Steven Horwitz recently wrote an essay on Cato Unbound on the issue of connection between the Austrian economic analysis and empirical studies.
Let us start with what Horwitz does not say. Contrary to many critiques that have already been made, he does not seem to say that necessarily true universal praxeological statements should or may be tested against the empirical data. He writes:
In short, Mises was making a Kantian claim about the human mind and the way in which minds are similarly structured across humans. We all have “a set of tools for grasping reality” that comes to us from our evolutionary heritage. The commonality of those tools allows us to engage in the reflection on action and the development of that core of economics as a set of necessary insights about how humans act. This core economic knowledge is not contingent but part of the very structure of human minds and is something that we can come to knowThe problem, however, is that, following Kant and Mises, Horwitz seems to view praxeological categories such as action as somehow distinct from the empirical world in the broad sense. But there is no reason for us to stick to this dualism and claim that the praxeological categories are somehow ontologically different from other things in the world we know.
There is, in other words, no need to ruminate on whether praxeological statements are truly apriori, what matters is that we know from introspection directly that they are true because we just cannot imagine things to be otherwise.
Adding to the fundamental praxeological statements some other universal truths about the world we live in, we arrive at fairly elaborate theories (which are part of a unified body praxeology) describing necessary causal connections between various human choices and their implications (though formulated, as Hulsmann has demonstrated, mostly in counter-factual form).
Then, comes the serious question of how we should use the theory (praxeology) in the analysis of concrete events. And it is at this point that Horwitz goes astray significantly, perhaps due to the Kantian dualism mentioned above:
However, beyond that, and especially including any claims about policy, economic arguments depend upon contingent claims about human behavior and preferences, the applicability of our assumptions, and the accuracy of our chains of argument. Good economics for Austrians means sound arguments, not just valid ones. Too much of modern economics consists of valid reasoning from false premises about human action.What Horwitz seems to be saying, is that when analyzing concrete events we should not just determine whether instances of certain praxeological categories are present in the phenomena at hand. We should somehow empirically (and that is where Horwitz's discussion about perceptions of the actors seem to fit in) demonstrate that praxeological categories "behaved" (Horwitz does not use this term) as they are supposed to "behave" in theory. But this is mistaken because if instances of praxeological categories are present they must behave in the way their behavior is described in theory.