Theoretical Contribution

Thomas Callahan: My girlfriend wrote a very interesting essay on who might have ordered the assassinations.
Gavin Vareek: Your girlfriend has a theory?
Thomas Callahan: Here, take a look at it. You'll get a kick out of how her mind works.
--The Pelican Brief

I first read Whetten's (1989) paper about what constitutes a theoretical contribution while I was in grad school. I'm reflecting on it here because I'd like to tighten up some conceptual research in process. Also, current genuflections made in deference to 'science' seem to lack understanding that legitimate science in any discipline is built on strong underlying theory.

When he wrote this, Whetten was editor of what remains arguably the most prestigious 'theory' journal in all of organization science. His primary audience was aspiring authors seeking to publish in the journal. His objective was to convey what he saw as the necessary ingredients of strong conceptual research so that author better understood what they needed to include in their papers to increase chances of publication.

Whetten suggested that a complete theory must contain four essential elements:

What. What factors (a.k.a. variables, constructs, concepts) should logically be considered in the explanation of the phenomenon of interest? The 'right' set of factors is both comprehensive (all of the relevant factors are included) and parsimonious (factors cannot be deleted without reducing value in the model). When formulating theory, it is usually better to err on side of too many factors initially as it is relatively easy to remove unnecessary or invalid factors as the theory is refined over time.

How. After identifying the set of essential factors, the researcher must explain how the factors are related. Figures that contain boxes (the factors) and arrows (the proposed relationships between the factors) help clarify the author's thinking and improve reader comprehension.

Together, the What and How constitute the domain or subject of the theory. When combined, Whats and Hows typically produce testable propositions. The difference between propositions and hypotheses is that propositions involve concepts while hypotheses require measures. Although they are not required for good theoretical contributions, propositions that are testable are generally very useful. They force the researcher to think about concrete applications of new or revised thinking, and they increase the likelihood that future researchers will use the theory to core arguments stemming from the model. When used, propositions should be limited to specifying logically induced research implications of the theoretical arguments. They should not re-plow old ground already confirmed by previous studies.

Why. What are the underlying economic, psychological, or social dynamics that justify the factors and the underlying relationships? The rationale constitutes the theoretical glue that welds the model together (to Whetten, model = theory). In theory development researchers, logic replaces data as the basis for evaluation. If they hope to have an impact, theorists must convince others that their proposed conceptualization makes sense. Moreover, if the research model is be interesting and useful, then all of the relationships in the model have not been previously tested, lest it is already 'classroom ready.'

Although propositions involving Whats and Hows can technically be tested without understanding the Whys in the model, this generally leads to empirically grounded discussions that lack conceptual foundations. Therefore, propositions should be well grounded in the Whys.

Together, What, How, and Why provide the essential ingredients of a simple theory: description and explanation.

Who, Where, When. Under what conditions does the theory hold, or not? Temporal and contextual factors set limits on generalizability and define the range of the theory. Although theorists cannot be expected to anticipate all possible boundary constraints, there is clearly value in conducting simple mental tests for generalizability of core propositions. How might proposed effects change over time? What if the phenomenon under consideration took place in a different land? While Who, Where, When is often fleshed out through subsequent empirical tests, theorists can add validity to their arguments by initiating the process of defining the theory's boundaries.

Most researchers do not generate new theory from scratch. Instead, they generally work on improving theory that already exists. What approaches can researchers take to extend theory in a meaningful direction?

What and How. When seeking to publish work in top tier journals, authors can rarely succeed by simply changing the Whats in an existing model. Adding or deleting variables rarely satisfies reviewers because the changes are rarely of sufficient magnitude to alter the core logic of the existing model.

That said, it may be possible to add a new factor that changes the accepted relationships (the Hows) among the Whats. Introduction of a new variable, for instance, might change the strength or direction of relationship between two existing Whats or might even obsolete a previously accepted relationship entirely. Theorists might keep an eye on empirical research that generates contrary or inconsistent results to motivate revision of models that are not completely satisfactory when subjected to test.

Why. Going after the Why is often the most fruitful, but also the most difficult, path to theoretical contribution. It commonly involves borrowing perspective from other fields that alters metaphors and gestalts that challenge underlying rationales of accepted theories. Whetten cites work that overlaps ecology and economics as an example here.

Who, Where, When. It is usually insufficient to point out new, previously undocumented conditions where an existing model either a) does not work well or b) performs as expected. In some cases, however, investigating qualitative changes at the boundaries could be theoretically interesting if something about the new condition suggests the theory should perform differently. The rule is that theorists need to learn something new about the theory itself when working with it under different conditions. New applications should improve the tool, not merely reaffirm its utility.

Three themes underlie these ideas for developing useful theoretical contribution (or theoretical critique). First, sufficient theoretical contributions rarely propose a single change to existing models. Interesting critiques focus on multiple elements, adding completeness to theoretical work.

Second, effective theoretical critiques should marshal compelling evidence. This evidence could be logical (e.g., the theory is not internally consistent), empirical (its predictions are inconsistent with data accumulated from past studies), or epistemological (its assumptions are invalid, given information from another field).

Third, good theoretical critiques should propose alternative or remedies. Critics are responsible for developing improved conceptualizations. Otherwise, it is difficult to know whether the extant model is inferior, or merely the best we can currently do in a complex world.

Advice that I hope to put to good use in upcoming projects. Would think any proponent of 'science' should aspire to do the same.

Reference

Whetten, D.A. (1989). What constitutes a theoretical contribution? Academy of Management Review, 14: 490-495.