Big Fish and Other School Effects on Academic Self-Concept

Abstract

A substantial amount of research indicates that academic self-concept is a function of both individual characteristics, and school effects that impact on the development of selfperceptions. Few studies have studied a cohort of students as they progress through the transition from elementary to middle school. The present study uses multi-level modeling to examine school effects on students’ academic self-concept in reading and math as they transition from elementary to middle school. Data come from the ECLS-K data set. Few school effects were found, but students’ SES was found to be a strong moderator of the relationship between reading achievement and self-perceptions of students’ ability and interest in reading. Educational researchers increasingly have become attuned to the possibility of “school effects” on educational outcomes. “School effects” refers to the idea that various outcomes may be, in part, due to school-wide characteristics, over and above individual student or teacher characteristics. For example, Lee and Smith (1997) found that mid-size high schools produce larger achievement gains from students’ freshman to senior years than do either small or large high schools, net of individual student characteristics, and that achievement in these schools was more equitably distributed across students’ SES. Similarly, Lee and Smith (1996) found that high schools in which teachers, as a group, believed in collective responsibility for student achievement produced stronger achievement gains than did schools whose teachers, as a group, held different attitudes. “School effects” are just one example of an “ecological effect”, in which features of the larger environment have demonstrable effects on various outcomes net of individual variables. The school psychology literature contains little research that truly explores ecological effects, although some recent studies (e.g., Rhodes, Roffman, Reddy, & Fredriksen, 2004) have included such variables. Testing for ecological effects also raises methodological issues. Most such research has used ordinary least-squares regression (OLS) to assess school effects net of individual variables through the straightforward practice of using ordered regression procedures in which individual student characteristics are entered as a block on the first step, and school-wide variables are entered on the second step. A significant increase in R 2 then serves as the measure of school-wide variables while controlling for individual variables. Alternatively, one might use an ANOVA and use covariates to partial out various effects. While these procedures have intuitive appeal and are reasonably easily understood, Lee (2000) delineates the methodological shortcomings of this approach: (a) aggregation bias – in which the same variable (e.g., SES) may have different meanings at different levels of aggregation (e.g., individual vs. school levels), (b) non-independence of cases (e.g., an individual student’s achievement may be related to the achievement levels of other students in the school), and (c) heterogeneity of regression. To counter these shortcomings, researchers may use multi-level modeling (MLM) methodology that allows for the modeling all of these effects. The most commonly used statistical program for performing such analyses is Hierarchical Linear Modeling (HLM) (Raudenbush, Byrk, Cheong, Congdon, & du Toit, 2004). Studies using HLM have begun to appear in the school psychology literature (Rhodes, et al., 2004; Stage, 2001). Most recently Clements, Bolt, Hoyt, and Kratochwill (2007) have championed the use of School Effects on Academic Self-Concept 3 MLM methodology to study school-based interventions. Research on school effects has focused primarily on achievement as outcomes, but such logic and analysis can be applied to social-emotional outcomes as well. Over a 20-year period, Herbert Marsh and his colleagues (e.g., Marsh & Hau, 2003) have conducted a substantial amount of research investigating the effects of school average ability on student’s academic self-concepts (ASC). Marsh and his co-workers discovered that school characteristics moderated the relationship between ASC and achievement. Humorously dubbed the “big-fish-little-pond effect (BFLPE)”, Marsh found that students with the same academic achievement level had somewhat lower ASCs when embedded in schools with higher average student ability, although the effect size is relatively small. Other recent research has further explored school effects on ASC. Trautwein, Ludtke, Koller, and Baumert (2006) found that the learning environment moderates the development of self-concept for a set of seventh grade students. Specifically, “meritocratic” schools produced more accurate ASCs than did “ego-protective” school environments. School average ability and meritocratic structure are excellent examples of “school effects”. None of these studies, above, investigated the effects of school transition on academic self-concept. Numerous other studies have looked at such effects in relation to goals, classroom practices, teaching styles and foci, and social status. In general, research has found that students’ academic self-concept declines during early adolescence and junior high school (Anderman & Midgley, 1997; Cantin & Boivin, 2004; Cole, 2001; Harter, Whitesell and Kowalski, 1992) which may be related to the change in atmosphere, change in teacher expectations, new grading practices, and a new social atmosphere. Harter, et al. investigated fifth through seventh graders experiencing school transitions and found that the effect of their academic self-concept was related to the transition as well as the anxiety and affect related to the transition. Wigfield, Eccles, Mac Iver, Reuman, and Midgley (1991) found that students’ academic selfconcept declined between sixth and seventh grades while their perceptions of social ability increased during seventh grade. Little research has focused on the possibility of school effects other than the BFLPE on academic selfconcept. Such school effects include the type of school (public or private), the size of the school, and the average socio-economic status of school. The goal of the present study was to extend the research on school effects on academic self-concept by tapping data from the recently released 8 th grade follow-up of students in the Early Childhood Longitudinal Study, Kindergarten cohort (ECLS-K). The 8 th grade follow-up allows for comparison of the influences on academic self-concept in elementary and middle school. Hierarchical linear modeling (HLM) methodology was used to address the following questions: 1. What individual and school effects are associated with academic self-concepts in 5th and 8th grade? 2. Are school effects on 8 graders’ academic self-concepts similar to those for 5 graders, after accounting for individual child characteristics, particularly academic ability in reading and math? 3. Do students who transition from elementary to middle school with substantially higher (or lower) school-wide achievement experience changes in their self-perceptions of competence?