Using Balanced Scales to Control Acquiescence

Abstract

The contamination of personality inventories and attitude scales by acquiescent response style is a continuing problem. Research with a recently-devised conservatism scale demonstrates empirically that the balanced keying of items is sufficient to prevent such contamination if item ambiguity is kept to a minimum. Such a scale also enables a statistical assessment of the degree to which unbalanced scales are biased by this response style. As is well known, psychological tests do not predict human behavior with anything the accuracy of, say, chemical tests in relation to the behavior of compounds. Part of the reason why many tests have low validity and/or reliability may be that they are not pure measures of the attributes under study; that is, the actual trait has been confounded with other measures, some of which are response styles. The purpose of this paper is to examine one kind of response style, known as acquiescence (although it includes both agreeing and disagreeing tendencies) and to report research on a recentlydevised scale of conservatism which may throw new light on a continuing problem. The term acquiescence was originally introduced to describe a tendency to agree rather than disagree with propositions in general. It has attracted attention because of its possible relevance to a wide range of tests, from proficiency tests to personality inventories to measures of vocational aptitude. Thus Cronbach has argued that in true-false tests some people have the habit of saying true when in doubt, while others are characteristically suspicious and respond false (Cronbach, 1960). Similarly, Lorge (1937) claimed that differences could be found in the way people responded with the like category to Strong Vocational Interest Blank items: one subject might use the