Correspondence analysis of ranked data

In the past correspondence analysis has been generally applied to two-way and multi-way contingency tables. However, data is sometimes presented in the form of rankings, and the analysis of treatments and their rankings often needs to be made. To apply correspondence analysis to ranked data, Anderson's chi-square statistic is considered instead of the classical Pearson statistic. The alternative approach to correspondence analysis discussed by Beh (1997) is also applicable to rank data if the researcher is interested in how the ranks and treatments compare in terms of location, dispersion and higher order components. A brief discussion on the interpretation of the transition formulae of Beh (1997) is also made by considering the 3x3 bean example of Anderson (1959).