Repository | Book | Chapter

It is not always possible to test an hypothesis with an appropriately designed experiment that will isolate the experimental hypothesis as the only possible explanation of the experimental data. When a correctly designed experiment is not feasible, a common strategy is to conduct two (or more) experimental studies, each of which contains one flaw in design, such that the imperfection in research design varies from study to study. According to an argument termed here the "Strategic Combination Argument", if these different studies exhibit different defects of design, positive results in all the studies should inspire confidence in the experimental hypothesis because, in some sense, the separate weaknesses cancel each other out. It is shown here that the "Strategic Combination Argument" rests on certain assumptions about what hypotheses deserve consideration, the superiority of parsimonious hypotheses as explanations, and the absence of interaction between theoretical processes and features of the experiments. Nevertheless, when the hypotheses that rival the experimental hypothesis are falsified in the experimental studies, and not simply ruled out as irrelevant or inoperative, the power of the "Strategic Combination Argument" is enhanced, although it still requires the assumption of no interaction. In short, the "Strategic Combination Argument" has merit and may improve upon the evidence provided by any one weak study in support of the experimental hypothesis; but the argument does depend on arguable assumptions and can never have the evidential impact of one correctly designed study.

DOI: 10.1007/978-94-010-2259-0_16

Full citation: