Same Technology, Different Outcome? Lessons on Dummy Variables & Dependent Variable Transformations

Abstract

There is long-standing body of empirical research concerned with the consequences of information technology for organization structure and processes. Several of those studies have reported that the same technology, when implemented in similar organizational settings, can be associated with vastly different, even diametrically opposing, organizational consequences. The seminal study in this stream of research is Barley's (1986) article entitled "Technology as an Occasion for Structuring: Evidence from Observations of CT Scanners and the Social Order of Radiology Departments." That study reported that two similarly-composed radiology departments implemented the same technology yet experienced different structural outcomes, i.e. that the two departments experienced different rates of decentralization and that they evolved through a different number of distinct phases of structuring. This difference in outcomes was attributed to differences between each departments' distribution of relevant expertise and "specific historical processes" (Barley, 1986:107) in which the technology was embedded. My reanalysis of the data uses different and arguably more appropriate research methods and shows that the failure to transform the dependent variable, as well as the exclusion, misspecification, and misinterpretation of several dummy variables, biased the regression estimates and led to erroneous conclusions.

The methodological contribution of this paper is that it underscores problems attendant to not recognizing two of the ways in which dummy variables can be interpreted: as a means for capturing intercept shifts and as a means for controlling for the effects of unobserved heterogeneity. The theoretical contributions relate to how the reanalysis impacts our understanding of the information technology -organizational structure relationship. In short, I conclude that research on the organizational consequences of IT, particularly ethnographic research, may need to (1) exchange the assumption of homogeneity among similarly-constituted organizations for one of heterogeneity (2) take both the observable properties of technology and its context of use explicitly into account and (3) and make more clear what is meant by "different structural outcomes