Are special elections like primaries, where turnout does not carry over to general?

Time to throw a new pitcher of cold water on the idea of there being a blue "wave" coming for the midterms, based on the outcomes of the special elections held so far -- supposedly a harbinger of the enthusiasm gap that will wipe out the GOP in the fall.

Of course, the GOP will do worse in at least the House and maybe just tread water in the Senate, but that doesn't mean either will flip, let alone by a large margin. We see this from the history of midterms for disjunctive presidents like Trump, Carter, Hoover, Cleveland, Buchanan, and Quincy Adams (those who were the last of their era, attempting to re-orient their ossified party, but failing and getting supplanted by an entirely new paradigm by their rival party).

Nor does polling support a "wave" for Democrats, who only enjoy single-digit leads in the generic ballot.

The main empirical basis of the "wave" narrative is the outcome of the special elections held so far (including state-level), and the other forms of collective action like the protests (women's march, anti-gun march, etc.). Democrats are so fired up that they're "swinging" their districts or states by so-many points from the 2016 presidential election -- just imagine the "wave" in the fall if even half of that swing holds!

The problem is that these special elections are not ordinary -- they're not the general election held in the fall of an even-numbered year. They're out of place in the year, or in the wrong-numbered year. They are too early to qualify in most people's minds as election-elections.

During the 2016 campaign, I discussed why primaries and generals are independent of each other, and why primary turn-out does not predict general turn-out over the years. Briefly, primaries are early-stage behavior -- and if there's nothing exciting going on early, few will show up. But when the "real" elections take place, when voting "really" counts, a whole lot of voters will show up who had tuned out the early activity.

It's possible that these special elections are functioning like primaries -- allowing people to participate early if there's something exciting going on within their side, or giving people a reason to keep sitting on the couch if it's boring on their side. Here, it is not the internal electoral battles to decide the future of the party, but just any reason to get out of the house and take part in collective action for your party, or against the other party.

As with primaries, though, just because a lot of people turn out early on doesn't mean so many more will turn out on the real date. Maybe a lot of folks who were already going to vote Democrat in the fall are turning out early because they're highly motivated to take collective action against Trump and the GOP. And maybe a lot of Republicans are eventually going to turn out in the fall, as they routinely do, but are just sitting at home early on because there's nothing exciting for them to do at this preliminary stage.

Or maybe the specials will resemble the ordinary elections. In the primaries, there is simply no correlation one way or another with early turn-out and general turn-out. My hunch is the specials are like primaries, where over the years there is no correlation one way or the other, and the early stage and final stage behaviors are independent of each other.

One of my unpaid interns out there somewhere can go through the historical data to test these ideas. Start by separating presidential from midterm years. Check which party did better during special elections -- either the partisan gap per se, or how the gap changed from the previous ordinary election. Then check the partisan gap per se in the next ordinary election, or the swing in the gap from the last ordinary election. There aren't a whole lot of these specials at the national level, and getting the greater range of cases at the state level would take some hunting around for the data. But someone could look into it.

Although I'm not empirically checking to see if my analogy is right between primary and special elections (both being "early"), neither is the side that just assumes the special elections are a harbinger of the ordinarily scheduled elections.

My argument is more congruent with the other signs we have of what will happen, so I'll stick with that until someone goes back over many years of data and shows that performance in specials predicts performance in ordinary elections.

If you're only going to pick one midterm season to investigate, it would be 1978 or 1930 -- the most relevant to this year's, as midterms during a disjunctive presidency.