This shows that on average the polls in 2010 were actually a little more accurate than in 2012. What these numbers also show is that polling (on average) overestimates the GOP. I am by no means saying that all polls are "skewed" to the right. What I am saying is that polling does not in general "skew" left as many have suggested. What it means is that Democrats tend to out perform their polls more than Republicans do. It must be remembered that these polls are a representative sample of the day they were taken. These polls are not exit polls. Therefore, even if the sample used for these polls are all representative of the voting population: who they actually vote for can change. Also, voters that end up voting for Democrats tend to more likely to report themselves as uncommitted to a candidate to a pollster.
Regardless, in 2010 only two races were missed by RCP (both were GOP projected and ended up going for the Dems) and once in 2012 (as a note there was also a second one, North Dakota, where the Republican was also projected to win and then ended up losing.).