ABSTRACT
This chapter examines a set of data on the number of females admitted into graduate school at the University of California, Berkeley. It describes polytomous data, that is, count data in which there are more than several possible outcomes. The chapter also examines comparisons between two samples of polytomous data, for example, comparing the numbers of females and males that are born in the different months of the year. It looks at comparisons among more than certain samples of polytomous data. The chapter considers a method of reducing large tables of counts that involve several samples of polytomous data into smaller more interpretable tables. Lancaster–Irwin partitioning is a method for breaking a table of count data into smaller tables. The reduced table involves all occupational groups but only the Protestants and Roman Catholics. In the collapsed table the occupational groups are unaffected but the Protestants and Roman Catholics are combined into a single row.
