Sigh. The evidence for this article comes from ... Democratic Underground. Anyway, let's count the errors:
1) Believing that mid-day polls were somehow more accurate than vote totals. People don't vote in equal number throughout the day, which skews the totals, and the totals become especially off when democratic groups were sending people to the sites where they knew the polls were taking place to skew the vote totals.
2) Believing pre-election polls (and only certain polls, i.e., the ones that showed Bush losing) were accurate. Republicans have reported being under-represneted in polls for years.
3)Believing that network internal polls, which were wrong in 2002, 2000, 1998, 1996, & 1994, were somehow accurate this time.
Arrrgh. Look. We did hundreds of threads about this around election time. The basic problem of the DU studies is simple: It assumes the exit polls were accurately modelled and unbiased. There is no evidence in either case that they were.
I've actually sent him information about the exit polls several times. It was the BS about vote-switching, the recount and voter suppression that I wanted to nail him on to get him to shut up once and for all.
There is a word for why the polls were skewed the way they were on election day: Zogbyism.
They wanted to supress the Republican vote by sheer biased media reporting. Make it up. Like they did with the faked National Guard memos.
Zogbyism continues to this day.