International Journal of Educational Research 35 (2001) 365–372 Chapter 2 Models, modeling, and the design of the study Sarah Ransdell*, Candace Hawkins, Ryan Adams Florida Atlantic University, Division of Psychology, 2912 CollegeAvenue, Ft Lauderdale, FL 33314, USA Abstract In this chapter three theoretical models linking verbal ability, nonverbal ability, family background, and study behaviors with college GPA are presented. The sample, consisting of 142 US and 42 Estonian University students, is described. Finally, the procedure used to conduct the study is summarized and the methods used to analyze the data are reviewed. r 2001 Elsevier Science Ltd. All rights reserved. Structural equation modeling offers a unique framework that organizes relations between variable causes and effects. The framework includes two types of models, conceptual and statistical. Conceptual models serve as a theory of why a pattern of relations should take a certain form. Statistical models are intended to provide confirmatory evidence for the hypothesized patterns in those conceptual models. When the data fit the conceptual model, the underlying theory has been supported. When the data do not fit the conceptual model, there are two choices. One may revise the conceptual model and test it again, or one may compare one model against others on the basis of theoretical arguments and empirical evidence. The latter approach is taken in this monograph. As mentioned earlier, Model 1 suggests that family background is the most basic predictor of GPA. But Model 1 also specifies that study habits will moderate the effects of family background. In turn, verbal ability and nonverbal ability will moderate the effects of study habits. In other words, GPA will be a complex result of the combination of verbal ability and nonverbal ability with study habits moderating the impact of verbal and nonverbal ability and study habits in turn moderating the effects of family background variables. In contrast, Model 2 suggests that family background and study habits are of relatively equal importance in the overall model. Finally, Model 3 suggests that verbal ability and nonverbal ability mediate the *Corresponding author. Fax: +1-954-236-1099. E-mail address: ransdell@fau.edu (S. Ransdell). 0883-0355/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved. PII:S0883-0355(01)00033-7