31 March 2014

In the shadow of overwhelming irregularities in the Ohio’s election, New Mexico has played out it’s own post-election drama almost unnoticed by anyone outside the state. Even before the November 2, New Mexico had been in the news for reports of malfunctioning voting machines and other problems. Hundreds of incidents were reported on Election Day. The state got some national attention for once again leading the nation in undervote rate. All this coupled with a small margin of victory in the presidential race (just 3/4 of 1%) led to concern about the accuracy of the results.



Soon after the November 23 certification of New Mexico’s election results, the Green and Libertarian presidential candidates requested a recount and deposited the $108, 000 required by the New Mexico State Code. By the middle of December over a thousand volunteers had been organized to act as observers and coordinators for the recount. On December 15, the State Canvassing Board granted the request for a recount, but also demanded that the recount advocates would be required to come up with $1.4 million by 10 a.m. the next day in order to proceed. That demand was contrary to New Mexico law as there is no legal requirement that candidates pay the full cost of a recount in advance, nor any way to accurately estimate the cost of a recount before it is completed. Recount proponents went to court to have the $1.4 million charge dropped. On December 17, a New Mexico court reaffirmed the decision of the Canvassing Board and on December 23 the New Mexico Supreme Court upheld the $1.4 Million demand.



Around this time the Greens and Libertarians offered a compromise in which they would pay in advance for a recount of 10% of the state's precincts, and then seek a full recount only if the results warranted it. Governor Bill Richardson repeatedly rejected this offer. In spite of this, Secretary of State Rebecca Vigil-Giron suggested last week that because recount advocates had never presented her with a list of 10% of the precincts she was sending a letter to county clerks giving them permission to clear the memory logs of the voting machines (http://www.abqjournal.com/elex/apmachines01-13-05.htm). But she and the Governor had never accepted the compromise and never requested the list of precincts. Saturday’s court decision to deny a temporary restraining order to protect the machine memory and the process of erasing that began in several counties immediately results in a forced electoral amnesia that precludes any chance of verifying the election data. In the case of the paperless voting machines used in 82% of the state on Election Day, the only record that remains is the certified canvass report.



But that canvass report is riddled with anomalies. In comparing the number of ballots cast (that is, the number of voters who signed the poll books or submitted an absentee or provisional ballot) with the total number of presidential votes in any precinct, there turn out to be three possibilities. There can me more ballots cast than total votes for president (these are called undervotes), the same number of votes counted as ballots cast (I’ll call those zero-undervotes), and alarmingly there is also the situation where the number of presidential votes exceeds the number of ballots cast (these are called ‘phantom votes’). Since these are all mutually exclusive results, it is extraordinary to find excessively high rates of all of three in the same state for the same contest, but that’s what happened in New Mexico according to the canvass report.



In 2004 New Mexico once again led the nation in undervote rate. According to the canvass report, 1 in every 37 ballot casts, 21,084 voters, did not cast a vote for any presidential candidate. On Election Day the rate was nearly 1 in 20 voters with 92 precincts reporting undervote rates over 10%. In spite of this, the New Mexico Secretary of State seems surprisingly unconcerned, commenting recently that she “doesn't spend a lot of time on undervote issues," going on to speculate that "some voters are just not concerned with the presidential race." (http://www.knoxstudio.com/shns/story.cfm?pk=MISCOUNT-FINAL-12-22-04&cat=AN)



Had she spent more time on undervote issues, she may have noticed that 77% ballots with presidential undervotes statewide were cast on push button voting machines – the ones without a paper trail, the ones that are now being erased. Further, in precincts that used optical scan machines for Early Voting and push button electronic machines on Election Day, the undervote rate on Election Day was almost six times higher. Somehow the unconcerned voters managed to live predominately in counties using electronic voting machines and chose to express their lack of concern on those machines.



But something must have gotten them to the polls. The large number of zero undervote situations in the state provides a way of determining whether to blame undervotes on the voters or the machines. In spite of New Mexico’s distinction of leading the nation in undervote rate, over half of the precincts in the state reported zero presidential undervotes at some stage of voting, with over half of the absentee precincts reporting zero undervotes.



By examining the undervote rate in down ticket contests on ballots that all contain presidential votes, it is possible to determine a normal undervote rate for each of the other statewide races, bond questions and amendments. If high presidential undervote rates indicate voters' lack of interest in the presidential contest as the Secretary of State maintains, one would expect the undervote rates in down-ticket contests to remain roughly the same regardless of the presidential undervote rate, but they do not.



A comparison of the normal rates to down-ticket rates in situations where there were presidential undervotes showed that the down-ticket rates increase by an amount that parallels the presidential undervote rate. The presidential undervote went from zero to 3.5%, the congressional undervote rate went from 1.8% to 5.3%, the justice's races went from 4.9% to 8.2%, the bonds and amendments from 14% to 17 1/2% - they all increased by the same amount that the presidential undervotes increased. This analysis strongly suggests that most of the 21,084 undervotes in the presidential election do not reflect unconcerned voters, but rather it suggests that the voting machines may have lost entire ballots.



You can blame the voters or you can blame the machines. Outgoing Director of the Bureau of Elections Denise Lamb, among other explanations she has offered, has blamed undervotes on the phenomenon of ‘fled’ voters – voters who enter the voting booth and then decide not to vote. At the same time there is compelling evidence that many machines may have failed to count votes. The only way to resolve the question of whether those votes were counted or not is through a meaningful recount in which the memory logs of the machines is examined. The Secretary of State has now eliminated the possibility of resolving that question by ordering the clearing of the machines.



And then there is the troubling phenomenon of phantom votes – cases in which the number of votes exceeds the number of ballots cast. The canvass report reveals 2,087 presidential phantom votes and over 13,000 in other contests on the ballot. The Secretary of State has maintained that phantom votes are not possible. While logically I would have to agree with her, the fact remains that the certified results report that in Early Voting in one Bernalillo County precinct, 178 ballots cast resulted in 319 presidential votes and that voters in one precinct in Dona Ana County apparently cast an average of 2 1/4 absentee ballots. In fact, presidential phantom votes were reported in 15 of the 33 counties and they appear down ticket as well. In one precinct in San Juan County 318 voters somehow managed to register 2,161 votes, 2,079 of them to incumbent county clerk Fran Hanhardt and just 82 for her opponent Glojean Todacheene. San Juan County was the first to announce its intention to clear the memory on their voting machines, initially disregarding the legal requirement of notifying all of the political parties involved in the election.



Since November 2, many have dismissed the situation in New Mexico as irrelevant. With only five electoral votes it couldn’t have effected the election or, as I’ve heard too often, “It wouldn’t have changed the outcome, so who cares?” Well it’s true that 140 votes would not have effected the outcome and the San Juan county clerk won the election that she was in charge of running by a comfortable margin, but it makes you wonder about the rest of the precincts. In fact, the canvass report cannot be considered an accurate record of the will of the people of New Mexico as expressed on November 2. Why have Governor Richardson, Secretary of State Vigil-Giron, and Chief Justice Maes opposed a recount?



--

Warren Stewart is a researcher and advocate for National Ballot Integrity Project. The data from the certified canvass report is available in an Excel file at http://www.votersunite.org/info/newmexicophantomvotes.asp. The research he has done with Ellen Theisen of VotersUnite.org is detailed in The Summary Report on the New Mexico Election Data available at the same site. The official canvass report of the 2004 New Mexico General election can be downloaded from the Secretary of State’s site.