Bridging the Lab and the Field: Three Essays on Experiments in Virtual Worlds

Peter B. Twieg

Advisor: Kevin A McCabe, PhD, Department of Economics

Committee Members: Daniel E. Houser, Omar Al-Ubaydli

Vernon Smith Hall (formerly Metropolitan Building), #5075
April 27, 2017, 01:00 PM to 10:00 AM


In many traditional laboratory experiments conducted within the purview of experimental economics, subjects are recruited into a laboratory and presented with decision-making tasks that represent an abstracted version of some microeconomic system of real-world interest. However, guaranteeing parallelism from laboratory results is not an easy task. Because of these difficulties, experimentalists have explored a wide variety of methodological tools that make different tradeoffs between internal validity and external validity in exploring various research questions. This dissertation focuses on one such tool: The use of online virtual worlds to generate contextually-rich and naturalistic environments within which new microeconomic systems can be implemented and investigated by experimenters.

There are several major goals of the essays that comprise this dissertation. First, I wish to convey to the reader a basic familiarity with the sorts of questions which virtual worlds have been used to answer and an appreciation for why virtual worlds were considered as an optimal platform for studying these questions. Second, I want to provide would-be experimenters with some useful tools and advice for getting started with the seemingly-daunting task of implementing a virtual world experiment - I believe that the unfamiliarity with the actual software tools that virtual world experiments tend to rely on is a major obstacle to their widespread adoption. Third, I aim to provide concrete examples of some of the virtual world experiments that I have performed and how they were used to contribute to our understanding of how humans establish informal institutions to solve social dilemmas.

The first chapter of this dissertation provides an introduction to virtual world research, outlining the advantages that virtual world designs can provide in terms of allowing experimenters to implement microeconomic systems with a deeper degree of control and parallelism than what can be afforded by alternative methodologies. Several different types of virtual world experiments are outlined before the narrative focuses in on the type of experiment that I'm most familiar with: Laboratory experiments using the OpenSim platform. Much of this chapter is devoted to providing general advice concerning OpenSim experimentation as well as outlining a general framework for the implementation of an economic experiment within OpenSim. A crucial goal of this chapter is to help experimenters overcome the fixed costs involved with becoming proficient in this software and how to use it in order to conduct experiments.

The next two chapters outline two specific experiments performed within virtual worlds. In Chapter 2, I explore the determinants of effective commons-management institutions and demonstrate that providing groups with opportunities for costly specialization generates stronger territoriality in resource foraging. Within this environment, groups are able to build these institutions and maintain them by employing several of the institutional design principles outlined in the work of Elinor Ostrom, including natural language communication, mutual monitoring and costly punishment. These results are demonstrated through the use of spatial econometric techniques that indicate that subjects exhibit less overlap in their geographic foraging ranges when costly specialization opportunities are present.

In Chapter 3, I explore how networks of trust and trade emerge among groups under conditions where opportunities for specialization and exchange are present but access to productive opportunities are scarce. In this environment, subjects can only maintain property rights over these productive opportunities while they remain spatially proximate to them - in essence, subjects can only exercise squatters' rights over productive areas. Consequently, my analysis shows that producers are more likely to be either weakly linked or not linked at all when the process of finding exchange partners involves an elevated risk of losing access to future production opportunities due to these weak property rights.