A Misguided Philosophy of Science

PrawfsBlawg: A Misguided Philosophy of Science:

During my first year as a eco­nom­ics grad­u­ate stu­dent, I spent at most two min­utes think­ing about the phi­los­o­phy behind empir­i­cal work. On the first day of my year-​​long econo­met­rics sequence, our pro­fes­sor quickly reminded us that hypothe­ses can­not be proven, only dis­proven. That was it. I don’t even think Karl Popper’s name came up. This is sim­ply not an issue that social sci­en­tists wres­tle with. Which is a prob­lem, since what we do is not what we think we do.

We think we are engaged in Popperian refu­ta­tion. Popper’s the­ory is rel­a­tively sim­ple to explain: Induction and con­fir­ma­tion are impos­si­ble, and all we can do is refute hypothe­ses. In other words, it is impos­si­ble to prove that all swans are white, no mat­ter how many white swans I see. In fact, to Popper, each addi­tional white swan pro­vides no addi­tional con­fir­ma­tion of that hypoth­e­sis. But all it takes is one black swan to refute the hypothesis.

The appeal of Popper’s appo­rach is that it avoids the prob­lems of induc­tion, known to us since the time of David Hume. Popper’s is a purely deduc­tive logic. Our the­ory makes pre­dic­tion X, we see that X is not so, so our the­ory is — and log­i­cally must be — wrong.

But the real prob­lem for me is that the social sci­ences are not engaged in the Popperian endeavor. If noth­ing else, our the­o­ret­i­cal mod­els are incom­pat­i­ble with it. Compare crim­i­nol­ogy to physics. In crim­i­nol­ogy, we may be able to make a guess about the direc­tion of the effect, but that is all: “more peo­ple in prison will lead to less crime” is the best we can do.

Physics pro­duces gen­uinely testable pre­dic­tions. The social sci­ences do not.

There is much more detail and explo­ration in the orig­i­nal post about the impor­tance that bring­ing more meta-​​analysis or “evi­dence based” prac­tices to the social sci­ences. His con­clu­sion that more rig­or­ous review is needed in the social sciences:

[O]ur fail­ure to pro­duce them is a fun­da­men­tal epis­te­mo­log­i­cal fail­ure. A sin­gle study can refute, but only an overview can con­firm. And con­fir­ma­tion is what we do. The explo­sion in empir­i­cal work makes such overviews all the more impor­tant, since the larger the lit­er­a­ture the harder it is to see the big pic­ture, espe­cially with an ever-​​growing pool of poorly-​​designed stud­ies mud­dy­ing the waters. Review essays are the very heart of empir­i­cal knowl­edge, and they should be treated as such.

An inter­est­ing and impor­tant begin­ning to look­ing into some of the more hid­den recesses of the social sci­ences, which have always struck me as “try­ing too hard” to jus­tify them­selves as “real sci­ence,” and thus (at times, not always) tend­ing to avoid self-​​reflection or analy­sis. I look for­ward to read­ing more on this topic.