%0 Journal Article
%T Power Analysis of Collapsed Ordered Categories with Application to Cancer Data
%A Arnab Kumar Maity
%A Jyotirmoy Dey
%J Calcutta Statistical Association Bulletin
%@ 2456-6462
%D 2018
%R 10.1177/0008068318803140
%X Ordinal data are often found in clinical studies. Sometimes the analysis of this data is done using binary logistic regression after collapsing the categories of ordinal responses. However, these analyses may not be appropriate in practice because either the assumptions are violated or because the information is lost. Cumulative logistic regression is shown to be a better alternative approach. The efficiency of cumulative logistic regression is demonstrated using simulation studies. A novel sequential testing approach is suggested in the context of cancer data. In addition, in the absence of the proper knowledge of the data, an automatic data-driven approach is also proposed. The efficacy of these proposals are illustrated via simulation studies. AMS 2000 subject classification code: 92D9
%K Cumulative logistic regression
%K logistic regression
%K non-proportional odds regression
%K proportional odds regression
%U https://journals.sagepub.com/doi/full/10.1177/0008068318803140