While most of the attention focuses on the presidential race, over at the Princeton Election Consortium Sam Wang has an interesting analysis of the data on the relationship of the national vote for the two major parties in each biennial election to the number of seats that they ended up holding in the House of Representatives.
In the figure the x-axis represents the popular vote margin for the Democrats over the Republicans and the y-axis represents the excess of House seats. The dots represent the results for each biennial election since 1946, with the 2010 results specifically indicated. The grey area represent the range of the data and you can visually estimate that a trend line would pass close to the origin (0,0), which means that if the two parties tie in the national vote, they should get an equal, or close to equal, number of seats.
Unlike the case for the presidency, where because of the electoral college system the winner need not be the person who wins the majority of the national vote, you would expect that control of the House would depend upon it. However, the black straight line represents what Wang thinks is the current relationship between these two numbers and the red line represents his range of predictions for this November’s election based on the current polls. You see that these two lines do not pass through the origin. There is now a built-in 2.5% bias towards Republicans, meaning that the Democrats need to get 2.5% more of the national vote than the Republicans just to expect to break even in the number of House seats.
Wang puts this down to gerrymandering of house seat districts following the Republican election landslide in 2010 that put Republicans in control of the redistricting process in the majority of states. He says that this will be a feature for the next decade, until redistricting is again done following the 2020 census.