| Herbert W. Marsh
University of Western Sydney
Multiple dimensions of self-concept in NELS:88: Their dual role as outcome and process variables
FINAL REPORT: Study 1: Using the NELS:88 to Evaluate Theoretical Models of Self-Concept: The Self-Description Questionnaire. Self-concept scales (Math, English, Parent Relations, Same-Sex Peer Relations, and Opposite-Sex Peer Relations) from the Australian Self-Description Questionnaire II (SDQII) were included in the National Educational Longitudinal Survey of 1988 (NELS:88). Mean differences that were based on responses by 17,544 (NELS:88) U.S. and 1,147 (SDQII normative archive) Australian students were small, and gender differences (girls higher for English and Same-Sex Relations but lower for Math) were similar for both countries. Structural equation models relating mathematics and English achievement scores, school grades, self-concepts, and school-average abilities replicated and extended previous results that were based on the internal/external frame of reference model and the big-fish-little-pond effect. The results support the construct validity of the SDQII responses in the NELS:88 data, and have implications for self-concept theory, measurement, and practice. Study 2: Evaluation of Self-Rating Scales in Specific School Subjects from the National Educational Longitudinal Study of 1988: An Application of Confirmatory Factor Analysis. The purposes of this study are to examine the specificity of academic attitudes to particular school subjects, to demonstrate and extend the use of confirmatory factor analysis (CFA) models of multitrait-multimethod (MTMM) data to this issue, and to develop appropriate guidelines for using items from the National Educational Longitudinal Survey of 1988 (NELS88) for the lOOOs of users of this important data base. As part of the NELS88 study, a large, nationally representative sample of eighth-grade students (N=24,599) rated their intrinsic motivation, extrinsic motivation, and anxiety in mathematics, science, social studies, and English. The CFA models showed that the school subject factors were contaminated with method effects so that simple scale scores are not appropriate. Evaluation of more complex models derived from the MTMM literature indicated that students clearly differentiated attitudes in relation to different school subjects, and that there was good support for the convergent and discriminant validity of their responses in relation to standardized test scores and schoal grades in these four school subjects. Study 3: Positive and Negative Global Self-Esteem: A Substantively Meaningful Distinction or Artifactors. Global Self-Esteem (GSE) items derived the Rosenberg scale are used widely as a relatively unidimensional scale. However, factor analyses suggest separate factors associated with positive and negative items and there is an ongoing debate as to whether this distinction is substantively meaningful. Consistent with this confusion, users of the National Longitudinal Study of 1988 were first told to treat GSE items as a single scale and then as two separate scales in a subsequent technical report. Confirmatory factor analysis (CFA) was used to evaluate alternative one- and two-factor models and to test hypotheses about how the factors vary according to reading ability. Responses reflected a relatively unidimensional factor and idiosyncratic method effects associated with negatively worded items, and these resuits were replicated in further analyses of followup data collected two years later. Similar negative-item effects are likely to contaminate ratings in other personality and social psychological research, and this CFA approach should be useful in evaluating whether factors associated with positive and negative items are substantively meaningful or artifactors .
Back to Funded Research Grants Page |