Trump might be right -- the GOP's House majority is not all but doomed

(CNN)Political prognosticators usually shake their head at President Donald Trump's electoral analyses. He doesn't tout all poll numbers, for example, and instead relies upon the rosiest figures from Rasmussen Reports.

But when it comes to the 2018 House of Representatives elections, I think the President may have a point. According to the New York Times, Trump disagreed with the assessment from White House legislative liaison Marc Short that "The GOP's House majority is all but doomed".
While Democrats are favored to take back the House, it's by no means a certain proposition.
Right now, Democrats hold somewhere in the area of a seven-point lead in generic congressional ballot polling over the last two months. Based upon movement in the generic ballot in midterm campaigns from this point in the cycle to the election since 1938, we'd expect the margin to remain roughly the same on Election Day 2018 (Democrats +7).
    A seven-point lead for the Democrats on the generic congressional ballot is far from a guarantee that they'd take back the House. In fact, my estimate based upon on how district lines are drawn suggest that Democrats would probably be a 50:50 proposition of gaining enough seats (23) to take back the House, if they won the national House vote by 7 percentage points.
    Now, it is true that Republicans are faring far worse in special congressional elections than you might expect in an environment in which they were down only 7 points on the generic congressional ballot. They have been outperforming the average partisan lean of the districts and states involved in special congressional elections by an astounding 17 percentage points. And as I have pointed out over and over again, there has been a clear correlation between special election results in midterm cycles and the House popular vote since 1994.
    Yet, keep in mind the sample size we're dealing with in special elections: nine. Although the signal from those nine has been fairly clear, nine elections is still a relatively small sample size.
    We're also only looking at races without incumbents when examining special elections. We know from history that incumbents tend to do better than the partisan lean of a district would suggest. About 80% of all Republican held seats up for grabs this fall will feature incumbents.
    To that point, take a look at some polling released last week by Patriot Majority USA (a pro-Democratic outside group) in 12 Republican held seats in which incumbents are running. The polls, which were not gold standard (i.e. did not call cell-phones) and do not meet CNN's polling standards, were meant to show Democratic strength in districts won by Trump.
    Yet, Democrats were only outperforming the partisan lean in these districts by 6 points -- significantly less than the special elections would indicate. That's not an awful result for Democrats given how powerful incumbency can be but given that these polls were released by a Democratic-leaning organization, it wouldn't surprise me if Republicans were in a bit better position in reality than these numbers showed.
    Indeed, the last time Democrats were doing so well in special elections, they didn't perform as well in the midterms as the special elections suggested. In 2006, they outperformed the partisan lean in special congressional elections by 15 points. That's a little more than the 17 points so far this cycle, though it is in the same ballpark. The end result in 2006 was that Democrats won the House vote by 8 percentage points. Democrats would still be favored to take the House if they won the national vote by 8 points, but it wouldn't be a sure thing.
    We can also look at district-by-district ratings produced by organizations such CNN, the Cook Political Report and Inside Elections. All of these rating systems point towards Democrats picking up seats, and the average outcome would be a Democratic takeover. But there's a wide range of possible outcomes given the predictiveness of these ratings at this point in the last few cycles. Anything from a Democratic pickup of roughly 10 to 50 seats would be in-line with these ratings.
    The national environment could certainly change over the next few months. But if we're looking at the polls and all the available evidence to us, the House is not a lost cause for Republicans at this time. Certainly not in the way described to Trump.
    Anyone categorizing the House as "all but" certain to be won by Democrats in 2018 is repeating the same mistake of 2016, when polls showed a Hillary Clinton win as more likely than not but far from inevitable.