ABSTRACT
Just as polytomous logit analysis presents a problem of homonymy for the hearing impaired, polytomous (or multinomial) quantal response bioassays present a problem for most biologists who are not familiar with the statistical methods involved in data analyses. Few examples of analyses of polytomous data are available in the biological literature. Either the data remain unanalyzed or the information is collapsed into a binomial structure to facilitate use of regular binary probit or logit analysis. Use of the polytomous (multinomial) model makes it possible to quantify effects that are masked by collapsing all effects into a binary model in which dead or alive are the only categories; it also permits estimation of individual dose–response probit lines for each effect. Polytomous quantal response analyses are more informative, especially for data that concern the effects of juvenile hormone analogues (JHAs) or other chemicals that are not classic poisons.
