On Monday, two polls in Michigan inspired a national pollster to blog about how odd he thought they were, and to pretty thoroughly dissect (and criticize) their methodology, Stephen Henderson writes in the Free Press.
His blog is a fascinating look at how polls work, why they sometimes differ in outcome, and at least infers a nice set of guideposts for readers trying to make sense of the flurry of numbers.
Nate Silver, whose fivethirtyeight.com blog I praised in my Monday column as painting electoral "portraits" that differ substantially from most polls' snapshots, didn't much care for the most recent polls released by Mitchell Research and Foster McCullom White Baydoun, two firms that are old hands at polling Michigan.
The Foster McCullom poll, from last week, showed Mitt Romney with nearly a four-point lead over President Obama in Michigan; the Mitchell poll, from Monday, showed the race tied in Michigan.
Silver posted on Twitter Monday afternoon that things aren't a "toss-up" in Michigan, and that there are "some REALLY bad polls there."
