Self-Reported Property Crime Patterns in 30 Countries: Adventures in Multivariate Exploratory Data Analysis
In this article, we investigate the use of a multivariate exploratory data analysis technique, nonlinear principle components analysis (NPCA). The technique is presented as a means for thorough descriptive analysis in a multivariate setting and with data of varying levels of measurement. The technique includes a strong visualization component of the kind previously only available for only one- and two-variable situations. This multivariate technique makes possible more thorough pre-analysis examination of data, a step frequently overlooked in most quantitative research. The data examined are self-reported prevalence measures for property crime from the Second International Self-Report Delinquency Study (ISRD-2). Our examination suggested that: (a) the property variable seems to group around two underlying dimensions (“Simple Object Dimension” and the “Complex Object Dimension”); (b) hacking proved to be a potentially very interesting but yet problematic variable due to interpretation of the item by the youthful respondents; and (c) the groupings and subgroupings of variables do show distinct nonrandom relationships among the 30 country property crime profiles in the analysis as well as the Esping-Andersen/ISRD-2 clustering system widely used in ISRD-2 research efforts.