CIRCUMPLEX STRUCTURE OF DATA

Gaelick, 1981), but these discussions have typically been highly technical and have appeared in psychomet-ric-oriented publications unlikely to be read by most personality and social psychologists. In this article we address this gap between the meth-odological literature and the substantive personality and social psychology literature on circumplex repre-sentations of data. We discuss conceptual and methodo-logical issues in testing the circumplex structure of data. We begin by reviewing the conceptual and structural assumptions implied by proposing that a set of constructs can be represented as a circular ordering on the circumfer-ence of a circle. Next, we describe the methods most commonly used to test circumplex representations and we evaluate the extent to which these methods accurately assess circumplex structure. We then present a relatively nontechnical discussion of a recently developed covari-ance structure modeling approach to testing circumplex structure (Browne, 1992). We describe this approach, illustrate its use with published data sets, and contrast it with more common methods of testing circumplex struc-ture. We conclude with a discussion of the practical im-plications of this new approach for testing circumplex theories in personality and social psychology. Assumptions Implied by Circumplex Representations of Data Positing a circumplex representation of data implies certain conceptual assumptions about the nature of the constructs under investigation (e.g., see Larsen & Die-ner, 1992). Furthermore, as psychometricians have long recognized, these conceptual assumptions in turn imply a set of structural assumptions (i.e., assumptions about the underlying structure of associations among the vari-ables in question; e.g., Anderson, 1960; Browne, 1992