which requires that the votes be counted at the polling place:
§ 9-100 At the close of the polls the inspectors of election shall, in the order set forth herein, lock the machine against voting, account for the paper ballots, canvass the machine, cast and canvass all the ballots, canvass and ascertain the total vote and they shall not adjourn until the canvass be fully completed.
An audit of the poll books and absentee voter lists for three of these eight polling places reveals that the preliminary hand count could not have been correct. In Louisville, there were 885 actual voters at the polls, but only 691 votes were counted for Congress on Election Night. In Waddington, there were 754 actual voters at the polls, but only 347 votes were counted for Congress on Election Night. In Rossie, there were 138 actual voters at the polls, but only 94 votes were counted for Congress on Election Night. 53 votes were counted later. Bill Owens got 50 of them.
One possible reason for the short counts on Election Night is that the Sequoia/Dominion ImageCast machines have two slots and two bins for ballots. There is a slot which sucks a ballot into the optical scanner, much like a dollar bill is sucked into a vending machine, and after the ballot is scanned it s into a locked box. There is another slot in the front of the machine which can be opened when the scanner breaks down and emergency paper ballots need to be segregated and counted by hand; these ballots into a separate locked box. It is possible that the Board of Elections initially counted the ballots from one box but not the other. But this is precisely why § 9-102.3(b) of New York State Election Law requires that the ballots be counted in public at the polling place, and why § 9-108.1 requires that the number of ballots be cross-checked with the poll books to be sure that all the ballots have been counted.
§ 9-102.3(b) Paper ballots and emergency ballots cast during voting machine breakdowns which have been voted shall then be canvassed and tallied, the vote thereon for each candidate and ballot proposal, announced and added to the vote as recorded on the return of canvass.
§ 9-108.1 The board of inspectors, at the beginning of the canvass, shall count the ballots found in each ballot box without unfolding them, except so far as to ascertain that each ballot is single, and shall compare the number of ballots found in each box with the number shown by the registration poll records, and the ballot returns to have been deposited therein.
Another problem with these voting machines is that it is mechanically possible to open both ballot slots, and both locked boxes, even while the optical scanner is operating. This opens the possibility that ballots could be deposited into the wrong ballot box, inadvertently or deliberately, and never be counted. An eyewitness who voted at the only polling place in Russell told me that she was not allowed to place her own ballot in the machine; a poll worker examined her ballot and placed it into the machine for her. This caused her to be concerned about both the privacy of her vote and the security of the vote count.
As previously reported, the number of “blank” ballots, or “undervotes,” is calculated by subtracting the number of votes counted for a given office from the total number of ballots cast. In the Congressional race, the highest percentage of “blank” ballots anywhere in St. Lawrence County was in Russell’s 2nd district. According to the poll book there were 590 actual voters at the polls, and there were 11 absentee ballots, for a total of 601, in Russell’s 1st and 2nd districts combined. According to the certified results there were 338 ballots cast, of which 23 (6.8%) were blank, in the 1st district, and 262 ballots cast, of which 27 (10.3%) were blank, in the 2nd district. It is highly unlikely that 10.3% of the voters made no choice among three candidates in one of the most hotly contested races in the nation.
The second-highest percentage of “blank” ballots for Congress was in Hammond. According to the poll book there were 569 actual voters at the polls, and there were 67 absentee ballots plus one special federal ballot, for a total of 637. According to the certified results there were 637 ballots cast, of which 51 (8.0%) were blank – again, a highly unlikely percentage for a hotly contested race. In Hammond, the preliminary (unofficial) results had shown 305 votes for Owens, 206 for Hoffman, and 37 for Scozzafava. The final (certified) results show 298 votes for Owens, 228 votes for Hoffman, and 60 votes for Scozzafava. The difference, which should represent the 67 absentee ballots, is minus 7 for Owens, 22 for Hoffman, 23 for Scozzafava, and, by subtraction, 29 blanks. Whether the in Owens’ vote total is an error or a correction is unknown. But there is simply no way that 29 (or even 22) of 67 voters who took the time and effort to cast an absentee ballot made no choice for Congress.
Hammond is not the only polling place where one candidate or another managed to lose votes subsequent to Election Day.
* In DeKalb’s 1st district, where there were 355 actual voters at the polls, the preliminary (unofficial) results had shown 201 votes for Owens, 128 for Hoffman, and 26 for Scozzafava. The final (certified) results show 189 votes for Owens, 132 votes for Hoffman, and 34 votes for Scozzafava. The difference, which should represent 15 absentee ballots, is minus 12 for Owens, 4 for Hoffman, and 8 for Scozzafava – a net increase of no votes at all.
* In Lisbon’s 1st district, the preliminary (unofficial) results had shown 146 votes for Owens, 149 for Hoffman, and 13 for Scozzafava. The final (certified) results show 121 votes for Owens, 159 for Hoffman, and 19 for Scozzafava. The difference, which should represent 19 absentee ballots, is minus 25 for Owens, 10 for Hoffman, and 6 for Scozzafava – a net decrease of nine votes.
* In Massena’s 9th district, the preliminary (unofficial) results had shown 108 votes for Owens, 87 for Hoffman, and 2 for Scozzafava. The final (certified) results show 119 votes for Owens, 69 for Hoffman, and 4 for Scozzafava. The difference, which should represent 14 absentee ballots and one special federal ballot, is 11 for Owens, minus 18 for Hoffman, and 2 for Scozzafava – a net decrease of five votes.
There are also places where more, not fewer, votes were added to the totals than can be explained by the reported number of absentee ballots. In most cases the discrepancy was only one or two votes, which could easily be due to corrections made during recanvassing of the vote totals as required by law. But some examples are not so easily explained.
* In Ogdensburg’s 1st district, where there were 305 actual voters at the polls, the preliminary (unofficial) results had shown 141 votes for Owens, 103 for Hoffman, and 10 for Scozzafava. The final (certified) results show 167 votes for Owens, 119 for Hoffman, and 16 for Scozzafava. The difference, which should represent 9 absentee ballots, is 26 for Owens, 16 for Hoffman, and 6 for Scozzafava – a net increase of 48 votes. Even now, there are reportedly 13 blank ballots out of 315, or 4.1% of the total. But more importantly, the electronic vote count on Election Night was short by 39 votes, or 12.8% of the actual total of 305. Either these were initially counted as blanks, or not counted at all, or some combination of the two.
* In Lisbon’s 2nd district, the preliminary (unofficial) results had shown 114 votes for Owens, 110 for Hoffman, and 9 for Scozzafava. The final (certified) results show 116 votes for Owens, 133 for Hoffman, and 12 for Scozzafava. The difference, which should represent 6 absentee ballots, is 2 for Owens, 23 for Hoffman, and 3 for Scozzafava – a net increase of 28 votes. Thus the electronic vote count on Election Night was short by 22 votes, or 8.3% of the actual total. (The poll books do not reveal the precise number of voters at the polls, because Lisbon was a multiple-precinct polling place, as were Massena’s 9th and 10th districts).
More examples, with somewhat less egregious numbers, could be cited for all of the categories presented in this article. But it suffices to show that there were suspiciously high percentages of “blank” ballots reported in Russell’s 2nd district and in Hammond; extraordinary declines in the vote totals subsequent to Election Day in DeKalb’s 1st district, Lisbon’s 1st district, and Massena’s 9th district; and lost votes on Election Night in Ogdensburg’s 1st district and Lisbon’s 2nd district. Each of these corruptions of the vote count can be attributed to electronic vote tabulation. Together with the breakdown or freezing of the Sequoia/Dominion ImageCast voting machines at eight polling places, there is more than enough evidence in St. Lawrence County to show that the court-ordered “pilot” election in New York’s 23rd
Congressional District was an utter failure, and that the time-tested lever machines were much more reliable.
Richard Hayes Phillips, Ph.D., is one of the leading election fraud investigators in the United States. His book on the 2004 Ohio election, Witness to a Crime: A Citizens’ Audit of an American Election, based on examination of some 30,000 photographs of actual ballots, poll books, and other election records, is available at http://www.witnesstoacrime.com