Correcting for Acquiescent Response Bias in the Absence of a Balanced Scale

Abstract

This article describes a method for purging data of the effects of acquiescent response bias—a tendency to respond in the affirmative to survey items irrespective of substantive content—in the absence of a balanced scale (a scale with an equal number of positively worded and negatively worded items). The method is relevant in situations where the scale of interest is not balanced but where items similar in form from another content area are available to help identify acquiescence; it is illustrated here using data on class consciousness. Structural equation modeling techniques are used to extract a factor that captures the covariation among the items that is due to acquiescence. A series of models is presented illustrating the procedure on data from the Comparative Project on Class Structure and Class Consciousness for the United States. The model is then cross-validated on samples from Canada, Norway, and Sweden.