I know that when constructing poll averages and box scores, one is pretty much duty-bound to include any serious survey research that is not blatantly partisan. But I have to register my deep suspicion of the Reuters/Zogby polling. Every time it comes out, the numbers just look too weird to be a reasonable take of the electoral mode. I’m not talking too Democratic-leaning here, just weird. For example, in the latest numbers, Democrats have large leads in places where, if they are in the lead at all, it is likely to be a much smaller margin. Examples would be Illinois 6, Connecticut 4, and New Mexico 1. On the other hand, Zogby has Drake beating her Democratic challenger in Virginia 2 by eight points, which seems way too high.
Not buying it. I know these polls are not the Internet-based ones that lots of folks have problems with, but there still seems to be a screw loose somewhere in the machine.