Four weeks before Election Day on October 2012, they had Tammy Duckworth, Bill Enyart, Brad Schneider, and Bill Foster all losing narrowly in Illinois, and Cheri Bustos ahead by a fraction (she won by 6.6 points).
A few days before Election Day, they adjusted their numbers to get a lot closer to the final on most of the races although they still had Schneider losing and flipped Bustos to losing by 2 points.
Regardless of how they did on Obama-Romney, Nate Silver said that WAA has a Republican house effect. Search here: http://fivethirtyeight.blogs.nytimes.com/2012/06/22/calculating-house-effects-of-polling-firms/
That was before the 2012 election though. House effects change every cycle. That being said, I'd probably peg WAA as a bit R-leaning, but I'd say this: I don't consider themselves particularly precise (or particularly accurate), so bias is less important than average error here.
I still maintain that Oberweis is in high single digits/low double digits territory at this point, because of Quinn's unpopularity.