The Similarity Structure of Organizational Populations

Abstract
Organizations are usually classified based on specific features, like their product structures, locations, sizes, or management structures. This paper builds on the view that beside these features, the perception of audience members should be taken into account as well. It proposes an Internet-based measure that captures the audiences' perceptions on similarity. Specifically, the perceived similarity of two organizations is measured with the overlap of pages linking to them. With the help of this measure, the perceived similarity structure of the U.S. higher education institution domain is analyzed. The validity of the measure is demonstrated by showing that it helps explaining the competition of colleges for undergraduates. Finally, implications of the findings and further research directions are discussed.

A Generalized Similarity Model for Relational Data

Abstract
A widespread approach to assess the similarity of objects is to measure the extent to which they have similar relationships to other objects or settings. For example, the notion of structural equivalence calls two persons similar if they have similar relationships to other persons. This paper generalizes this approach and views two objects similar if they have similar relationships to similar objects or settings. After proposing a geometrical representation for this generalized approach, we reanalyze two classic datasets: the Davis et al. (1941) data on social event attendance of 18 women, and the roll-call data of the U.S. Senate. We show that the proposed representation surpasses previous models of relational data, and illustrate how it opens up new possibilities for sociologists and social scientists in general.

Selective Sampling of Empirical Settings in Organizational Studies (with Jerker Denrell)

Abstract
Most studies in organization theory are retrospective and rely on historical data. Because more data are available about widely diffused practices or about large populations, studies typically focus on these. Using simulation we demonstrate that such selective sampling of empirical settings has important implications for two major research programs in organization theory. Diffusion researchers typically study practices that have diffused widely. We show that this implies that they will underestimate contagion effects. Researchers in organizational ecology often study populations that have become large. We show that such selective sampling of populations can generate spurious non-monotonic density dependence. We discuss the implications of such selective sampling of empirical settings and suggest ways to correct for the bias.

Similarity and attraction: homophily revisited (with Michael Macy)

Abstract
The principle that "likes attract" is one of the few empirical regularities that holds across the sciences. Organizational ecologists point to an exception - a high concentration of similar others can lead to competition for scarce resources. We extend the ecological effect to the pairwise probability of tie formation between similar organizations. Using Web data for U.S. colleges and universities, we show that the probability of a page link between schools increases with similarity, as the theory of homophily would predict, but only up to a point. Above a critical level, the effect of similarity reverses. Further analysis suggests that the reversal is due to the likelihood that highly similar schools see one another as competitors.

Niche Width and Scale in Organizational Competition: A Computational Approach (with Glenn Carroll)

Abstract
Two major components of macro organizational behavior's niche theories, the Principle of Allocation (PoA) and Scale Advantage (SA), contradict each other in some applications. While PoA claims that organizations with wide niches get punished, SA claims that large organizations gain an advantage because of scale efficiencies. Analyzing these theories implies a possible trade-off between niche width and size: if both PoA and SA are strong, then organizations must be either focused or large to survive, resulting in a dual market structure, as proposed by Carroll (1985)'s theory of resource partitioning. This paper builds a computational model to study this possible trade-off, and investigates the properties of organizational populations with low/high SA and low/high PoA. We first validate the model by showing that it generates three basic expectations for the "corner" solutions: (1) the dominance of large organizations in the strong SA setting, (2) the proliferation of narrow-niche organizations in the strong PoA setting, and (3) to a bifurcated or dual market structure if both SA and PoA are present. We then turn to questions about organizational action and environmental resource distributions. For organizational action, we aim to identify circumstances under which narrow-niche (specialists) or wide-niche (generalists) organizations thrive. For environments, we explore the claim that concentrated resource distributions are more likely to generate partitioned or bifurcated populations.

The Advantage of Rare Types (with Jerker Denrell and Tomasz Sadzik)

Abstract
Combining individual level learning with population level selection reveals a number of interesting phenomena. One of them is what we call 'the advantage of rare types.' The basic idea is straightforward: types that are rare have an inherent advantage over types that are common, because agents with rare types have more opportunity to learn how to act against agents of common types than vica versa. This gives a survival advantage to rare types, and leads to diversity. Although the idea is familiar to biologist under the name of 'frequency dependent selection' (Allen 1988), we found no game theoretic modeling of this phenomena, and no social science applications have been proposed and explored so far. This project takes on this dual task: to try to come up with a framework unifying the learning models and selection models of game theory; and to show how this mechanism explain social phenomena.