BIVARIATE PROBIT, LOGIT, AND BURRIT ANALYSIS
Abstract
The problem of a mixture of two stimulants in a biological quantal assay is investigated from a mathematical standpoint. The basic assumption is made that the response region does not depend on biological considerations - i. e., given a specified mixture of stimulants z, the response region is defined by the point z' in the p-variate space where there are p stimulants under consideration; instead, the probability functions, themselves, may take on different forms. A general form is proposed and investigated. Three analytic models (one utilizing the bivariate normal distribution, one a bivariate logistic distribution developed by Gumbel (1961), and one a bivariate Burr distribution developed by this author) are employed in this investigation. The investigation includes the analysis of data, under the three analytic models, which had been classified by previous investigators as examples of synergistic action, simple similar action, independent action, and additive action. The residual analyses are included as well as the FORTRAN IV subroutines used in evaluating the functions, the partial derivatives and the weights.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 08, 1969
- Accession Number
- AD0694438
Entities
People
- Frederick C. Durling
Organizations
- Southern Methodist University