Essays on the similarity of organizations

This dissertation investigates how the similarity of organizations can be inferred from the similarity of their relationships to other organizations and settings. The dissertation consists of an introductory chapter, which argues for the importance of studying similarity in the social sciences, and three essays that discuss theoretical and methodological issues of conceptualizing relational similarity.;In the first essay, I review earlier studies on organizational classification and similarity measurement. Here, I conclude that there is a need for an audience-based organizational classification. I propose an Internet-based similarity measure for organizations. Specifically, I propose that the similarity of two organizations can be measured with the overlap of webpages linking to them. To illustrate the proposed measure, I analyze the similarity structure of the U.S. higher education institutions. The validity of the measure is demonstrated by showing that it helps explain the competition between colleges for undergraduate students. The second essay, co-authored with Michael T. Hannan, illustrates how the similarity of organizational categories can be measured by their tendency to co-occur. For example, we infer that the category "Argentine" is similar to the category "Steakhouses", because the organizations that are categorized as "Argentine" tend to be categorized as "Steakhouses" as well. We analyze the categorization of the on-line dataset, Yelp.com.;The third essay introduces a novel representation for relational data and relational similarity. This representation generalizes earlier measures of co-locational similarity. While correlation and structural equivalence measure similarity by the extent to which the objects have similar relationships to other objects or settings, the proposed model views two objects similar if they have similar relationships to similar objects or settings. I examine the behavior of the proposed similarity model through simulations. With the help of the proposed representation, I re-analyze two classic datasets: the Davis et al. (1941) data on club membership and the roll-call data of the U.S. Senate.