Hypotheses Testing in Economic Science

Abstract

Formulating and testing hypotheses is the heart of the scientific method to most agricultural economists. Yet formal hypotheses testing is now rarely found in agricultural economics literature (King, p. 840). This note identifies some reasons why formal hypotheses are not used more widely in economic science. It also suggests future directions for hypotheses testing. Various philosophies of science define the role and use of hypotheses, the most important being logical positivism. It holds that the approximate nature of reality can be learned from observing behavior in a world of unknown and unknowable true causal relationships. Because true causal relationships never can be observed, our estimated economic expressions always contain stochastic error along with systematic relationships. Economic science as an evolving set of unrejected hypotheses is insufficiently descriptive of the real world to allow us to prespecify our empirical models precisely. This raises questions about using statistical procedures for selecting as opposed to testing hypotheses.