Heading into the 2010 midterm elections, Democrats’ incumbency advantage looked like a seawall against an electoral storm.

For the past decade, incumbency, along with the degree of partisanship in a district, has been the most important predictor of House elections. In 2002, 2004, 2006, and 2008, incumbents in both parties ran considerably ahead of their parties’ presidential candidates in their districts and ahead of the vote received by their parties’ candidates running for open seats in similar districts. In those elections the incumbency advantage ranged from 9 to 15 percent, a level comparable to the incumbency advantage from the 1970s through the end of the 1990s.

Although Democrats expected to lose at least 25 seats in the House last November, their low rate of retirements seemed to buoy their chances. Democrats had only twenty open seats; in 236 districts, an incumbent stood for reelection. Republicans had 21 open seats and 158 incumbents running for reelection.

Forecasting models that considered the usual factors—partisanship, the economy, and a ten-point average incumbency advantage—predicted at most a 45- or 50-seat gain for the Republicans. Instead, they gained more than 60.

What the forecasts got wrong was the incumbency advantage. Rather than the usual ten points, the advantage last year collapsed to just five points—as low as it has been measured since the 1960s. Strangely, the drop was the same for both parties in the House. Neither party’s incumbents performed at their usual level.

Even more puzzling, the Senate did not exhibit the same decline in the incumbency advantage. From 2002 through 2008, incumbent Senators of both parties enjoyed an average advantage of eleven points. In 2010 the Senate incumbency effect was ten points. (Others seeking reelection to statewide offices did not fare as well.)

What explains the change in a decades-long pattern? One possibility is that this was a one-time event. Much of the rhetoric leading up to the midterm expressed voter frustration with Congress, Washington, and incumbents. We have heard such rhetoric before—in 2006, for instance. But calls for ousting incumbents have usually been code for frustration with one party or another, rather than with incumbents per se. And in past years words have not matched actions: voters have chosen to keep their incumbents even in the worst of times. So the change may have been a one-time shock to the incumbency effect, a product of an unusually bad economy or unusually deep partisan division in Washington, D.C.

Yet 2010 also could portend a more lasting change in congressional elections. The infusion of party and independent expenditures has lessened the long-held campaign-spending advantages of incumbents, especially in competitive districts, and the rise of partisan discipline in the House has made congressional elections increasingly national affairs. Historically, incumbents have staved off the effects of national campaigns and ideological shifts by tending to a variety of needs of their constituents. But in 2010 voters were focused on national issues.

The decline in the incumbency effect means legislators’ activities on behalf of their constituents, such as earmarks and constituent services, provide less protection from short-term political forces than they once did. If these trends continue, we can expect control of the House to seesaw over the coming years—surely an opportunity for wealthy parties and interest groups to shape the composition of Congress.