Self-concept: The generalizability of research on the SDQ, Marsh/Shavelson model and I/E frame of reference model to united Arab emirates students

Social Behavior and Personality, 2000 by Abu-Hilal, Maher M, Bahri, Tawfik M

This study examined the structure of self concept (SC) and the internal/external (LIE) frame of reference model for elementary (n = 276) and junior high (n = 293) students in the United Arab Emirates. The results provided support for the multifaceted interpretation for both groups of students. However, the study provided stronger support for the hierarchical structure with junior high than with elementary students' data. Higher-order confirmatory factor analysis (HCFA) of each group revealed that the second-order factor model was the best fitting model. The pattern of correlations indicated that SC in various areas becomes more differentiated with age. Self concept dimensions were more distinct for junior high students than for elementary students. SDQ-I subscales were more reliable with junior high students (range: .72-.94) than with elementary students (range: .74-.90). The study provided partial support to the propositions of the I/E frame of reference model for both groups of students. MANOVA revealed that elementary students demonstrated significantly higher mean scores than did junior high students on two of the seven SC subscales (math and school SCs).

The self concept (SC) structure and its relations to other constructs have attracted much attention in the past two decades. This attention centered on the multidimensional, hierarchical structure and construct validity of SC. Numerous within-network and between-network studies in the West have established the multidimensionality, hierarchy, and construct validity of SC. This paper examined the generalizability of results derived from: (a) the Self Description Questionnaire (SDQ), Shavelson, Hubner & Stanton (1976) theory of SC, and (b) Marsh's revisions to the theory and the internal/external (IJE) frame of reference model. This study is important in that it provides an opportunity to examine the situated generalizability of results in the context of different linguistic and cultural boundaries.

Marsh ( 1993) stated that "self concept researchers often evaluate age or gender effects in different components of self concept" (p.843) in an implicit indication that SC is a developmental construct. Shavelson et al. (1976) and Marsh (1985) posited that children's self concepts become more differentiated with age. Marsh (1988) explained that as children grow "they incorporate more external information into their self concepts" and " as children incorporate more information about their actual skills and abilities as well as feedback from others, into the formation of their self-concepts in different areas, their self concepts will also become dif ferentiated" (p.124). Accordingly, maturity would lead to smaller correlations among various dimensions of SC, and to larger correlations between SC and external constructs.

Shavelson et al. (1976) demonstrated that self concept becomes multifaceted with age. Marsh, Barnes, Cairns and Tidman (1984) found that younger children scored significantly higher than older children did on most of the SDQ scales. Marsh et al. ( 1984) compared the responses of preadolescent children with those of adolescent children and concluded that the effect of age was linear for preadolescent years and non-linear during adolescent years. However, Marsh (1990) reported that studies that employed overall scores for SC had found no evidence for any age effect. The evidence for age effect on SC dimensions, or even overall SC, is scarce in the published work of Arab researchers. Since age has been one of the main features of the Shavelson et al.'s theory of SC, it is important that the relationship of SC with age, as indicated by grade level, be considered with nonwestern samples.

Shavelson et al. (1976) proposed a model of SC which has been considered by many as a well-defined model, and which has been supported by subsequent research (e.g., Byrne & Shavelson, 1986; Marsh, 1988; 1990). Some important modifications, however, have been added to the model (Marsh & Shavelson,1985; Shavelson & Marsh, 1986). These revisions pertain to academic SC and the revised form of the model has been designated the Marsh /Shavelson Model of Academic SC. Marsh and Shavelson (1985) and Shavelson and Marsh (1986) posited that academic SC is not a uni-dimensional construct. Rather, math and reading SCs have been proposed and found to form two weakly-related facets. This position was supported by the majority of studies conducted in the West (see Marsh 1986, 1988, and 1990, Marsh & Shavelson, 1985); however, it was not supported by non-western samples (Abu-Hilal & Aal-Hussain, 1997).

The frequently used construct that has been linked to SC is academic achievement. Lyon ( 1993) highlighted the importance of the relationship between SC and academic achievement, and stated that "academic self concept is theoretically more closely related to academic achievement than most other cognitive variables" (p. 203). Marsh, Parker and Barnes (1985) indicated that the substantial correlations between different areas of SC and academic ability and performance measures provide evidence for the construct validity of SC. Marsh (1986) presented the relationship between achievement and SC in a novel and unique fashion with his phenomenal internal/external (I/E) frame of reference model. The I/E model has been considered as "a major development in self perception theory" (e.g., Skaalvik & Rankin, 1995) which provides an explanation for the surprising relationship among SC dimensions. Also, it clarifies the cross-links between math and verbal achievements, and math and verbal SCs. Based on the correlations between math and verbal achievements (r's=.50 to .80), Marsh expected that math and verbal self concepts would be similarly related. Surprisingly, Marsh and his associates have found that math and verbal SCs are nearly uncorrelated-leading to the revision of the original Shavelson et al. (1976) model and to the proposal of the IHE model (see Marsh et al. 1985; Marsh, 1988, 1990; Marsh, Byrne & Shavelson, 1988).