False Vote Counts in Four Counties in NY-23


Northern NY News

Written by Richard Hayes Phillips, Ph.D.  

Wednesday, 02 December 2009 07:08


CANTON, NY – It is now widely known that zero votes were initially reported for Doug Hoffman in numerous election districts in New York’s 23rd Congressional District.  What has not been previously reported is that these votes were shifted to other candidates.  While most of these counts were corrected during recanvassing, they never should have been reported in the first place.


This vote switching is best illustrated in Madison County, where the Board of Elections (to its credit) released, for each election district (or precinct), its preliminary results, before the recanvass, and its final results, as certified to the State.  A comparison of the two reveals what really happened on Election Night.


In the initial vote count, Hoffman got zero votes in three election districts in Madison County.  In Fenner, the count was 157 for Owens, 248 for Scozzafava, and zero for Hoffman.  In Hamilton’s 3rd district, the count was 75 for Owens, 79 for Scozzafava, and zero for Hoffman.  In Sullivan’s 2nd district, the count was 173 for Owens, 251 for Scozzafava, and zero for Hoffman.


Somebody should have noticed this.  On Election Night, Scozzafava was awarded 578 of 983, or 58.8%, of the votes in these three districts, while winning only 583 of 16,770, or 3.5%, of the votes in the rest of the county.  This illustrates perfectly why election results need to be released at the precinct or district level.


These numbers were corrected during recanvassing of the results, and absentee ballots have since been added to the totals.  In Fenner, the certified count is 159 for Owens, 242 for Hoffman, and 21 for Scozzafava.  In Hamilton’s 3rd district, the certified count is 76 for Owens, 77 for Hoffman, and 4 for Scozzafava.  In Sullivan’s 2nd district, the certified count is 174 for Owens, 250 for Hoffman, and 11 for Scozzafava.  This amounts to a gain of 4 votes for Owens, a gain of 569 votes for Hoffman, and a net loss of 542 votes for Scozzafava.


The Board of Elections has attributed the false initial numbers to human error.  Poll workers mistakenly read the wrong line on the computer tape, or so the story goes.  But votes were not only denied to Hoffman; they were delivered to Scozzafava.  What obviously happened is that vote counts were switched.  Hoffman’s tallies on the Conservative Party line were given to Scozzafava, and Scozzafava’s tallies on the Independence Party line were given to Hoffman.  If all of Scozzafava’s 36 rightful votes in these three districts were on the Republican Party line, the result would be false tallies of zero votes for Hoffman.


Thus, for the “human error” explanation to be true, poll workers in three different polling places must have made the same two mistakes.


Also in Madison County there were two other districts with egregious errors that somebody should have noticed:




Countywide, the initial count for Madison County, reported on the morning after the election by the Watertown Daily Times, was 7743 for Owens, 8110 for Hoffman, and 1128 for Scozzafava.  With corrections and adjustments made, and absentee ballots counted, the final (certified) count is now 8290 for Owens, 9155 for Hoffman, and 724 for Scozzafava.  Thus, Hoffman’s lead of 367 votes on Election Night has grown to 865 votes – a net gain of 498.


For Oneida County, at 11:50 P.M. on Election Night, the Albany Times-Union posted these vote tallies: 3510 for Owens, 2432 for Hoffman, and 274 for Scozzafava.  Owens was reportedly winning Oneida County by 1078 votes, with 56% of the total.  The next morning, the Watertown Daily Times reported very different numbers: 2024 for Owens, 2779 for Hoffman, and 362 for Scozzafava.  Owens was now losing Oneida County by 755 votes, with only 39% of the total.  This represents an overnight reversal of 1833 votes.  But by that time, Hoffman had already conceded the election.


Preliminary precinct results obtained a few days after the election contained no votes from Lee’s 2nd and 5th districts.  The partial results from elsewhere in the county match what was reported in the Watertown Daily Times, so these were the only two districts not reporting.


But even the corrected partial results were incorrect.  In Camden’s 2nd district, the Board of Elections was still reporting 100 (74%) for Owens, 23 (17%) for Scozzafava, and 12 (9%) for Hoffman.  Somebody should have noticed this.  By comparison, Hoffman’s lowest percentage anywhere else in the county was 43% in Boonville’s 4th district.  In Camden’s other two districts, Hoffman received 66% and 67% of the vote.  The Camden example demonstrates clearly the methodology for vote switching.  Hoffman was awarded 12 votes, not zero.  These votes had to come from somewhere.  The simplest explanation is that Hoffman’s tally on the Conservative Party line was given to Owens, and Owens’ tally on the Working Families Party line was given to Hoffman, who suffered a net loss of at least 75 votes.  If these votes were shifted not to Scozzafava but to Owens, the other leading candidate, the margin was affected by 150 votes.


Whether these numbers from Camden’s 2nd district have been corrected is not certain, because Oneida County has not released its final precinct results.  The final countywide results show 2243 for Owens, 3225 for Hoffman, and 459 for Scozzafava, which represent, since the corrected partial results reported the morning after the election, gains of 219 votes for Owens, 446 votes for Hoffman, and 97 votes for Scozzafava.  Either way, Hoffman’s countywide percentage has grown from 39% on Election Night to 54% today.


The Oneida County Board of Elections has confirmed that optical scanners were used only in the Town of Marcy.  Lever machines were used elsewhere.  Thus it seems likely that the numbers from Camden’s 2nd district would have been corrected during recanvassing, because the true ballot positions that correspond with the vote tallies are plainly visible on a lever machine.


In Jefferson County, Sean M. Hennessey, Democratic elections commissioner, said that poll inspectors in four districts reported that Hoffman had received zero votes after inadvertently reading the wrong line of the poll system’s printout.  Hennessey said that results in some other districts were either incorrectly relayed by the poll worker or incorrectly typed by the part-time staff answering phones at the Jefferson County Board of Elections office.


Jefferson County election officials blamed the mistakes on “chaos” in their call-in center, and on inspectors who read numbers incorrectly when reporting results over the phone.  “The machines were not at fault,” said Jerry O. Eaton, Republican elections commissioner for Jefferson County.  “It’s all human error that happens every election.”  Jefferson County has not conducted a hand count of the paper ballots from the election districts where the zero vote counts were reported.


The initial vote count reported in the Watertown Daily Times was 9996 for Owens, 9439 for Hoffman, and 1155 for Scozzafava.  By the time the Jefferson County Board of Elections provided its preliminary precinct results to one of the involved campaigns, three days after the election, the zero vote counts had been corrected in all four districts.  The corrected preliminary results were 10,238 for Owens, 10,358 for Hoffman, and 1179 for Scozzafava.  This represented net gains of 242 votes for Owens, 919 for Hoffman, and 24 for Scozzafava, and a change of 677 votes in the countywide margin.  The combined increase of 1185 votes (5.8%) indicates that not all districts had reported their results when the Watertown Daily Times went to press, and suggests that vote shifting had altered the margin by about 640 votes.  But more importantly, the ratio of the newly counted votes (Hoffman got 78% of them, Owens 20%, and Scozzafava 2%) indicates that, in the four districts with the zero vote counts, most of Hoffman’s votes had gone to Owens.  The “tally sheets” from these four election districts should tell the tale.


But even the corrected preliminary results were not correct.


·        In Wilna’s 5th district, the count was 122 for Owens, 154 for Scozzafava, and 7 for Hoffman.  Apparently, Hoffman’s tally on the Conservative Party line was switched with Scozzafava’s tally on the Independence Party line.  In Wilna’s other four districts, Scozzafava got only 5.2% of the vote, but in Wilna’s 5th district she got 54% (of 283), suggesting that about 140 votes were switched from Hoffman to Scozzafava.


·        In Watertown’s 15th ward, 4th district, the count was 92 for Owens, 5 for Scozzafava, and 3 for Hoffman.  Apparently, Hoffman’s tally on the Conservative Party line was switched with Owens’ tally on the Working Families Party line.  In the other four districts of the 15th ward, Hoffman got 43% of the vote, but in the 4th district he got only 3% (of 100), suggesting that about 40 votes were switched from Hoffman to Owens, thus affecting by 80 votes the margin between them.


Perhaps those numbers have been corrected.  A second corrected preliminary count was posted in the Watertown Daily Times, still prior to the counting of absentee ballots.  This, the third count, was 10,460 for Owens, 10,884 for Hoffman, and 1179 for Scozzafava.  This represented net gains of 222 votes for Owens, 526 for Hoffman, and zero for Scozzafava.  The combined increase of 748 votes appears inexplicable; the precinct results previously given by the Jefferson County Board of Elections contained numbers for all 91 election districts in the county.  But the ratio of the newly counted votes (Hoffman got 70% of them, Owens 30%, and Scozzafava zero), indicates that adjustments had been made.


Now the absentee ballots have been counted in Jefferson County, and the final (certified) results are 10,902 for Owens, 11,354 for Hoffman, and 1414 for Scozzafava.  Hoffman was reportedly losing Jefferson County by 557 votes on Election Night, and ended up winning by 452 votes – a turnaround of 1009.


In Oswego County, according to the Valley News, “problems with the reporting of numbers on election night led the Oswego County Board of Elections to remove the results from its web site on Wednesday, the day after the election.”


On Election Night, Hoffman was reported to lead by only 500 votes (exactly) with 93% of the districts (115 of 124) reporting, but inspectors found that Hoffman actually won by 1748 votes – 12,748 to 11,000 (exactly) – and that Scozzafava had received only 950 votes.  These corrected unofficial numbers, reported on Saturday, November 7th, included 100 percent of the districts in the county.


The initial count had been 10,882 for Hoffman, 10,382 for Owens, and 1339 votes for Scozzafava.  Thus the recanvass resulted in a net gain of 1,866 votes for Hoffman, a net gain of 618 votes for Owens, a net loss of 389 votes for Scozzafava – and an increase of 1248 votes in Hoffman’s countywide margin.


Because Oswego County has not released their precinct results for any stage of the vote counting, we do not know if Hoffman received very few votes or zero votes in the districts with the false numbers.  But we do know, from the countywide comparisons posted online by the Valley News, that the initial point spread was 2.21%, and that the corrected point spread was 7.07%, which is 3.2 times higher.  This suggests that the reported 500-vote margin should have been 1600 votes.


We also know that Scozzafava’s initial percentage was 5.93%, and that her corrected percentage was 3.85%, suggesting that her initial count should have been about 870 votes (not 1339), and that about 470 votes had been shifted from Hoffman to Scozzafava.  As this does not account for the entire discrepancy, some votes, probably about 315, must have been shifted from Hoffman to Owens in order to bring the false margin down to 500 votes.  Inspection of the tally sheets from the affected polling places should tell the tale.


Both the initial count and the corrected count should be viewed in light of a statement made by Oswego County Democratic Elections Commissioner William Scriber, as reported in the Palladium-Times.  Scribner said: “No votes changed from election night, to election morning, to the day after, right up to today.”  Note that Hoffman’s reported lead of 500 votes on Election Night had became 1748 votes four days later, so Scribner’s statement, reported on Friday, November 20th, is not correct.


We have carefully preserved all the newspaper articles cited above, lest they disappear from cyberspace.


Now the final results have been certified by the Oswego County Board of Elections.  The certified countywide totals are 11,552 for Owens, 13,300 for Hoffman, and 1158 for Scozzafava.  The certified absentee ballot count is 437 for Owens, 461 for Hoffman, and 208 for Scozzafava, for a total absentee ballot count of 1106 (emergency ballots, and affidavit, or provisional ballots, have been added to the totals also).  According to Scriber, as reported in the Palladium-Times, 1145 absentee ballots were received in Oswego County, so 39 (3.4%) of them must have been blank, void, or write-ins.  Countywide, Oswego County is reporting 1338 blank ballots for Congress out of 27,394 ballots cast, or 4.9% of the total.


Actually, the posted election results are intended to reveal very little.  Not only is there no district by district breakdown, but we do not know the number of ballots cast at the polls or the number of absentee ballots, because any blank, void, or write-in ballots are listed as such and are not included in the numbers for “machine” or “absentee.”  But some things are nonetheless revealed.


Most importantly, the total number of ballots cast adds up to 27,462 countywide.  The ward totals and the town totals tally up to 27,462.  The total number of ballots cast in the 25 districts of the Oswego County Legislature tally up to 27,462.  The same number is given in the .pdf file posted by the Board of Elections for total ballots cast in three countywide races, to wit: County Treasurer, Proposal Number One, and Proposal Number Two.  The total number of ballots cast for State Supreme Court Justice is exactly twice this number, 54,924, because voters were choosing two candidates, not one.  But the total number of ballots cast for Congress, countywide, is reportedly 27,394, and 30 of these are “special federal” ballots cast for Congress only (being the only federal office on the ballot), so the count of ballots cast for Congress is short by 98, unless there were fewer ballots cast for the special Congressional election than for the general election.


What the data for “emergency” paper ballots tell us is the locations where the voting machines broke down.  There were 85 “emergency” paper ballots, and all but one was cast at four locations: Fulton City’s 4th Ward, Oswego City’s 2nd Ward, Oswego City’s 4th Ward, and the Town of Hastings.  There are no more than 22 “emergency” paper ballots listed for any town or ward in the county.  This cannot be the cause of the “problems” on Election Night that caused the Board of Elections to rescind the election results posted on its own web site.  How hard can it be to count 22 paper ballots?  There were only seven contests on the ballot in Fulton City and Oswego City, and ten in the Town of Hastings.


In summary, switching of votes from Hoffman to other candidates occurred in four counties in New York’s 23rd Congressional district.  In Madison County, about 540 votes were shifted from Hoffman to Scozzafava in three districts.  In Oneida County, about 75 votes were shifted from Hoffman to Owens in one district, and a mysterious 1833-vote alteration of the margin, to Hoffman’s detriment, was reported on Election Night.  In Jefferson County, votes were shifted from Hoffman to other candidates, mainly to Owens, in six districts, thus altering the margin by about 860 votes.  In Oswego County, about 470 votes were shifted from Hoffman to Scozzafava, and about 315 votes were shifted from Hoffman to Owens, in an unknown number of districts.  Altogether, vote switching in four counties altered the reported margin between Owens and Hoffman by an estimated 2,650 votes.


And this is only what we know about.  These are the districts, or precincts, where the entire vote counts on the Conservative Party ballot line were shifted to other candidates.  With concealed electronic vote counting, partial shifts of the vote count could occur without a trace, and not be readily apparent in the election results.  And yet the New York State Board of Elections is expected to certify these election results and the untrustworthy machines that produced them.


For the record, here is the comparison of the initial results reported on Election Night, and the final results certified to the State, for Jefferson, Madison, Oneida, and Oswego counties:




Election Night            Owens             Hoffman          Scozzafava


Jefferson             9,996   48.5%              9,439   45.8%            1,155   5.6%

Madison             7,743   45.6%              8,110   47.8%            1,128   6.6%

Oneida               3,510   56.5%              2,432   39.1%               274   4.4%

Oswego           10,382   45.9%            10,882   48.1%            1,339   5.9%


Four Counties   31,631   47.6%            30,863   46.5%            3,896   5.9%


Certified Results        Owens             Hoffman          Scozzafava


Jefferson           10,902   46.1%            11,354   48.0%            1,414   6.0%

Madison             8,290   45.6%              9,155   50.4%               724   4.0%

Oneida               2,243   37.8%              3,225   54.4%               459   7.7%

Oswego           11,552   44.4%            13,300   51.1%            1,158   4.5%


Four Counties   32,987   44.7%            37,034   50.2%            3,755   5.1%


 Author's Note: Since this article was written, the Gouverneur Times has learned that the central point of my first article is correct.  “Blank votes,” or “undervotes,” are calculated, not counted, in the St. Lawrence County Board of Elections software.  Thus the final column in their chart, for “blank votes,” will always equal the number required for the votes to add up, whether or not the vote count is correct, even if a negative number is required to make this happen.


The reason for this, also acknowledged to the Gouverneur Times by the St. Lawrence County Board of Elections, is that the software for the Sequoia / Dominion ImageCast voting machines does not count the “blank votes.”  The importance of this cannot be overstated.


The vote count is supposed to be a breakdown.  All the votes for the individual candidates, plus the “blank votes” or “undervotes” and the “void ballots” or “overvotes,” are supposed to add up to the total ballots cast.  If one of these categories is merely calculated by subtracting the other categories from the total, there is no check upon the count, and it will always balance.


Think of it this way.  If you are a cashier, and you have the tape indicating how much money is supposed to be in the cash register, you cannot count the paper money, subtract that number from the total on the tape, and report the remainder without counting the change.  You will lose your job.


The insidious thing about voting machine software not counting the “blank ballots” is exactly what was alleged in my article.  “Phantom votes” can be introduced into the system.  The computer can be programmed to add votes to one candidate’s total, and the unsuspecting Board of Elections will dutifully subtract all the candidates’ vote counts from the total ballots cast, report the remainder as “blank votes,” all the numbers will add up perfectly, and no one will be the wiser – except, of course, if negative numbers turn up in the column for “blank votes.”


I do not know that “phantom votes” were actually introduced anywhere in the 23rd Congressional district.  But that is not the point.  The voting machine software could allow it.  Thus the voting machines should not be certified, and perhaps the election should not be certified.


The mechanical lever machines did not count the “blank votes” or “undervotes” either.  The number was derived by subtracting the vote totals for the candidates from the total votes cast.  (Overvotes are not allowed; two levers cannot be pulled for the same position).  There is a simple remedy for this.  Add a row at the bottom for “none of the above,” and require each voter to pull one lever in each column before opening the curtain that causes the votes to be cast.  The Boards of Elections can then add a column for “blank votes” to their worksheets, as if it were another candidate, and subtract all the candidates’ vote counts from the total ballots cast.  The final column, the remainder, should always equal zero.


Last Updated on Wednesday, 02 December 2009 08:30