A big Washington Poll Error

This speaks to what I was bringing up: a pre-election poll this week by a Republican pollster had Gregoire (D) over Rossi (GOP) in a tight 44-41% general election. But look at the actual primary totals from this week.

The actual election totals were about 470,000 Democratic votes to about 316,000 Republican votes, of about 899,000 votes cast. (The remainder are third party and dogged non-partisans - who knows what they're smokin') If it is a reasonable assumption that voters will stay with their primary party, then this is a very large underestimation of Democratic voters in this Washington poll.

Other factors you should consider: this is the first closed primary; primaries tend to draw out partisans; turnout which should have been down with a closed primary (see Bollenbach, 1986) was steady. Also, a recent poll had a solid but not overwhelming 8 point Kerry lead - but I find this curious. This is where I think Democratic turnout is being underestimated.