This graph (click image for a larger version) shows, with the dark line, the number of US House seats that were won “marginally” in any given election from 1946 through 2004.
This takes a rather generous definition of what is “marginal”: Any seat that the party holding it won with no more than 60% of the two-party vote. The grey line is the number of seats the “out” party would have to win in the next election for control of the House to shift.
The trend shows clearly the decline in the number of relatively close seats in recent years, although it also shows that the current low number of marginal seats is not exactly unprecedented. There were 75 seats won with 60% or less in 2004, but there were 66 such seats in 1988. Following the 1988 election, there was a sharp uptick in the number of marginal seats, and when Republicans finally took the House in 1994, there were 162 seats that had been won at the previous election with less than 60%. The number of marginal seats has declined steadily since the Republicans took control.
While the low number of marginal seats in the current House is bad news for Democrats’ hopes of retaking the chamber, the good news for them is that the majority they are fighting against is much narrower than that which the Republicans faced heading into the 1994 election. In 1992, Democrats had won 258 seats and Republicans 176, meaning the out party needed to win of at least forty-two new seats. They got fifty four. (Actually, they took 58 seats from Democrats, but 4 other seats swung in the reverse direction.)
In 2006, Democrats need a net swing of only fifteen to retake the House (counting Bernie Sanders’ seat in their column, and assuming they hold it as he vacates it for his Senate bid).
If we break the marginal seats down into those held by the current majority versus those held by the “out” party, we see that the Democrats’ order is a bit taller than was the Republicans’ was in 1994, though this “tall order” is somewhat compensated for by the narrowness of the current Republican majority (noted above).
Despite the “sweep” pulled off by the out party in 1994, the Republicans actually won less than half the “marginal” seats. There were 90 seats that Democrats had won with 60% or less of the two-party vote in 1992, and Republicans won 40 of these. (They also won 18 seats that the Democrats had won with more than 60%.)
The national two-party vote swing from 1992 to 1994 was just over six percentage points. If swings were relatively uniform, we would expect nearly all the seats that Democrats held with less than 55% of the two-party vote to have swung. But even these more-marginal seats–46 in all–swung at a less than 50% rate: Republicans won 22 of them. This underscores the extent to which House elections in the USA are fundamentally not national partisan contests to the degree they tend to be in parliamentary systems: Even in a year branded by pundits as a “revolution,” the erstwhile “in” party kept a majority of its marginal seats. (I hope to post some parliamentary-election swings for comparison at some point.)
As for 2006, of the 75 seats won with less than 60% of the two-party vote in 2004, forty-seven of them are held by the “in” party (Republican) and 28 by the “outs.” When we look at the seats won last time with 55% or less, the two parties are defending an equal number of such close seats: ten each.
If the Democrats in 2006 were to win just over a third of the seats that the Republicans won in 2004 with 60% or less, they would retake control, provided they did not lose any seats of their own.
With Democrats running about 8-10 points ahead in most recent generic partisan polls and with the Democratic candidate having gained eight points in a (previously) very safe Republican district in the recent special election in California, an alternation in the majority party is feasible. But, given the low number of marginal seats being defended by the majority, the Democrats can’t be any less successful than the Republicans were in 1994 at nationalizing the election. And these data show Republicans were less successful at that task than the conventional wisdom might lead one to believe. Nationalizing congressional elections in the USA is just very hard to do.