N.H. vote count shows disparity in machine vs. paper ballots

From Citizens for Legitimate Government:

2008 New Hampshire Democratic Primary Results -- Total Democratic Votes: 286,139 - Machine vs Hand (RonRox.com) 09 Jan 2008
Hillary Clinton, Diebold Accuvote optical scan: 39.618%
Clinton, Hand Counted Paper Ballots: 34.908%
Barack Obama, Diebold Accuvote optical scan: 36.309%
Obama, Hand Counted Paper Ballots: 38.617%
Machine vs Hand:
Clinton: 4.709% (13,475 votes)
Obama: -2.308% (-6,604 votes)

2008 New Hampshire Republican Primary Results -- Total Republican Votes: 236,378 - Machine vs Hand (RonRox.com) 09 Jan 2008
Mitt Romney, Diebold Accuvote optical scan: 33.075%
Romney, Hand Counted Paper Ballots: 25.483%
Ron Paul, Diebold Accuvote optical scan: 7.109%
Paul, Hand Counted Paper Ballots: 9.221%
Machine vs Hand:
Romney: 7.592% (17,946 votes)
Paul: -2.112% (-4,991 votes)

It wasn't only Clinton and Romney who benefited (or Obama and Paul who suffered) in those precincts that used Diebold machines. According to an analysis of the vote data by the Ron Paul partisans at www.ronrox.com, Republican Mike Huckabee of Arkansas won an average of 13.294 percent of those votes counted by hand -- but only 10.627 percent of those votes counted by the Diebold machines. Others who did decidedly worse on the Diebold machines were John McCain (36.631 percent on Diebold machines vs. 39.293 percent on paper), Bill Richardson (4.328 percent vs. 5.561 percent) and John Edwards (16.843 percent vs. 17.584 percent).

Is there something about the precincts that use Diebold machines that differ from those using paper ballots that could explain these disparities? Or is something else going on here?

In addition, Bev Harris's Black Box Voting site reports that the 31 votes cast for Ron Paul in the town of Sutton, N.H. were completely left off the tally sheet released to the public, raising more questions about yesterday's outcome.