Sciences are distinctive in many ways, but they all have in common the drive to minimize Type I error. This, you will recall, is the risk of accepting a hypothesis when it is actually false, as against rejecting it when it is actually true—the Type II variety. Sciences are progressive, advancing over time, because they have systematic procedures for expunging falsehoods. Other fields of human achievement have much to offer, but they lack this particular trait. Biology in 2011 is “better” than biology fifty years ago in a way that music or politics isn’t.
If you are still wondering whether economics should be considered a science, think about all the articles you’ve read that claim to be testing theories, and where the key language is “is consistent with”. “The starred coefficients in Table 9 are consistent with the properties of equation 18", “The greater incidence of such episodes in countries in Panel A is consistent with the predictions in our model”, etc.
The long form of “is consistent with” is “we should be more willing to accept this theory because it could be explaining the data”. The short form of this long form is “rejecting this theory risks Type II error”. Remarkably, most economists think this approach is what makes economics scientific.
If you take Type I error seriously, you have to ask, does the evidence preclude any other explanation? Am I at risk of accepting a false explanation because there is another which is actually correct? In practical terms, this means two things: taking all plausible explanations into consideration and not just the one you want to support, and searching aggressively for all the elements in your data that might contradict your pet theory. This second admonition includes examining subsamples whenever feasible, for instance.
Because economics, as it is practiced, is more concerned about Type II than Type I error, it propagates and defends a vast array of dubious propositions, and there is little methodological resistance.