Correspondence Analysis

Correspondence analysis is part of a family of multivariate statistical methods which aims to visualize the main dimensions of a rectangular table of data, facilitating interpretation and understanding of patterns in the data. Within this family it is probably the most versatile technique, because it is primarily aimed at categorical data and incorporates the idea of weighting the rows and columns of the table. This method has long been used in the social and environmental sciences and is increasingly popular in marketing research, archaeology, psychology and linguistics. In this course, the first day will be spent giving a comprehensive introduction to simple correspondence analysis, emphasising the geometric concepts that underlie the method. On the second day we shall discover multiple correspondence analysis, which analyses several categorical variables simultaneously and which is used extensively in survey research. Finally, on the third day, some related topics are discussed, such as data coding for other types of variables (preference, paired comparison and continuous data) as well as the analysis of square tables.