Reply #3: There Were Some Bizarre Fluctuations in NM, WA, and ME [View All]

[On the Road]

during the counting, especially in Dean's numbers. I would be interested in hearing anything about it.

I looked at the precinct-level data in NH. The distribution among the candidates is pretty much what I would have expected: geographic advantage for Dean near VT, for Kerry near MA. No 6-sigma deviations or anything. The optical/manual difference seemed to be a function of geography rather than vote tampering.

I don't know what to make of the Tom Dark experience. It does make he mistrust the polls in certain situations. Polls are vulnerable to response variables -- people saying one thing and voting another. They may say a candidate's name just to avoid being ignorant, or say they support the canvasser's candidate just to get rid of them, or avoid reavealing their loyalties if their preferred candidate is unpopular. Often these things cancel each other out, but not always. Dean may have been overrported in the polls at first, but is now underreported.

I find it easy to believe that polls are sometimes dishonest, but many polls put together are usually on the mark.