Bayesian analysis for sociologists

Journal/Book: Sociol Method Res. 1999; 28: 2455 Teller Rd, Thousand Oaks, CA 91320, USA. Sage Publications Inc. 7-34.

Abstract: This article provides an applied introduction to Bayesian statistics for sociologists. Unlike frequentist statistics, which attaches repeated-sampling frequencies to test statistics, Bayesian statistics directly describes uncertainty about unknown statistical parameters with a probability distribution With this foundation, much of Bayesian statistics follows from basic rules of probability theory Three areas of Bayesian statistics are especially relevant for sociologists. First, hierarchical regression models allow several levels of uncertainty into an analysis. Second Bayes factors provide a useful approach to the problems of model selection, model averaging, and posterior inference about model indexes. Third, recent breakthroughs in estimation methods offer valuable new tools for analysis of Bayesian models that were previously intractable.

Note: Article Western B, Princeton Univ, Princeton,NJ 08544 USA

Keyword(s): CHAIN MONTE-CARLO; MULTILEVEL MODELS; SOCIAL-RESEARCH; STATISTICS; UNCERTAINTY; INFERENCE; SELECTION

Zurück | Weiter

Bayesian analysis for sociologists - An introduction