Faculty Scholarship 1994 - Present
Continuous Dependent Variables and Organizational Ecology: Toward a more Perfect Union
In this paper, I review all previously publsihed organizatoinal ecology research which utilizes continuous development variables. I unearth twenty-one such studies, half of which were published within the past four years. The broadening array of dependent variables in this field is a most welcome development. However, each of these papers has at least one methodological limitation, in specification of cross-unit effects and/or controls for autcorrelation. Perhaps the most serious problem is the assertion that the fixed effects research design solves the problem of autocorrelation. I demonstrate that this assertion is untrue. I conclude with advice on the proper way to model continuous dependent variables in organizational ecology research, as follows: (1) Consider omitting all organizations which do not exist for more time periods than the number of independent variables. (2) Test for autocorrelation, report the results, and correct for autocorrelation if the test indicates that it is a problem. (3) Use a fixed effects model, and justify it based on the nonrandomness of the data.