Actions for ESTIMATION AND TESTING FOR TWO-WAY CONTINGENCY TABLES WITHIN ORDER-RESTRICTED PARAMETER SPACES

Email RIS file
Bookmark
Report an Issue

ESTIMATION AND TESTING FOR TWO-WAY CONTINGENCY TABLES WITHIN ORDER-RESTRICTED PARAMETER SPACES

Author
LEMKE, JON HAROLD
Physical Description
180 pages
Additional Creators
Pennsylvania State University
Access Online

Availability

I Want It
I Want It

Finding items...

Summary
Two different classes of models for two-way contingency tables are investigated under assumptions of orderings of the cell probabilities. The first class of models is for I X I square tables with row and column response variables which have the same ordinal categories. These models place structure on the odds of the pairs of cross-diagonal cell probabilities. The assumption that the cross-diagonal-odds are all at least one provides the order restricted parameter space. The second class of models is the class of log-linear models for rectangular tables with two response variables. Here, two particular types of order restrictions on the cell probabilities are investigated; they are the assumptions that (i) the modal cell is known, and (ii) the probabilities are monotone within each row and within each column.
For each class of models the maximum likelihood estimates are obtained by maximizing the likelihood of a given model while simultaneously restricting the estimates to the order restricted space. The likelihood-ratio test statistics investigated have asymptotic distributions which are chi-bar-square distributions.
Other Subject(s)
Dissertation Note
Ph.D. The Pennsylvania State University 1983.
Note
Source: Dissertation Abstracts International, Volume: 44-05, Section: B, page: 1512.
Part Of
Dissertation Abstracts International
44-05B
View MARC record | catkey: 13611958