Elias Isquith took issue with the House part of it, in a post called "Jonathan Bernstein is probably wrong," which is an excellent way to get my attention! More posts called that, please. Anyway, Isquith:
While this is all plausible it’s sort of a longer way of acknowledging that Dems need to win the House vote by 7 percent to reclaim it, but insisting that could happen. And it could! But it’s certainly not normal, that kind of landslide election; and to point to 2006 and 2010 is to ignore the historic nature of 1. the Iraq War and 2. the greatest economic catastrophe since the Great Depression.I didn't give a lot of numbers in the original column, but since Vital Statistics is easily available now, how about the large out-party wins in percentage of national House votes, postwar, defining out-party here as the House minority party:
Again, I’m not saying developments on the order of that magnitude can’t happen — but they’re certainly not likely to happen. Barring some truly dramatic and historic convulsions, then, it’s fair to say the House is as it will be until a new census (and new state legislatures) change the parameters of the game once again.
Year/Party Margin of Victory
So depending on what Democrats really need for a majority, it's happened once, twice, or maybe five or six times since WWII. Does that make it "likely"? I don't know! We can play with the data all sorts of ways. For example: 1946, 1954, 1994, 2006, and 2010 are all midterms during periods of unified control of government. The other cases like that? 1962, 1966, 1978...and that's it. In 2002, there was unified control of House and White House, but not Senate. So: five out of the eight times there was unified party control at a midterm, the outparty had at least a five percentage point national win.
1982 might be relevant there, too -- Democrats had the majority in the House (with Reagan in the White House and a Republican Senate), but the conservative coalition came pretty close to controlling the chamber, winning several key votes. The result? Democrats won big, by 11.9 percentage points. Now, that's not quite the same -- the incumbency advantage was with them, not against them -- but it's also far more than 7 points.
What about the other scenario -- 1948, when a Democrat won the White House and brought the House with him? The minority party in the House during divided government retained the White House in 1948, 1956, 1972, 1984, 1988, 1996, and 2012, only getting one winning surge. But several of those had at least modest surges for the president's party. A scenario in which Democrats lose only a few seats in 2014 and then win the House in 2016 isn't likely, but it's plausible -- although it would surely be a very slim margin and one that would be at serious risk in 2018.
Then there's a Republican presidential victory in 2016 and retention of the House in 2018, and then a Democratic win in 2020. A party switch in the White House to the party with the House minority yielded a new House majority in 1952, but failed to do so in 1968 and 1980. The incoming president's party did reasonably well in each of the three, although none of them were anywhere near close enough to hit that 7 point apparent GOP bias that's opened up now.
Put it all together, and I'd say: a Democratic takeover in 2014 would be absolutely unprecedented. A Democratic takeover in 2016, with Democrats retaining the White House, is fairly unlikely, but certainly not impossible. A Democratic takeover in 2018 if Republicans win the presidency in 2016 is quite plausible, perhaps even likely; if Democrats won in 2016, it's very unlikely. If a Democrat wins back the presidency after a 2016 Republican win, then it's quite plausible. Oh, one more: what if Democrats really do win four presidential elections in a row? That's pretty unlikely, by historical standards -- which makes me think that if it does happen, there's a chance that it might actually indicate a real tilt towards the Democrats, which in turn makes it more likely that either 2016 or 2020 will be enough of a landslide to push the House to them, too.
I guess the main thing here is that the combined districting and incumbency bias are not as important as the political context -- the big reason Democrats are unlikely to win the House soon is because Barack Obama is president (and therefore the conditions for a Democratic surge in the House are unlikely). And that since political context is more important, and that's unpredictable out into the future, it's also hard to predict out-year House results with any accuracy.
And while it's not really relevant to the main point, it's worth pointing out again that gerrymandering is not a major factor here; gerrymandering is part of the districting advantage, which in turn is only part (with incumbency) of the GOP advantage right now, which in turn is less important than the political context.