Latent Class Analysis Results for "Acquaintances" Data, Three Class Solution
|Parameter||Class 1||Class 2||Class 3|
|Note: αi = Prob (observation in class i); θi1 = Prob|
(Being able to confide in name/class i); θi2 = Prob (Name
being missed/class i); θi3 = Prob (Name being regarded
as a friend/class i); θi4 = Prob (Name being an active
particular good as judged by their chi-square values, the interpretation of the classes found is relatively straightforward and intuitively sensible. In the two- class solution the division is clearly into "close friends" and perhaps "just names"; in the three-class solution this division is again clear, and the additional class might be described as "friends whom one could not confide in." A full description of the example is given in Leff et al. ( 1990).
The analysis of categorical data is an important topic in many scientific disciplines and particularly in the behavioral and social sciences and medicine. In this chapter a number of useful methods have been discussed but the coverage has of course, been less than comprehensive. An area not covered at all is that of ordinal categorical data and readers are referred to Agresti ( 1984) for details. Again nothing has been said about measures of association for contingency tables; such measures are discussed in the series of papers by Goodman and Kruskal ( 1954, 1959, 1963, 1972).
Agresti, A. ( 1984). Analysis of ordinal categorical data. New York: Wiley.
Barnard, G. A. ( 1984). "Discussion of Yates (1984)". Journal of the Royal Statistical Society, Series A, 147, 449-450.
Cochran, W. G. ( 1954). "Some methods for strengthening the common x2 tests". Biometrics, 10, 417-451.
Cox, D. R. ( 1984). "Discussion of Yates (1984)". Journal of the Royal Statistical Society, Series A, 147, 451.