On the Interpretation of Density Dependence in Rates of Organizational Mortality.(Reply to Petersen and Koput)

This paper presents the authors' opinion on claims of Petersen and Koput (PK) regarding a problem in the usual interpretation of tests of the theory of density-dependent legitimation and competition as applied to rates of organizational mortality. PK argue that the negative first-order effect of organizational density on mortality rates is compatible with a simpler and thus preferable alternative explanation that relies on the operation of unobserved heterogeneity instead of legitimation. Their strongly stated assertions of generality fail to acknowledge that the artifactual effect of density occurred in only half of their simulations. Even substantively trivial effects are likely to be statistically significant when so many observations are generated by the simulation. Age of population and density are so highly correlated in the data sets produced by the simulation structure of PK that we could not estimate models including these two variables because we could not invert the Hessian. The most unrealistic feature of the simulation of PK concerns its treatment of founding processes. For unexplained reasons, they chose a deterministic founding process: The same fixed number of foundings occur each year in each frailty class.