Estimation of Synergy in Risk Analysis Using Orthogonality Principles, Ancillary Variables and Multivariate Analysis

Abstract

This paper addresses an issue an issue in risk assessment which has often been acknowledged but for which reliable statistical analysis methods have not been synthesized: estimating synergy in the onset of risk event. Biological, environmental and habit-related factors all influence the detriment or detriments to health which we observe and research. Often, however, our data collection and data analysis strategy has had only the principles of “one-at-a-time” experiments — the possible risk factors have not been varied simultaneously in the planning of data collection. A “directed data collection and analysis” using multivariate statistical methods is advocated in this paper. The present author developed two systems, PRE-PRIM and POST-PRIM, which provide sampling and data collection specifications for research studies which explore synergistic effects. Design templates from PRE-PRIM use matrix orthogonality principles to assist in data collection for prospective or retrospective studies. POST-PRIM interfaces the data input matrix of hypothesized and ancillary variables with multiple output measures. A profile of multiple input-output links is thus generated.