Search

In this paper we build on previous research on affiliation bias at the group level. Drawing on recent theoretical work on Respondent-Driven Sampling, a network-based statistical procedure for sampling and analyzing hidden populations, we present a dual homophily model that divides homophily into two components, one due to affiliation preference and one due to differential degree. Affiliation homophily is a term that captures the extent to which affiliation is based on like associating with like; consequently it measures social differentiation. In contrast, degree homophily is a term that captures the extent to which differences in affiliation patterns reflect differential status, as reflected in differences in network sizes; consequently it measures social inequality. We derive formulas for each component and show how homophily can be exclusively expressed as a function of them. The paper concludes with an empirical example of how the dual homophily model can be used to enrich research on social networks and HIV/AIDS using a Respondent-Driven Sampling sample of 305 Latino men who have sex with men from San Francisco. [ABSTRACT FROM AUTHOR]

This paper is motivated by the observation that while both students and faculty are well aware of grade inflation, relatively few attempts at presenting a mechanism for grade inflation have been made. Furthermore, the mechanisms which have been suggested are largely macro level processes, such as University level free riding or department level compensation, for which a clear connection to micro level grading has not been made. This paper uses agent based modeling to consider the impact of three micro level processes and their interactions on the overall grade achieved and effort required to achieve that grade. We find that all three micro level processes (student effort adjustment, grading on the curve, and voting with feet) and their interactions lead to significant increases in overall grades and significant reduction in the effort required to achieve those grades. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]

Based on a pilot study of 159 university undergraduates, this paper presents Respondent Driven Sampling, a new method of studying hidden populations, as a viable method of understanding overall network structure. The method was found to be highly effective and efficient. Examples identifying bridge groups, centrality, and multiplex association are discussed. [ABSTRACT FROM AUTHOR]

Respondent-Driven Sampling (RDS) is a chain referral or "snowball"-type sampling and analysis methodology originally developed for studying hidden or hard to reach populations. The method goes beyond other snowball sampling techniques by incorporating sampling procedures and analytical tools that allow for the calculation of unbiased population estimates. Now widely used in studies of hidden populations, RDS data provide a wealth of information and potential for social network analysis by shifting the unit of analysis from nodes to ties in the network. This paper utilizes unique additions to an RDS dataset that allow the empirical evaluation of an RDS affiliation measure easily conducted with standard RDS data. Additionally, two techniques for measuring transitivity with RDS that require supplemental network data are discussed. [ABSTRACT FROM AUTHOR]