Most philosophers of science are empiricists. They believe that observational evidence is ultimately what should lead us to accept or reject the claims of scientists about the real world. Economics is a science of sorts. Though it may differ methodologically from the so-called harder sciences (e.g., physics), it is similar in that it is supposed to give us knowledge about the real world.
At a recent Show-Me Institute book club, I was discussing the methodological views of Ludwig von Mises with others. Mises’ view of economic methodology was essentially Kantian. Kant’s view relied on two distinctions: analytic-synthetic, and a priori-a posteriori. Analytic statements are definitional, while synthetic statements are about the world. The truth of a priori statements can be determined prior to experience, while a posteriori statements require experience to determine their truth. Here’s a trusty table I constructed from Wikipedia examples:
Analytic:”All bachelors are unmarried.”
Synthetic: “All bachelors are unhappy.”
A priori: “7 + 5 = 12.”
A posteriori: “Tables exist.”
Before Kant, no one believed that there were such things as synthetic a priori statements, i.e., statements about the world that can be determined prior to experience. Kant argued that the axioms of Euclidean geometry were examples. In his view, no one could argue that these axioms were false (since they were definitional) even though they revealed knowledge about the real world. Non-Euclidean geometry has not been especially kind to Kant; nor have philosophers of science.
Mises’ view is essentially Kantian; it merely treats the basic axioms of economics as his ‘synthetic a priori’ truths.
What I tried to communicate at book club was that the postulates of economics are not like the axioms of Euclidean geometry since they first require induction in their formulation. J.S. Mill (whose methodological views I tend to favor) was acutely aware of this when disseminating his methodological view all the way back in 1836. You can find a summary here.