Economics is only now beginning to come to terms with the external consistency demands posed by research in psychology, sociology and other related disciplines.
I’ve been thinking a bit more about science, pure and applied. My two criteria for “science-ness” thus far have been explanation (providing the causal mechanisms that generate outcomes) and giving priority to the minimization of Type I error. These have to do much more with pure science than applied, however. When a geologist assesses whether a site is stable enough to build on, she is concerned quite as much, if not more, with false negatives than false positives.
Where the two overlap, however, is in the role given to what can be called external consistency. Internal consistency is the property of logical coherence. Economists and others use math, for example, to test for it. An argument is externally consistent, however, if it does not contradict claims thought to be true made by other researchers, typically in adjacent fields. For instance, to be externally consistent a geological theory must adhere to the recognized results obtained in physics and chemistry.
The case for external consistency can be seen as a corollary to Type 1 error minimization: if your hypothesis butts up against arguments or evidence with enough credibility elsewhere, you are at heightened risk of being wrong. But what counts as enough credibility? There is no getting away from the fact that this is an elastic criterion. On one end we have the more-or-less indisputable results of real, fully tested sciences: if you violate any of them, you can’t be right. (I once had a student who proposed a project design for urban rainwater collection that would have violated the second law of thermodynamics. This was extreme.) On the other we have plausible results that have not yet been subjected to critical tests. How much credence we give them, and how we trade off this consideration against others, is a matter of judgment—one characteristic of the quality of a practitioner.
Economics is only now beginning to come to terms with the external consistency demands posed by research in psychology, sociology and other related disciplines. A good applied economist is someone who knows this work, can evaluate how compelling it is and how much impact it should have on the validity of economic models and methods.
4 comments:
This is the key theme in the work of Herbert Gintis. He sees game theory as a methodology to integrate and unify the social sciences.
Edward O. Wilson's "Consilience: The Unity of Knowledge," published in 1999, has a section on economics (p.p. 212-224) that starts with this:
"The enterprise within the social sciences best poised to bridge the gap to the natural sciences, the one that most resembles them in style and self-confidence, is economics. This discipline, fortified with mathematical models, garlanded annually by its own Nobel Memorial Prize in Economic Science, and rewarded with power in business and government, deerves the title often given it, Queen of the Social Sciences. But its similarity to 'real' science is often superficial and has been purchased at a steep intellectual price."
Might we say "The Queen is dead, long live ...."?
One of Wilson's goals with Consilience wass to apply priniciples of the scientific method (real science!) to certain of the social sciences. I wonder what Wilson might have to say regarding the recent and continuing financial crises. Frankly, when I read Consilience back in 1999, I thought his goals were lofty, well beyond his experience with ants.
"Scientists are taught not to answer questions, but, rather, to question answers."
--anonymous
This noble goal is frequently disrupted by the client who demands bankable answers from his hired haruspicers.
--ml
Martin, I don't think that I'd buy that definition. My own graduate schooling, a while ago, was in the field of experimental psychology. The field made a major effort to be as scientific as possible. The focus was on improving one's method of observation and measurement of the variables in question. Some modern day social scientists seem to think that the answers are in the mathematics brought to bear. That while ago the mathematics were seen only as a necessary tool, not as an answer.
Experimental design, control of variables, valid operational definitions and correct measurement procedures were the focus of training. Are these any longer important steps in the graduate training process?
Post a Comment