Universal-deterministic and probabilistic hypotheses in operations management research: a discussion paper

Abstract

Based on Karl Popper’s approach to science, this conceptual article aims at promoting a discussion on important issues debated in the OM literature. Our focus is on the different types of hypotheses used in OM investigations and the implications of this for the formulation and testing of theory with an emphasis on case study research. Given that Popper’s methodology calls for an adequate testing of scientific propositions (universal-deterministic or probabilistic), we illustrate how case study research can be used to conduct a severe test of a scientific theory. It is also explained how case studies can be utilized to propose universal-deterministic hypotheses that should then be tested. Our investigation further discusses why it is important to clearly acknowledge the differences between the two types of hypotheses. Otherwise, inadequate research designs, deficient policy advice, and other similar problems, can arise. Topics for future research and discussion are also offered.

Keywords:

• Universal-deterministic hypotheses; probabilistic hypotheses
• methodology
• case study research
• case studies
• individual observations
• testing
• Karl Popper

Notes

Acknowledgements

We wish to thank Professor David Miller for his insightful criticisms of a previous version of our investigation. All errors, inaccuracies and the like are our sole responsibility.

Notes

1 Hypotheses with universal scope can also be expressed in the form of non-existence statements (e.g. ‘there is no’ statements). Therefore, a universal statement such as ‘all swans are white’ can be restated as ‘there is no non-white swan’. This assertion, evidently, negates the existence of swans that are non-white.

2 Although Popper does not deal explicitly with the topic of the evidence resulting from case study research, he explains in detail the role that the evidence found in individual observations has when testing universal-deterministic and probabilistic hypotheses. Moreover, in a seminal article, Flyvbjerg (2006) already proposed that the evidence found in case study research can have the same ‘operational’ value that individual observations have within the context of Popperian epistemology. Therefore, our likening of the evidence found in case studies to that of individual observations in Popper’s methodology is justified.

3 ‘… a severe test [is] … one in which the prediction [is] unlikely given the rest of your background knowledge’ (Dienes 2008a, 30). Also: ‘A test is severe if it makes the theory stick its neck out, that is, if it is likely to falsify the theory if the theory is false’ (Dienes 2008b).

4 Schmenner and Swink (1998, 106–107) also interpret Skinner’s model as a universal-deterministic theory: ‘A manufacturing plant cannot simultaneously have the highest levels of product quality, flexibility, and delivery, and the lowest manufacturing costs amongst all competitors’.

5 See for example Popper (2002) and Mellor (2005) for some comments on the possibility of deriving singular probability predictions/statements from probabilistic hypotheses.

6 For a detailed discussion on this and closely related topics, we recommend Popper (2002) and Miller (2016).

7 For example, when searching for documents containing the keywords ‘operations management’, ‘universal-deterministic’, ‘probabilistic’ and ‘Popper’ in several databases, we identified just over 40 documents. These results prompted us to do a second and more specific search using the keywords ‘case studies’, ‘operations management’, ‘universal-deterministic’, ‘probabilistic’ and ‘Popper’. This second search resulted in the identification of two dozen documents containing these keywords. The fact that only a few documents (out of hundreds of management articles dealing with these themes) refer to the topics specified in our two searches arguably corroborates our views.

8 In a series of investigations, Sarmiento et al. have done a thorough literature review and critique on the topic of strategic trade-offs. For example, Sarmiento (2013) concludes that there is no valid and credible evidence contradicting/refuting/falsifying Skinner’s model. To achieve brevity in this paper, we refer the reader to Sarmiento, Thurer and Whelan (2016) for a more recent analysis of the literature on strategic trade-offs.

9 By ‘independent’, we mean individual units that have not been subject to analysis already. For example, let us suppose that Skinner proposed his hypothesis on the basis of a detailed case study with one manufacturing firm in the United States (‘firm W’). A candidate for an independent unit of analysis that could be used in a legitimate and severe test of his theory would be any ‘non-firm W’ manufacturing system anywhere in the world.

10 Although Flyvbjerg is correct in affirming that according to Popper, evidence from a single case study can falsify a scientific proposition, he does not specify that a Popperian falsification can only occur when the scientific proposition is presented as a universal-deterministic hypothesis. For a detailed analysis of Flyvbjerg’s work, see Sarmiento, Whelan and Sprenger (2018).

11 One of the referees suggested that more research should be done in order to explore the possibility of utilizing case study research in the severe testing of probabilistic hypotheses.