Levels Issues I: Homogeneity and Heterogeneity
| Nicolai Foss |
Issues that relate to levels of analysis are some of the most vexing ones in social science, both theoretically and empirically. I plan to post on levels issues over the coming week or so. Today’s topic: Homogeneity and heterogeneity across levels of analysis.A prevalent idea is that concentrating on one level of analysis — say, the industry — implies that lower levels of analysis are claimed to be populated by homogenous entities. Thus, the industry analyst when conducting his analysis assumes that incumbent firms are homogenous (an argument that I have often heard adherents of the resource-based view launch against Porter style strategic management theory). Or, capabilities theorists must assume that lower level entities — that is, individuals — are more or less homogenous, an argument made by Teppo Felin and William Hesterly in their forthcoming AMR paper, “The Knowledge-based View, Heterogeneity, and New Value Creation.”
Sometimes the stronger argument is made that this assumption somehow follows with some kind of necessity; thus, analysis at one level of analysist must imply that lower level entities are homogenous (the much cited 1994 AMR paper by Klein, Dansereau, and Hall seems close to asserting this).
It is far from clear on what basis these arguments are made.
One possibility is that those who make these arguments are making an ontological claim. This is how I interpret e.g. Teppo and Bill’s paper. They essentially argue that capabilities theorists really think that people/employees are homogeneous (and can therefore be “rounded out”), so that all the action (e.g., competitive advantage) takes place on the level of capabilities.
Another possibility is that homogeneity is introduced for epistemic reasons, specifically the analyst’s bounded rationality — á la “We know that the micro-structure is exceedingly complex; therefore, we concentrate on macro phenomena, and we don’t lose any essential information by neglecting the heterogeneity of the micro phenomena — i.e., we might as well assume that they are homogenous”.
Both interpretations seem to me to make sense. What does not, however, make sense is the position that concentrating the analysis on one level of analysis must necessarily imply that the lower levels are populated by homogenous entities. For example, when economists do price theory the analysis takes place at the level of a market. However, analysis of individual agents and their choices are necessary to “build” the analysis. This does not imply with any degree of necessity that the analyzed agents are homogenous. Sure, it may be convenient to make this assumption and focus on a single representative agent (as in the Pigou/Robinson analysis of the firm in price theory), but there simply is no logical necessity involved.