I’ve been saying for a while that I think the Democrats will probably retake the House in 2018. More recently the probability seems to be getting higher and higher if you look at the generic ballot.
But I noticed something on Twitter and made an observation which I think perhaps I should put here: the conscious Republican gerrymandering after 2010 opens the possibility for greater Democratic gains because of tail risk. I was prompted to this comment after seeing a distribution of likely outcomes of the November 2018 election. The shape of the likely number of Republican/Democratic representatives wasn’t Gaussian. Rather, there was a much longer Democratic tail to the distribution. I hypothesized that this was the outcome of massive Democratic gains if the wave was high enough, and gerrymandering districts begin to overtop and flip.
The logic is pretty straightforward. Republican gerrymandering involves packing Democrats into some districts and dividing others between very Republican districts. The packing decreases the proportions of Democrats in some Republican districts. But the dilution of Democrats across very Republican districts, leaving somewhat less Republican, but still reliably Republican, districts, is where my point comes in.
If the national generic ballot swing toward Democrats is large enough, then some safe Republican seats come into play. Distribution of Democrats across these districts in a normal year does not entail anything more than a trivial shift in probabilities. But in a wave election, the standard operating procedure might not hold. If the Democratic votes were in a single district, then the Republican districts that remained would be more robust to a wave. As it is, removing these safe Democratic districts and distributing them across Republican districts made these districts a little less robust to a wave.