Haavelmo's epistemology for an inexact science

The first two chapters of Trygve Haavelmo’s Probability Approach provide a very rich epistemological framework for understanding what is involved in finding laws outside the laboratory. Even though these laws will be inexact, a framework was developed to specify for which conditions laws could be found. Therefore, Haavelmo defined two different kinds of influences, potential and factual influences. To decide whether a regression equation represents a law, it is not the strengths of the factual influences that are relevant, but whether the potential influences are significant. Unfortunately, by passive observations only it is difficult to obtain knowledge about potential influences, particularly when they have not revealed their strengths (yet).