A/Prof. Eric Beh

Biography

The primary focus of my research is on correspondence analysis, in particular the case where categorical variables consist of ordinal responses; this is often the case in many social and scientific studies - such variable structures may form part of (for example) responses from questionnaires and surveys. Correspondence analysis is a statistical procedure (often linked to the family of multivariate data analysis techniques) that allows for a graphical summary of the association between two or more categorical variables. While investigating the theoretical properties of correspondence analysis, I have paid particular attention to ordinal contingency tables.

My work with ordinal categorical data also extends to partitioning quantities commonly used to measure the extent to which two (and more generally multiple) categorical variables are associated. In particular, such partitions have been applied to the Pearson chi-squared statistic (for symmetrically related variables) and the Goodman-Kruskal tau index, Marcotorchino index and the Gray-William indices (all used for asymmetrically associated variables).

Other work I have undertaken, in collaboration with my colleagues, has resulted in an iterative free (direct) procedure for estimating the linear-by-linear association term commonly included in ordinal log-linear models. Work is currently being developed to look at the properties of these direct procedures and their extension to more general cases. I have also been very interested in ecological inference - a technique used to identify the behaviour of individual/cellular level data when only aggregate data is available. In particular, I am interested in the properties of a single 2x2 contingency table where the cells of the table are unknown or missing

Qualifications

• Doctor of Philosophy, University of Wollongong, 1999
• Bachelor of Mathematics (Honours), University of Wollongong, 1995

Research

Research keywords

• Categorical Data Analysis
• Correspondence Analysis
• Ecological Inference
• Measures of Association
• Statistics

Research expertise

My research interests cover the broad statistical topic of categorical data analysis, with particular emphasis on the mathematical development and application of correspondence analysis, measures of bivariate and multivariate association, log-linear models and ecological inference.

The primary focus of my research is on correspondence analysis, in particular the case where categorical variables consist of ordinal responses; this is often the case in many social and scientific studies - such variable structures may form part of (for example) responses from questionnaires and surveys. Correspondence analysis is a statistical procedure (often linked to the family of multivariate data analysis techniques) that allows for a graphical summary of the association between two or more categorical variables. While investigating the theoretical properties of correspondence analysis, I have paid particular attention to ordinal contingency tables.

My work with ordinal categorical data also extends to partitioning quantities commonly used to measure the extent to which two (and more generally multiple) categorical variables are associated. In particular, such partitions have been applied to the Pearson chi-squared statistic (for symmetrically related variables) and the Goodman-Kruskal tau index, Marcotorchino index and the Gray-William indices (all used for asymmetrically associated variables).

Other work I have undertaken, in collaboration with my colleagues, has resulted in an iterative free (direct) procedure for estimating the linear-by-linear association term commonly included in ordinal log-linear models. Work is currently being developed to look at the properties of these direct procedures and their extension to more general cases. I have also been very interested in ecological inference - a technique used to identify the behaviour of individual/cellular level data when only aggregate data is available. In particular, I am interested in the properties of a single 2x2 contingency table where the cells of the table are unknown or missing.

Collaboration

My research interests cover the broad statistical topic of categorical data analysis, with particular emphasis on the mathematical development and application of correspondence analysis, measures of bivariate and multivariate association, log-linear models and ecological inference. I am actively engaged in collaborations with national and international researchers including

• Professor Luigi D'Ambra, University of Naples, Italy
• Professor Thomas B. Farver, University of California - Davis, USA
• Associate Professor Rosaria Lombardo, Second University of Naples, Italy
• Dr Biagio Simonetti, University of Sannio, Italy
• Professor John Rayner, University of Newcastle
• Professor Derek Smith, University of Newcastle
• Dr Clovia Holdsworth, University of Newcastle

Fields of Research

Centres and Groups

Centre

Programs

• Bachelor of Mathematics
• Bachelor of Science (Mathematics co-major)

Courses

• STAT1060 - Business Decision Making