(Lack of) Accuracy of election polls

Factual Questions
21 

Pollster          Dates              N/Pop  Stevens  Begich  Bird  Haase  Other  Undecided  Margin  

Hays Research (D)  11/2/08            400 A  42          49          -       -           2         7               +7D  
YouGov/Polimetrix  10/18-11/1/08  358   RV    42        44          -       -            -          -               +2D

Here are the last two polls given by Pollster.com. The YouGov Poll was tied within margin of error and the Hays poll had a large undecided factor. Either poll showed an even split in the voting between Stevens and Begich. If the undecided broke for Stevens at the last minute, which seems probable, then the polls were trending correctly. As I said earlier, polls are backward-looking not forward looking. People always seem to forget that.

22

I owe an apology to pollsters everywhere. I thought they were missing a lot of Obama supporters, but it looks like they got it about right. (Including state polls in Georgia.) So hats off!