A quasi-experiment is different from a true experiment. In a true experiment, a researcher randomly assigns subjects to a treatment or a control group. If the treatment and control group exhibit different patterns on some other variable (the "dependent variable," in technical jargon), that is because of the treatment. How do we know? Because the treatment was randomly assigned. In social science, and political science in particular, we don't often get to do true experiments. We can, however, look for ways to take advantage of "quasi-experiments." An exogenous shock will occur, changing the state of the world. Compare the before-and-after circumstances, and as long as the shock is truly exogenous, the difference you observe between before and after is the result of the exogenous shock.
Then, there are what we call
Donald Trump was so different from every other candidate that he was... a shock to the system. The question is whether or not he can be considered an exogenous shock. Not... really. He didn't come out of nowhere. From Newt Gingrich to Sarah Palin to birtherism... put all of that together, and the movement within the GOP towards Trump was there going back decades. I'm not going to say, "we should have seen it coming," but it isn't appropriate to say that he was a truly exogenous shock. However, he didn't exactly have uniform support, even within the GOP in the primaries.
Next, valence. The tricky thing about valence is as follows. If you ask Republicans right now, a lot of them really will tell you that they think that Trump is competent and honest. Why? Because assessments of valence characteristics are often endogenous. They are rationalizations for how we assess candidates, and that is mostly about partisanship and other factors. So, Republicans decide they like Trump. Within the GOP, Trump gathered supporters because he led the birther movement. These people then decide, because they liked Trump, that Trump must be competent and honest. The problem is, what's the direction of causation?
This is where quasi-experiments come in. In normal elections, with candidates who are at least vaguely comparable, distinctions are hard to make and direction of causation is hard to assess. Take any two normal candidates, and Democrats will say the Democrat is more competent and honest, and Republicans will say the Republican is more competent and honest. Beyond that, what do you do? If the candidates are comparable... you see the problem.
At the congressional level, there are real distinctions because most congressional challengers are schlubs, and there is some interesting work that has been done with expert assessments of the candidates to rate them on their valence traits. Lots of good work here by Walter Stone, and various co-authors. Check out his new book for a really good take on valence at the congressional level. The challenge there is that the discrepancies are big, but voters frequently don't know much of anything about the low-valence candidates.
The problem is, we need an election in which there is a really low valence candidate, well-known by the voters. What would happen? How would that candidate's party evaluate him?
What would happen if you found the single most incompetent, dishonest, sleazy, vile, possibly treasonous candidate... someone who clearly lusts after his own daughter, brags about committing sexual assault... just try to imagine any bad trait, and give the candidate that bad trait. Put it all together in one package. What would happen? Would voters of that candidate's party still rationalize those traits away and support him? Would those voters still say, in surveys, "yeah, he's actually competent and honest, at least relatively speaking?"
Of course, we have the endogeneity problem. Trump didn't come out of nowhere. The GOP nominated him because the Republican Party has been trending in that direction for decades, but from a research design perspective, the assessments of those who voted for candidates other than Trump in the primaries are still going to be informative. From a social science perspective, this is a tractable problem. This can be done. The analysis can be conducted.
So, what's the problem? Simple. The analysis rests on the common assumption that Trump is an exogenous shock because he is so uniquely vile, and such a clearly low-valence candidate that we must assume, as an analytic point, that anyone who rates him highly on any valence trait is simply rationalizing.
I am willing to do that, and to say that. I have never seen any signs of intelligence or competence from Trump. He lies more than any politician on record. The corruption in his campaign and White House is beyond comprehension. As far as I am concerned, these points are beyond dispute. Anyone claiming that Trump is competent or honest is not someone with whom I would bother to engage in a debate.
And yet, could I publish this? Nope.
Too bad.
I call this "Part I" because I suspect I'll be doing more of these.
