January 11, 2008
Election Fraud in NH Primary? Or Not?
The internet has been abuzz since Tuesday night with wild claims that Hillary Clinton “hacked” the NH Primary — or that someone else perpetrated the dirty deed to help Clinton and McCain triumph in the Granite State. Some point to differences between the margins in hand-counted precincts vs. optical scan precincts. Others claim that the pre-election polling could not be so far off from reported results.
Both of these cries of “fraud”, and “hacking” are based on flawed logic — and stunning ignorance or basic misunderstanding of statistics. They also fail to look at the demographic make-up of precincts that produced different margins for the candidates. On the other hand, there has been an almost universal failure to consider whether well-documented problems with the type of optical scan machine used in New Hampshire offers at least a partial explanation of how this happened.
Let us first take a look at fundamental concepts of statistics. First of all, polls use samples, not the entire universe of possible voters. How these samples are constructed can bias the results by over-sampling some groups of voters and under-sampling others. This skew in the sample will yield different results. The timing of the poll in a volatile election can also lead to different answers being given –e.g., pre-Hillary tears vs. post-Hillary tears. Lastly, all polls have a margin of error.
The concept of margin of error is one of the least-understood ideas about polling. When a pollster says that Candidate A is polling at 35 per cent with a margin of error of 4 here is what is really being said: “Candidate A has support between 31 and 39 per cent.” So we really should not be surprised when Candidate A gets as little as 31 per cent or as much as 39 per cent. The pre-election polling in New Hampshire showed a virtual dead heat with Obama and Clinton having overlapping margins of error. Why are people so surprised that Obama got votes at the lower end of his range and Clinton got votes at the higher end of her range?
Demographic differences between precincts have been almost universally ignored amidst all the cries of “election fraud”. In New Hampshire the hand-counted precincts are in rural areas and small towns, while optical scan machines are used to count votes in urban areas like Manchester. Is it so hard to comprehend that rural and small town voters might cast their ballots in a different mix than voters in bustling Manchester? Indeed, if there was uniformity across precincts with different demographics there would, in my opinion, be more cause for concern that the count had been manipulated in some way. The variation between rural and urban precincts is normal and natural, not a reason to cry “fraud.” As was pointed out elsewhere:
The difference between the machine-counted precincts and the hand-counted precincts, writes blogger Ed Morrissey, “is that the former tend to be in the bigger cities such as Nashua and Manchester where Hillary had significant polling leads before the primary. The hand-counted precincts were in areas known to be Obama territory.” http://blog.wired.com/27bstroke6/2008/01/new-hampshire-t.html
So what about the machines that are used in New Hampshire — could they be part of the problem? The answer to this question is a resounding “maybe.” The machines that count the votes in the Granite State are the Premier Election Solutions (formerly Diebold) AccuVote optical scanners. A private company, LHS Associates, provides programming and technical support for all the machines in New Hampshire and other New England states. There is no public oversight to this process — and this raises the hackles of those who believe that elections should be transparent to the citizenry and fuels the suspicions of that something is fishy in New Hampshire. But this murkiness in the process alone does not support claims of election fraud. It simply creates a feeding frenzy among conspiracy-theorists.
There are well documented problems with optical scanning machines in general and the AccuVote model used in New Hampshire in particular. Three common issues with optical scanners are poor machine calibration, failure to “read” certain types of inks that voters use to mark their ballots, and mis-programming of the ballot definition files. Let’s look at each of these in turn and see whether they might have contributed to problems in the recent New Hampshire primary.
- Machine calibration: Optical scan machines use beams of light to “read” the marks on the paper. If the light beams do not “look” at the right part of the paper ballots they will not “see” the voter’s marks and will not count the votes correctly. It is possible that some of the machines in New Hampshire were mis-calibrated.
- Inks: Optical scan machines do not “read” the inks of some popular pens that voters use to mark their ballots. So some votes might not be counted if the wrong pen is used. Marks made with inks that are visible to the the human eye are undetected by the optical scan machines.
- Ballot definition programming: The information about which spot on the paper ballot the machines should “look”to find votes for a particular candidate is know as a ballot definition. This information changes every election. Therefore new data is loaded into the optical scanners each time. This programming is provided by the private company LHS Associates. Mistakes have been documented with ballot definition programming before and this is one place to look if seeking a possible explanation for inexplicable election results.
A hand recount of the voter-marked optical scan ballots can easily correct for all three of these errors and provide an accurate tally of the votes.
Finally, could these machines have been hacked? Yes, this is the same model machine that was hacked by Harri Hursti in Leon County FL. Anyone with access to the machines and a modicum of technical skill could have swapped memory cards and changed the reported election tally. So how do we know whether the machines were tampered with? If these had been the paperless touchscreen machines we would have no way of knowing. Fortunately for all of us, New Hampshire has paper ballots that can be re-counted.
And thanks to candidates on both sides of the aisle (Democrat Dennis Kucinich and Republican Albert Howard) there will be a recount of New Hampshire’s paper ballots that will prove once and for all whether there was “fraud” or “hacking” of the vote.
My hunch is that some errors will come to light, probably not enough to substantially change anything — and the specters of “fraud” and “hacking” will be laid to rest for the New Hampshire primary.
Ah, but there are 48 more states to go … and many of them are not fortunate to have paper to either audit or recount. So controversy is likely to continue through November, and perhaps beyond, especially in states that rely on non-recountable, non-auditable paperless voting machines. At last count there were around 20 states that had some paperless voting machines. One of these is South Carolina, which comes up shortly on the primary calendar.
One point that needs to be made is that the New Hampshire primary is a perfect example of why paper ballots alone are not the ultimate answer to election integrity. Post-election audits are needed to demonstrate that the results are to be trusted. They should be routine, mandatory and scientifically sound. Several states have already passed laws requiring post-election audits but New Hampshire is not one of them.
On the federal level Rep. Rush Holt (D-NJ) has proposed legislation (H.R. 811) to mandate audits for all Federal elections but the bill has not made it to the floor of the House of Representatives, despite having 216 co-sponsors. Now Rep. Holt has decided to introduce an emergency opt-in bill to fund a transition to paper ballots and post-election audits for those states which choose to take advantage of it.
In light of the controversy swirling around the New Hampshire primary, please join me in urging Congress to act on this bill quickly so the nation is spared similar situations in November. The memories of Ohio 2004 and Florida 2000 still linger — American voters deserve better in 2008.