Bringing the Arrows Back In: On the Recursivity Assumptions in School Process Models

School process models in the tradition of those introduced by Sewell and his colleagues have identified interpersonal influences as central to the maintenance of differences in goal orientations across SES levels. These differences in goal orientations, in turn, are the most immediate causes of the attainment differentials observed later as youth complete their schooling and enter the world of work. Documentation of this social-psychological process typically has come from recursive structural equations models in which this particular step-wise flow of influence is entertained and other plausible possibilities are precluded. The present study examines the tenability of these recursive constraints by estimating two alternative specifications: (1) a nonrecursive model in which cross-sectional measures of interpersonal influence, educational plans, curriculum placement, and school performance all are simultaneously determined, and (2) a longitudinal recursive model, in which earlier school process measures ar e allowed to affect each later school process variable. These estimations indicate that significant others offer support in response to students' expressed plans for college but are themselves ineffectual as agents of socialization. This pattern is in striking contrast to that obtained under the recursive specification, and suggests that the evidence in favor of the social-psychological interpretation may be fundamentally flawed.