"I estimated that Democrats would have to win the national popular vote by 2.5% in order to have a 50-50 chance of gaining control. I also predicted that the House popular vote margin would be D+0.0%, for Democratic gains of 2-22 seats . As of now, counting the leader in each undecided race, the new House will be
235 R, 200 D, a gain of only 7 seats.
ThinkProgress reports a popular-vote tally of 50.3% D to 49.7%, a margin of
D+0.6%. Both results are within range of my prediction.
However, this is quite notable. The popular vote was a swing of more than 6% from the
2010 election, which was 53.5% R, 46.5% D. Yet the composition of the House hardly changed and the party that got more votes is not in control. This discrepancy between popular votes and seat counts is the largest since 1950.
Did I underestimate the tilt of the playing field? Based on how far the red data point is from the black prediction line, the
structural unfairness may be higher as much as 5% of the popular vote. That is incredible. Clearly nonpartisan redistricting reform would be in our democracys best interests."
http://election.princeton.edu/2012/11/09/the-new-house-with-less-democracy/#more-8865