Correcting Factual Errors in Salon: "Was the New Hampshire Vote Stolen?"

Jan. 22, 2008
by Bruce O’Dell and Theron Horton, Election Defense Alliance

Problems with Manjoo’s Source Data

In Farhad Manjoo’s article, “Was the New Hampshire vote stolen?” the author acknowledged in the text that he had based his article on an anonymous partisan website,, stating

“The most thorough analysis I've seen was performed by an anonymous supporter of Ron Paul.”

The origin of the site was described by its webmaster as follows:

This site was birthed out of a late night whim to try to see if I could make better sense of the numbers I saw coming in from the New Hampshire primaries. It was originally just an exercise in data formatting, but during the night, with the great help of fellow Ron Paul supporters from, I began to add data points to my charts.[1]

However, this “most thorough analysis,” though well intentioned, was initially based on incorrect data and on ad-hoc statistical categories.

In fact, the webmaster, who has consistently acted responsibly to correct all issues with his new site/project, subsequently removed the voting district size categories that Manjoo used as the basis of his January 11 article. According to a screenshot from the evening of January 12:

I have changed the size of the town breakdown because it was brought to my attention by another person offering vote stats that for the hand-counting in large towns, there wasn't enough data for it to be statistically significant.”[2]

Nor did Majoo check his source’s primary data. The official tallies were available on the Official New Hampshire Secretary of State Website on the morning of January 10. Yet as-of 9:18 AM CST on Monday, January 14, 2008, a full six days after the New Hampshire Presidential Primary, still listed the total vote for Clinton as 112,238 and the total vote for Obama as 104,757, while the New Hampshire Secretary of State website’s official vote total for Clinton was 112,610 and the total vote for Obama is 105,007. Also as of that time, had not corrected the voting methods for numerous voting districts.

Errors on Size of Hand-count Voting Districts

In paragraph 11 Manjoo, utilizing the data, states “in places with more than 1,500 votes, here, Clinton, not Obama, did better in hand-count areas”.

However, there were no hand-counted voting districts that had “more than 1500 votes”. The largest Democratic hand-counted voting district had only 1,172 votes, in Newport in Sullivan County. When we brought this the attention of the author, he indicated that he was referring to “Claremont and Franklin Counties”.

The three voting districts that comprise each of these towns - rather than “counties” - had been consolidated into a single number on the website at the Concord Monitor Online[3]. The consolidation of the voting districts’ vote totals distorted Manjoo’s analysis of the data. But there is a more serious problem with his analysis.

Error on Voting Method in Claremont

If Majoo truly intended to use Claremont as one of the “places with more than 1,500 votes” to make the assertion that, “Clinton, not Obama, did better in hand-count areas” there is a problem. According to the Official New Hampshire Secretary of State Website, Claremont Wards 1, 2 and 3 are actually not hand-count voting districts; they use Accuvote optical scanners. In fact, the Claremont City Clerk’s Office verified by phone that they Accuvote[4]. Again in this case, Manjoo was using as his source. has subsequently acknowledged that the Claremont voting method data that they had originally provided was incorrect.

Error on Clinton Performance in Large Hand-count Voting Districts

Additionally, the hand counted paper ballot analysis provided in paragraph 12 states (emphasis from the original document) :

Manjoo: “Let me say that again: In large areas, Clinton did better in places where votes were counted by hand than where votes were counted by machine.”

First, “large areas” is a spurious category; second, there is no apparent factual basis for Manjoo’s statement.

According to the Secretary of State’s original posted data, Clinton won 35.56% of the vote in “the larger hand-count voting districts” (defining those, as voting districts between the maximum size of 1,172 votes and ones half that size, or 550 voters) and Clinton got 41.02% of the vote in the “larger optical scan districts” (defining those as voting districts between the maximum of 5,542 and one half that size, or 2,771 votes). Even if you simply look at the 32 largest hand count v. the 32 largest optical scan voting districts, Clinton’s margin was 35.56% for the 32 largest hand count districts, and 40.85% for the 32 largest optical scan districts.

Finally, paragraph 13 identifies Amherst as a “county”; it is a voting district.

Errors Should Be Corrected

These factual errors should be corrected, and the thesis of the article should be re-examined.

Why Is It that People Vote Differently on Optical Scan?

We wish that it was possible to dismiss concerns about the outcome in New Hampshire simply by saying optical scan tallies vary from hand-count results because people who use optical scan equipment are clustered in communities that tend to vote differently. If you accept the basic integrity of the current voting system, that’s a reasonable and comforting position; but assuming the integrity of the vote as a premise of your argument against those impeaching the integrity of the vote is circular reasoning. It is also possible that optical-scan voting districts have different voting patterns than hand-counted paper districts simply because that’s how the equipment says they tend to vote.

There is no certain way to resolve which alternative is correct as long as votes are counted in secret by machine.

Where Do We Go from Here?

That’s why we do agree with Manjoo, when he says that that “…as many voting-reform experts have argued, manually counting the votes should be a routine in any race… we should, at least, conduct a randomized, accountant-approved audit of ballots” - but we would add, only if ballots can one day be handled with the same care – and controls - we apply to cash. You’ll not likely soon see boxes of cash at the corner bank stored in a cardboard boxes secured with stick-on paper seals, and grudgingly brought in to be double-checked weeks after their deposit - only if and when someone else offers to pay for it.

We wish we could share Mebane and Manjoo’s confidence that the host of well-documented computer system and voting procedural vulnerabilities are not being actively exploited, and specifically, their stated faith in the official outcome of the Ohio Presidential election and recount there in 2004. The jailed election officials who gamed the Cuyahoga County recount and their counterparts in the fifty seven Ohio counties that lost or destroyed their 2004 ballot records, despite court orders to preserve them, hardly require “conspiracy theorists” to call attention to their actions.

No one calls an IT auditor at a bank an “embezzlement theorist” when he or she states that security vulnerabilities discovered in accounting procedures or in the bank’s software might actually have been exploited.

[1] See: (note correction: changed from .../home on 01/23)

[2] See: screenshot of the website 1/12/07 10:06 PM CST


[4] Claremont County Clerk’s contact information: VOICE 603.542.7003 • FAX 603.542.7014