Why economists disregard economic methodology

Economic methodology has been strongly in!uenced by philosophy. Indeed, it can be considered to be a part of philosophy, as the normally used name ‘philosophy of science’ aptly conveys. The scholars engaged in this endeavour often have been educated in philosophy or at least have strong philosophical inclinations. They approach the subject in two steps: "rst they inquire what is the ‘right’ way of doing economics according to what they consider to be the ‘scienti"c’ standard, and then they exhort the economic researchers to act accordingly. Those who do not follow the principles laid down are considered not to be real scholars or to at least have a character de"ciency. The result of this approach is an almost total split between what economic methodologists prescribe and what economists do in their practical research. The gap is so vast that economists do not even take the trouble to reject the exhortations o¡ered by the philosophers of science. They simply disregard them. The picture of preaching philosophers of science, and economists totally disregarding them, is perhaps less surprising where fundamental issues are concerned. It would be too naive to expect practical economists to change completely because some ‘outsiders’ advise them to do so. An example would be the sort of issues addressed by practical economists, which tend to be selfde"ned within economic science rather than being reactions to what may be seen to occur in the ‘real’ world (see Frey and Eichenberger 1992). But the advice o¡ered by economic methodology is also disregarded when it comes to speci"c research procedures. A pertinent example are the problems arising when the ‘best’ econometric result is presented after having run virtually thousands of estimates. The statistical tests, in particular the t-values, are nevertheless presented as if the data had been used only once. This is mistaken (see the special issue of the Journal of Economic M ethodology 7, June 2000). Another example is the interpretation of ‘statistical signi"cance’ in empirical economic research in terms of substantive importance (see McCloskey 1997). A high t-value associated with an estimated parameter is taken to mean that Journal of Economic M ethodology 8:1, 41–47 2001