Incumbent Advantages in U.S. House Races: The reason why most entrenched incumbents are safe in U.S. House elections is not simply that they represent districts leaning in their party's favor. Incumbency is an electoral advantage in itself. Incumbents typically gain that advantage through greater campaign spending ability, more press coverage, "franked mail" privileges, past campaigning in the district, more experienced campaign operations, and ongoing delivery of constituent services for their district. In addition, incumbents may have had opportunities to help shape the boundaries of their home district to their advantage.
We can quantify incumbency advantage by calculating the difference between incumbents' winning percentage and their districts' projected partisanship. In the seven U.S. House elections from 1996 to 2008, for example, incumbents on average had a winning percentage nearly seven percent higher than their districts' partisanship would have predicted-translating into a 67%- 33% win in a district that favors their party 60% to 40% and an average victory margin boost of 14% over a likely open seat result be in that district. While incumbent advantages continued in 2010-2012, they declined to less than 5% in both years- a smaller incumbency bump, to be sure, but still providing a margin advantage of more than 8% for an incumbent over what a non-incumbent of the same party would likely win in the district.
Measuring Incumbency Bumps to Determine Partisan Swings: Given that incumbents of both parties are able to secure the same incumbency advantages in comparable ways, the differences in incumbency advantage by party in a given election can be used as a tool to determine the underlying tilt in the nationwide partisan division in a given year. In 2008, for example, the median Democratic incumbent ran 10% higher than their district's partisanship, while the median Republican incumbent only received a 2% bump. We interpret this as a year in which the mood of the electorate was generally favorable to incumbents, with their percentage likely to be 6% higher than a non-incumbent of their party, but it was a better year for Democrats. Democrats ran 4% ahead of that norm and Republicans 4% below it, suggesting that the underlying partisan preference of voters that year was 54%-46% Democratic.
Support for the Parties in 1996-2012: In the basic preference of voters in these nine elections, Democrats have been favored six times, including in 2012, and Republicans three times. But only three times has one party had a preference edge of more than 52% of votes: Democrats in 2006 (53%) and 2008 (54%) and Republicans in 2010 (53.8%). The fact that Republican in 2013 will have a secure hold on the House is tied to the fact that the median district is a 52.2% Republican district. All things being equal, therefore, Democrats need to have a particularly strong national advantage to overcome that Republican bias - as they did in 2006 and 2008 - and then avoid another year like 2010 where Republicans are more easily able erase past Democratic gains.
See chart below summarizing Incumbency Bump analysis for 1996-2012. See a clearer pdf of the same chart here.
 District partisanship measures the degree to which a congressional district's partisan division of votes is likely to deviate from the national presidential partisan division.