FWIW, since I wanted to know the Chicago Tribune's record in polling, I was given a link to this site that detailed it.
http://www.ourcampaigns.com/cgi-bin/r.cgi/PollingFirmDetail.html?&PollingFirmID=65
They have done 34 races and have averaged overstating the D support level by +4.6364 and have understated the R support level by -5.6526 (and overstated the L support level by +2.9702).
Seems to me that their record is not that great, imo.
Actually, a firm called Market Shares Corp. did the Chicago Tribune polls, and according to the website that you linked to, Market Shares has a +1.5316 D, -0.0601 R bias -- very close to Mason-Dixon's accuracy (+1.1260 D, -0.0119 R). The data on that site looks stale, however ...
