NIOSHTIC-2 Publications Search
Min and max scorings for two sample partially ordered categorical data.
Sampson AR; Singh H
J Stat Plann Inference 2002 Sep; 107(1-2):219-236
For certain types of designed experiments the outcome variables are ordinal categorical variables. Such experiments are typical in many behavioral science and social science settings, as well as in certain types of clinical trials. The representation, analysis and summarization of the results of such experiments can be difficult and complex. We present one analytical approach to these settings that uses scalings for a simpler presentation of the results. The resulting scalings can be viewed as a technique for reducing the complexity of the multivariate ordinal response data and providing insightful data summaries. Utilizing some of the results for designed experiments with univariate ordinal responses, we introduce new techniques to handle responses taking values in a partially ordered set. These latter results are applied to scaling multivariate ordinal data. Several data sets are analyzed using this new approach.
Behavioral-testing; Behavioral-tests; Clinical-tests; Clinical-techniques; Analytical-methods; Statistical-analysis; Author Keywords: Partial order; Ordinal data; Scoring; t-statistic; Cochran-Armitage statistic; Stochastic ordering; Chi-square; Max-correlation
Issue of Publication
Journal of Statistical Planning and Inference
Page last reviewed: June 16, 2022
Content source: National Institute for Occupational Safety and Health Education and Information Division