“The Case for Inductive Theory Building”

Professor of business management Edwin Locke sent me an article:

This paper argues that theory building in the social sciences, management and psychology included, should be inductive. It begins by critiquing contemporary philosophy of science, e.g., Popper’s falsifiability theory, his stress on deduction, and the hypothetico-deductive method. Next I present some history of the concept of induction in philosophy and of inductive theory building in the hard sciences (e.g., Aristotle, Bacon, Newton). This is followed by three examples of successful theory building by induction in psychology and management (Beck’s theory, Bandura’s social-cognitive theory, goal setting theory). The paper concludes with some suggested guidelines for successful theory building through induction and some new policies that journal editors might encourage.

Like most social scientists (but maybe not most Bayesians), I’m pretty much a Popperian myself, so I was interested to see someone taking such a strongly anti-Popper position.

My quick resolution of Locke’s and my ideas is to say that Popper is great on checking models (and Popper’s protege Lakatos is great on the larger question of how model checking fits into the processes of normal science and scientific revolutions), but that none of those guys had much to say about where new ideas come from. My current thinking is that new models are derived from a language-like process. Remember that writing is non-algorithmic. I’ve noticed the same sort of “aha” process associated with scientific ideas, also occurs when I write sentences and reformulate them to say what I want them to say. The fractal nature of scientific revolutions and all that.