Files
Abstract
The purpose of this study was to investigate the ways in which social science data can be collected in virtual environments. In particular, this project focuses on sampling strategies when the population characteristics are not known, how virtual worlds residents may or may not differ from a nationally representative sample of Americans, and the factors surrounding how virtual worlds residents construct their identity and the relationship that has with physical and mental health. Using the largest virtual world in existence, Second Life, as the virtual platform for data collection, respondents were asked a series of demographic and health-related questions. The most significant contribution of this project is the methods used to design a quasi-random sampling frame for individuals in Second Life. A sample of 297 virtual worlds residents completed a survey instrument that contained replicated questions from the NHIS 2009 adult core questionnaire as well as questions designed specifically for this project. Results suggest that attempts at random sampling in virtual environments may not be worth the extra time and cost over convenience sampling methods since the differences between the two samples appear to be relatively slim only choice of survey language and primary language spoken were significant at the .05 level. When SL respondents are compared to a nationally representative sample of US adults, using the NHIS 2009 adult sample, there are significant differences in age (the NHIS sample is about 13 years older, on average), gender (there are more males in the NHIS sample), and physical and mental health indicators but no significant difference in income levels between the two groups. When age is controlled for, gender remains significant but the differences in health conditions largely disappear. This suggests, at least preliminarily, that residents of virtual worlds may not be that different from the population as a whole.