Two aspects of continuity, stability of individual differences and means, were examined in a longitudinal study from the middle elementary through the high school years. Two hypotheses regarding individualdifference stability were supported with structural equation modeling in both the general-verbal and math domains: (a) Academic intrinsic motivation is a stable construct throughout these years, and (b) with advancement in age, academic intrinsic motivation becomes increasingly stable. A third hypothesis, that the mean level of academic intrinsic motivation declines over these ages, also was supported, and significant linear trends were obtained, but it was also found to be modified by particular subject areas, with math showing the greatest decline and social studies showing no significant change. The combination of these 2 aspects of continuity places those with low motivation early in their schooling particularly at risk.

A hierarchical facet model of self-esteem proposed by Shavelson, Hubner, and Stanton was partially tested in a previous study by Fleming and Watts. Their three factors, which were labeled Self-Regard, Social Confidence, and School Abilities, corresponded to three of the four dimensions posited by Shavelson et al. Predictions of other individual-difference variables from these factors were also tested by correlational analysis. Improvements to the instrument-the Self-Rating Scale (which is an outgrowth of Janis and Field's Feelings of Inadequacy Scale)-led to a replication of their three factors plus two predicted physical factors: Physical Appearance and Physical Abilities. A second-order factor analysis yielded a single, superordinate factor of global self-esteem, supporting the hierarchical interpretation of the facet model. Construct validity was further examined by replication of the correlational findings of Fleming and Watts and by correlations with other measures of personality and adjustment, including a global measure of self-esteem: the Rosenberg scale. The facet model as presently operationalized measures dimensions of relevance for the intended population, but these dimensions are not so broadly defined as to be redundant with related constructs.

and the families participating in the Fullerton Longitudinal Study for help in various aspects of this study. Thanks are also extended to the reviewers for their valuable suggestions.

The hypothesis that self-esteem is'a multidimensional construct-as suggested by Shavelson, Hubner, and Stanton and others-was examined via factor analysis of a test adapted from Janis and Field. Three factors were found and labeled Social Confidence, School Abilities, and Self-Regard. These are comparable to three of the four dimensions discussed by Shavelson et al. Correlations of total selfesteem with measures of several other constructs were consistent with findings from previous studies; however, the correlations of the factors with these variables revealed that factors were differentially related to the constructs in some cases. These findings were seen as supporting a hierarchical conceptualization of self-esteem.The quantity, if not quality, of research on self-esteem has been impressive (e.g., see Wylie, 1974). As McGuire and Padawer-Singer (1976) have stated, "This continuing fascination with the self-concept is easy to understand: What we think about ourselves is probably the central concept in our conscious lives" (p. 743). Interest in the self-esteem construct in psychological research has emanated from several areas. These include studies of personality variables (e.g.

The templating effect of the tetrafluoroborate ion leads to assembly of four Co ions and six bridging ligands around this anion to give a tetrahedral complex with a bridging ligand along each edge and the anion trapped in the central cavity (shown below). Surprisingly under identical conditions but with Ni a simpler dinuclear complex forms.