Some Problems in the Analysis of Multiresponse Models

摘要

Engineers and chemists frequently need to analyze multiresponse data. For example, when studying a chemical reactch reactor, for each setting of a group of 'input' variables determining the reaction conditions (such as temperature, pressure, reaction time and the initial concentrations of the reactants) a number of 'output' variables or responses (such as the final concentrations of each of the chemical substituents) may be measured. The capability of making such multiple measurements has greatly increased since the advent of better analytical tools such as the gas chromatagraph. But problems frequently arise in the analysis of such multiresponse data. One major source of problems arises from dependencies among the responses. Different kinds of dependencies can exist giving rise to different kinds of problems. The object of this paper is to consider certain aspects of the analysis of data of this kind and to present an illustrative example. (Author)