Towards a Comprehensive Approach to the Analysis of Categorical Data.
CARNEGIE-MELLON UNIV PITTSBURGH PA DEPT OF STATISTICS
Pagination or Media Count:
Current methodology for the analysis of categorical data can be traced to two papers published in 1900, one by Pearson and one by Yule. The keys to the linking of their ideas have been the use of linear models and inferential tools due to Fisher. After 50 years of research, statisticians have come close to developing a comprehensive approach to categorical data problems that stresses three basic themes interpretability, flexibility, and computability. This paper surveys the evolution of this comprehensive approach and classes of problems for which it has proven useful. A concluding section contains some speculation on unsolved methodological problems of current interest and on future developments. Author
- Statistics and Probability