This course provides an overview of Categorical Data Analysis. The course will take place on three days. Each day will involve both theoretical and practical sessions, and during the latter the most important R packages for categorical data analysis are presented. Because of the history of Categorical Data Analysis, most of the applications will come from the Social Sciences or the Humanities.
On the first day, the fundamental concepts of categorical data are introduced, such as joint probability, conditional probability, marginal probability, and stochastic (in)dependence. The rational of Pearson’s chi-square test will be explained, followed by the introduction of the bivariate loglinear model (independence, saturated, linear-by-linear association, diagonal parameter model). Mosaic plots are introduced as a means to visualize the associations in a frequency table in a concise way.
The second day starts with the exposition of Log-linear Analysis as the technique for modelling the associations or (conditional) independence in multiway tables. The hierarchical principle in modelling will be explained, along with commonly used statistics for model selection such as BIC and AIC. Considerable time will be spend on interpreting parameters in the Log-linear model. Extentions of the model to the analysis of ordinal variables will be discussed, for instance linear-by-linear association.
The third day consists of explaining the relation of Log-linear Analysis to various Generalized Linear Models for categorical data. Attention will mostly be paid to Poisson regression and logistic regression, but Negative Binomial regression for over-dispersed data will also be covered.
February 20: Henri Dunantlaan 1, Gent (HILO-GUSB), PC room 1
February 21: Henri Dunantlaan 1, Gent (HILO-GUSB), PC room 1
February 22: Henri Dunantlaan 2, Gent (FPPW), PC room 2
Agresti, Alan (2013). Categorical Data Analysis. Third Edition. Hoboken, New Jersey: John Wiley & Sons.
- PhDs and postdocs of a Flemish University: € 0
- Other academics: € 180
- Non-profit/Public sector: € 300
- Private sector: € 600
Anyone working with categorical and/or cross-tabulated frequency data.
- Understanding of basic inferential statistics: hypothesis testing (e.g. t-tests), p-values, statistical significance.
- Basic skills in R (vectors, indexing, functions, ...).