Growth Spline Modeling

Although researchers in many fields have long theorized about how the contributions of antecedents to outcomes might change across time, only recently have statistical tools necessary to examine such relationships become commonplace. In these situations the flexibility to simultaneously model ever smaller groups of observations both cross-sectionally (i.e., how the relative contribution of the antecedents differs among individuals) and longitudinally (i.e., how the relative contribution of the antecedents differs through time) is desirable (Singer, 1998). The SAS MIXED proceedure is a relatively new tool which provides this requisite flexibility. In this paper we will present an extensible, hybrid statistical approach comprised of spline modeling and growth modeling which allows for an examination of how the relative antecedent contributions to an outcome change through time while simultaneously controlling for past effects.