Larry Summers and The Wrong Questions: Ad Hominem and The Perils of Searching For The Wrong Causal ConclusionsPosted: June 13, 2013 | |
When Larry Summers was President of Fair Harvard, he dared to suggest that women just might not like math and science.
His logical process flowed something like this:
P: Women are participating in and producing scientific insights at a lower rate than men.
P: We’ve instituted programs to boost female participation that have been in place for multiple academic generations.
A: [Effects of social pressures, then, should have been largely ameliorated for girls that would have been interested in pursuing the sciences.]
C: Maybe women just hate science, or somehow aren’t biologically equipped to thrive in it.
The reaction was swift and merciless. Every response was that Larry was wrong because he was biased, and that he shouldn’t be allowed to continue spouting his hate – in essence, an ad hominem flanking maneuver designed to forestall further exploration of his conclusion.
Let’s ignore, just for a moment, whether Larry was right, and focus instead on the fallacies in the reaction. Ad hominem means “to the man,” and refers to an attack on the source of an argument rather than the merit of the argument itself.
Why, though, is ad hominem fundamentally a fallacy? Sure, it doesn’t speak to the validity of the conclusion, but why? To figure out what’s really underlying the fallacy, we need to ask a slightly reworked question:
Was Larry Summers’ conclusion correct? Notice how I didn’t ask if Larry was correct, although they admittedly coincide. We’re after the logic behind the statement, not the logic in judging the man. Ergo the reframe.
Unfortunately, the critics asked the other question: Was Larry Summers correct? And by the logic of the reaction, it doesn’t matter. Larry’s critics are following a different logical path:
P1: There are few women in science.
P2: If girls hear that they don’t like science or are innately unable to do science at the highest levels, they may not pursue science.
P3: Larry Summers told girls they might not like science or might be bad at science at the highest levels.
SC: Larry Summers’ question may have the effect of dissuading girls from science.
A: [Low female participation in math and science is BAD.]
SC: Larry Summers is causing BAD.
P4: We can stop BAD by castigating its source until it no longer has the ability to comment.
A: [Stopping BAD is GOOD.]
A: [The inverse of P2 is also true.]
C: If we silence Larry Summers, GOOD will occur, mainly because more girls may participate in science.
This reaction is problematic, in no small part because it never answers Larry’s question in the first place – which is not only the point of the academic endeavor, but also goes directly to the heart of his critics’ final assumption, and thus their ultimate conclusion. If we were to design an actual logical framework to test his conclusion, we’d do it like this:
1. Is Larry Summers’ conclusion true? In other words, are women actually less inclined towards science, or worse at it at high levels?
2a. If 1 is true, why?
2b. If 1 is true, should we attempt to raise female participation in the sciences?
2c. If 1 is true AND 2b is true, how do we best go about raising female participation in the sciences?
3a. If 1 is false, what in the data is suggesting 1 as a conclusion?
3b. If 1 is false, why else is there low female participation in the sciences?
3c. If 1 is false, how do we best go about raising female participation in the sciences?
Note that once you assume 2b, or that female participation in science is GOOD, the ultimate question converges. Aha! If the ultimate question remains the same, than Larry’s critics were right! The answer to his now-intermediate question doesn’t matter!
Wrong. Why? Because.
Because is the operative term here. Because indicates a causal conclusion. Look at 2c and 3c – our ultimate questions. Those are asking for causal conclusions: whatever factor is underlying 2a and 3b, once properly addressed, will cause a rise in female participation in the sciences. Conversely, properly dealing with the root causes can give us the necessary feedback to evaluate our conclusions, by testing our underlying assumption. Causal conclusions always have a built in assumption: this cause, and no other, is really the cause of this effect. And, to hammer a point home, we can’t properly conclude anything without showing our assumptions to be correct.
Look back once again at Larry’s critics’ reasoning above. See the real conclusion? Social messages cause girls to not participate in the sciences. Is it true? How would we ever know?
Here’s the real fallacy behind the Larry Summers ad hominem witchhunt: Larry was asking to find causal conclusions just like his critics were, but he was willing to explore all the data – questions 2a, 3a, and 3b – to find his cause. His critics, meanwhile, assumed his statements were their cause and then cut off potential inquiry into the subject, which left them no room to test the adequacy of their assumption.
Ironically, testing the robustness of a causal conclusion is, well, science. It must be excruciating to rail against a conclusion with venomous fervor, only to fail in the realization that you were strengthening it all along.
 For those of you paying attention, which logical fallacy am I committing in this statement?