Discovering concepts Principles of theory-building research

In theory-testing, the concepts of a proposition are operationalized in procedures that specify how they can be validly and reliably measured see Appendix 1: “Measurement”. Although it looks as if the criteria of measurement validity and reliability do not apply in theory-building because the concepts emerge from the data, we consider it essential to good theory-building research in comparison to mere exploration that the emerging proposition is proven to be true in the instances of the object of study that are selected for the study. This implies that each concept that emerges from the research must be defined pre- cisely after its “emergence” and that it must be assessed whether the data, on the basis of which the concept was developed, can be con- sidered valid and reliable indicators of the value of that concept in the instances studied. Measurement validity and reliability are, thus, equally important criteria for the quality of theory-building research as they are in any other type of research. The same kind of reasoning as applied here to the measurement of concepts which are discovered and described in the study itself applies to the type of relations between concepts that is discovered in the study. If a proposition is developed in theory-building research, it should be demonstrated that the proposition is true in the instances from which scores were obtained internal validity. This implies that the result of a theory-building study is not only one or more new propositions but also an initial test of them within the study.

8.3 Research strategies in theory-building research

We have described how the type of proposition either deterministic or probabilistic determines which research strategy is preferred in theory-testing Chapter 4. However, the reason why we want to design a theory-building study is precisely because we do not yet have any proposition. How could we make a reasoned choice for one specific research strategy? We think that the most important criteria for this choice are efficiency and convenience. Because the only aim of the study is to generate propositions that need to be tested in further studies anyway, it makes sense to keep the theory-building study as simple and cheap in terms of time and costs as possible, i.e. at the minimum level of investment that is necessary to generate some relevant propos- itions or concepts. This is consistent with our preference for exploration in situations in which propositions need to be formulated. Chapter 8 The relative efficiency or convenience of different research strategies will differ for different topics or phenomena. But in general we advise selection of an appropriate research strategy in the following way: 1. decide whether experimental research would be useful and feasible, if not; 2. conduct a theory-building comparative case study. Regarding point 1, above, theory-building experimental research is useful in principle in two following situations: ■ if an independent concept A is known and an independent concept B must be found; and ■ if both the independent and dependent concepts are known but not yet the type of their relation. In these two situations an experiment could be designed and conducted in which the value of concept A is manipulated and the effects are observed. If the dependent concept B is known, its value will be meas- ured in the different experimental situations defined by different val- ues of the independent concept A. If the dependent concept B is not yet known, it must be discovered first. The value of the independent concept is experimentally varied and the experimenter attempts to dis- cover interesting differences between the different experimental con- ditions as well as with the control condition, if any. As mentioned earlier in the context of theory-testing research Chapter 4, experimental research is usually not feasible in business research. If an experiment is not feasible, then the principles of convenience and efficiency point to selecting only a small number of instances for observation point 2, above. On the other hand, a minimum number of instances is required for several reasons. First, if the researcher does not know what the relevant factors or effects could be and does not know how the concepts in the resulting proposition will be related e.g. in a deterministic or probabilistic way, it must be discovered whether there is a range of relevant causes or effects and, therefore, a range of diverse instances needs to be compared. But, second, if it is already known from the start that there is some evidence for a deterministic relation, an effect must be found consistently in more than one case in order to find a can- didate sufficient condition or a cause in more than one case in order to find a candidate necessary condition. In order to find other types of candidate relations such as a deterministic relation or a probabilistic relation even more cases are needed – three is the bare minimum. This means that the comparative case study is the preferred research strategy.