The Self-Description Questionnaire (SDQ) is a self-concept instrument designed to measure seven components of self-concept derived from Shavelson's model. Separate factor analyses of responses by children (6 to 11 years of age) from four grade levels (Grades 2 to 5) all clearly identified the seven SDQ factors, and correlations among these factors were consistent with the hierarchical organization of self-concept hypothesized by Shavelson. However, several observations suggested that these factors become more distinct with age. Large sex differences for two factors (boys' self-concepts were higher in Physical Abilities; girls' were higher in Reading) and small sex differences for several other factors were consistent across different ages and consistent with previous findings with older children. There was a strikingly linear, negative relation between grade level and the total self-concept scores, the three academic self-concepts, and the two physical self-concepts, but not Relationship with Parents and Relationship with Peers. A separate examination of responses to negatively worded items demonstrated a substantial method/halo bias that was systematically related to grade level, arguing against their use on questionnaires for young children. These findings have important implications for the study of preadolescent self-concept and also support the construct validity of interpretations based on the SDQ and the Shavelson model.Self-concept is a hypothetical construct whose usefulness must be demonstrated by investigations of its construct validity. Within-network studies explore the multidimensionality of self-concept and attempt to demonstrate consistent, distinct, and theoretically defensible components. They typically employ factor analysis. Between-networks studies attempt to demonstrate a theoretically consistent (or at least logical) pattern of relationships between measures of self-concept and other constructs. These studies correlate self-concept indices with variables such as measures of ability/performance, self-concept ratings inferred by significant others (e.g., teachers, parents, peers), family background variables, behavioral observations, experimental manipulations, and other self-report measures.The authors would like to acknowledge the insightful criticism and helpful comments provided by Ray Debus on earlier drafts of the manuscript.

The study of dispositional differences in self-attributions has important implications for educational settings, but critical issues have been largely ignored in much attributional research. The purposes of this study are to clarify the distinction between dispositional and situational approaches to attribution research, to review particular issues that are important for the study of individual differences in self-attributions, to examine these issues with respect to results from a new self-attribution measure, and to demonstrate how self-attributions are related to dimensions of self-concept. Conclusions based on the literature review and empirical findings both demonstrate that (a) individual differences in self-attributions cannot be explained in terms of the bipolar dimensions that have been found in research that manipulates situational components of the attribution process (e.g., the internal-external, stable-unstable, and controllable-uncontrollable dimensions), (b) attributions for success and failure outcomes differ in ways that have not been recognized by attributional theorists, and (c) ability attributions (but perhaps not attributions to effort and external causes) are specific to particular areas of academic content. The results also demonstrate a clear and predictable pattern of relationships between dimensions of self-attribution and self-concept that supports the convergent and discriminant validity of responses in each of these areas.

College students were asked to complete the Self-Description Questionnaire III (SDQ III;Marsh & O'Niell, 1984) and to ask the person who knew them the best to complete the SDQ III as if they were the person who had given the particular student the survey. The purposes of the study are to examine psychometric properties of the SDQ III, to determine the ability of an external observer to accurately infer multidimensional self-concepts, and to describe new methodological approaches to this type of study. Separate factor analyses of both self-ratings and responses by significant others identified the 13 dimensions of self-concept that the SDQ III was designed to measure. For each set, the internal consistences of the 13 scales were high, whereas the average correlation among the factors was close to zero. Selfother agreement was quite high (mean r = 0.58), demonstrating that significant others are able to accurately infer multidimensional self-concepts of someone who they know well. Multitrait-multimethod (MTMM) analyses, based on Campbell-Fiske criteria and confirmatory factor analysis, supported both the convergent and divergent validity of the SDQ III factors. A methodological extension of the MTMM design is introduced and its implications for other research are discussed.

The Self Description nuestionnaire II (SDQII) wawadministered to 901 students s11 to 18 years old) in grades 7 through 12 hho attended one public coeducational high school. Factor analysis clearly identified the 111SDQIIcales, each scale eas reliable (median alpha = .86)) and correlations smong the factors were small (median r = .17). All of the SDQ II scales wese significantly correlated with sex and/or age, though ghe effects of sex and age were smalllnd independent of each other. The direction of the sex effect taried dith tht particular scale, and was not significant for the sum of all the eSQ II scales. This total score, and most of tho separate scales, had a quadratic age effect where self-concepts started out high, reached their lowest level in gradg 9, and ,hen increased. At tvery grade devel academic criterion measurea were significantly correlated with every academic scale, but not with thw nonacademic scales. Verbal achievement was most highly correlated wiih Verbal self concept, while mathematics achievement was most highly yorrelated with Math self concept. These findings not only demonstrate the multidimensionality of self-concept, but also show that iti relationship to other constructs cannot be adequately understood if fhis multidimensionality is ignored. The findings have important implications for the study of adolescent self-concept and support tth construct validity of the SDQ II and the Shavelson model on which it is based.Self-concept is a hypothetical construct whose usefulness must be demonstrated by investigations of its construct validity. Marx and Winne (1978) argue that the demonstration of consistent, distinct, and theoretically Q2.2. at NORTH DAKOTA STATE UNIV LIB on June 29, 2015 http://aerj.aera.net Downloaded from SELF-CONCEPT defensible components of self-concept is prerequisite to the study of how self-concept is related to other constructs. Systematic reviews of selfconcept research (e.g., Burns, 1979; Shavelson, Hubner, & Stanton, 1976; Welles & Marwell, 1976; Wylie, 1974, 1979) emphasize the lack of a theoretical basis and the poor quality of measurement instruments used in most studies. Shavelson et al. (1976) reviewed theoretical and empirical research in this field and used the review as the basis of a self-concept model that incorporates aspects from most theoretical positions. Of particular relevance are Shavelson's assumptions that self-concept is multifaceted, hierarchically arranged, and becomes increasingly multifaceted with age. This theoretical model was the basis for the design of the three Self Description Questionnaire (SDQ) instruments. THE MULTIDIMENSIONALITY OF SELF-CONCEPT The self-concept dimensions proposed by Shavelson et al. (1976), as well as their hypothesised structure, are heuristic and plausible, but they were not empirically demonstrated. Despite the assumption of multidimensionality of self-concept, which is explicit in Shavelson's model and implicit in other research, factor analyses of the most commonly used instruments typically fail to identif...

The authors would like to acknowledge Samuel Ball, Raymond Debus, and Lee Owens for their helpful suggestions at various stages of this research.