I meant to comment on something Matt Yglesias linked last week: A Washington Post report on recent psychological research suggesting that power tends to stunt empathy. You might call this another instance of researchers confirming conventional wisdom—”power corrupts.” But there’s a still more general lesson that might be drawn from this and a raft of other similar research, which is that the perpetual emphasis on character in politics may be misguided, precisely because “character” is so much a creature of context.

An excellent book by John M. Doris, Lack of Character, makes the case against character in its strongest form—and, indeed, puts the case rather more strongly than I think the evidence justifies. Still, there’s a powerful body of evidence suggesting that if “character” is not quite a phantom, it is generally predictive only in relatively familiar situations, and that we are all prone to attributing to deep and permanent personality traits behaviors that owe far more to circumstance. In most cases, circumstance is stable enough that character is a passable heuristic: Ceteris paribus, someone with a reputation for honesty in business dealings probably will be honest in future business dealings. But President of the United States is such a unique circumstance that perhaps the heuristic breaks down there. If we’re not quite prepared to throw out character, then, we should at least be asking which data points about candidates are situationally relevant to the office and which are sufficiently rooted in radically different contexts that they can be ignored.