So why might this model not be enough for Democrats? There are three reasons. First, at least for now, Republicans are polling considerably better than they were in 2006. As of today, President Trump’s job approval sits at 44 percent in the RCP poll average – nearly eight points higher than Bush’s – while the Democrats’ lead on the generic ballot is in the mid-single digits. This suggests that despite the similarities, Republicans tend to be performing at least somewhat better.

Second, 2006 can’t be analyzed without acknowledging that it involved a number of fluke-ish victories for Democrats. These include: Tom DeLay (resigned under scandal, Republicans were unable to place a different name on the ballot); Bob Ney (convicted in a coin-trading scheme shortly before the election); Don Sherwood (accused of choking his mistress); Mark Foley (accused of having sexual relationships with pages); Sue Kelly (caught on camera running into the shrubs to avoid questions about Foley); and John Sweeney (accusations of domestic battery arising shortly before the election). There were also a number of semi-flukes, such as Curt Weldon having his congressional office raided by the FBI before the election, or Richard Pombo, who was caught up in the Jack Abramoff scandal. Regardless, if one doesn’t count the six “clear” flukes, the Democrats’ actual gain was 24 seats, which would barely be enough for them to win the House today.

Finally, there is the issue of exposure.