The Role of ‘Intervening Variables’ in Clinical Research

Abstract

The study of the relationships between behavioural variables and drugs has two distinct purposes. The firts purpose is to specify drug indications and efficacy. The second purpose is to better characterize the relationship between cerebral structures and behaviour on the basis of the known mode of action of drugs. To serve either purpose, we use several behavioural variables (several items on a scale or several scales). The action of a drug is thus usually studied on one or two dozen variables. The recourse to a variety of variables is justified on several grounds. Firstly due to our very unprecise means of observation and criteria for description and quantification of behavioural data, an evaluation based on only one item would be very unreliable. By using a relatively high number of traits, the risk is minimized: thus an inadequate rating of one item is compensated by a better rating of the others. The second reason lies in the fact that actions are subject to multiple determinations. Each behavioural trait is the result of several factors and drugs do not necessarly act on all of them. In other words, one behavioural trait cannot be considered as a sufficient index of drug action and as representing the behavioural target of the drug.