Abstract. This study examined the hypothesized relationship between social adjustment, as measured by perceived social support, self-concept, and social skills, and performance on academic achievement tests. Participants included 27 teachers and 77 fourth- and eighth-grade students with diverse academic and behavior competencies. Teachers were asked to select one student for each of the three participant nomination categories: undeveloped academic competence, undeveloped behavior competence, and proficient academic and behavior competence. Multivariate analysis of variance results indicated that each participant group differed significantly on social skills, and students with proficient academic and behavior competence demonstrated significantly greater levels of self-concept than did those with an undeveloped behavior competence. None of the groups differed significantly on perceived social support. Structural equation modeling analyses revealed that the model predicting academic achievement from self-concept, social skills, and academic competence adequately fit the data. Indicators of social adjustment were discussed as intervention targets for programs intended to improve students' social competence and academic achievement.
Students with learning and behavior difficulties tend to present relatively lower levels of social adjustment than their typically achieving peers. Key indicators of social adjustment include perceived social support, self-concept, and social skills. Researchers have demonstrated that levels of these three indicators vary by students' academic and behavior competencies (Demaray & Malecki, 2002; DiPerna, Volpe, & Elliott, 2002; Merrell, 1991; Wentzel, 1991). These studies have included students in 3rd-12th grades, individuals with and without disabilities, and roughly one third of participants in each study were racial or ethnic minorities.
Several researchers have also shown that academic achievement can be predicted from indicators of social adjustment. In their longitudinal study of the prosocial foundations of children's academic achievement, Caprara, Barbaranelli, Pastorelli, Bandura, and Zimbardo (2000) used children's third-grade social behavior to predict their eighth-grade academic achievement. Caprara et al. (2000) found that early prosocial behavior robustly predicted later academic achievement, but early aggressive behaviors did not. Third-grade prosocial behaviors significantly predicted eighth-grade academic achievement, even when controlling for third-grade academic achievement. Interestingly, third-grade academic achievement was not a significant predictor of eighth-grade academic achievement when controlling for third-grade prosocial behaviors. The results of this study suggest that social skills significantly contribute to later academic achievement, more so than do problem behaviors and even early academic skills.
Other researchers have included the construct of academic competence in their predictive models of academic achievement. DiPerna and Elliott (2002) defined academic competence as a multidimensional construct comprised of students' skills, attitudes, and behaviors that contribute to school success. The components of academic competence are categorized as either academic skills or academic enablers. Academic skills include a student's aptitude in content areas such as reading and math, whereas academic enablers are the attitudes and behaviors (i.e., motivation, interpersonal skills, etc.) that facilitate a student's learning (DiPerna & Elliott, 2002). Academic competence has been operationalized as teacher ratings of a student's performance across multiple domains including content area achievement (e.g., reading, math), motivation, classroom behavior, and parental encouragement (DiPerna et al., 2002; Gresham & Elliott, 1990).
Although researchers have demonstrated an association among social adjustment and academic and behavior competencies, the methods for identifying students of varying levels of academic and behavior competence have relied on either disability status or the same instruments used to examine the association with social adjustment. This study employed a distinct teacher nomination procedure to select students with diverse academic and behavior competencies and included several indices of students' social adjustment in the prediction of academic achievement.
Because prior research findings suggest that indicators of social adjustment are associated with academic performance (Caprara et al., 2000; Demaray & Malecki, 2002; DiPerna et al., 2002; Wentzel, 1991) and that interventions targeting students' social adjustment have led to improvements in academic achievement (Zins, Bloodworth, Weissberg, & Walberg, 2004), it is important to examine the pathways through which social adjustment influences academic achievement. A reasonable hypothesis is that academic competence, the combination of a student's academic skills and enablers, is one such pathway, as it has been demonstrated to be associated with indicators of social adjustment (Demaray & Malecki, 2002; Gresham & Elliott, 1990) and academic achievement (DiPerna et al., 2002). For this reason, the predictive model of academic achievement tested in the present study was based on the hypothesis that social adjustment would have an indirect influence on academic achievement through its direct influence on academic competence.
Two research questions were addressed in this study: (a) Are there significant group differences in social adjustment, as measured by perceived social support, self-concept, and social skills, based on students' teacher-identified academic and behavior competencies (as indicated by their placement in one of three participant nomination groups: undeveloped academic competence, undeveloped behavior competence, and proficient academic and behavior competence)? (b) Is there empirical support for a predictive model of academic achievement based on social adjustment (perceived social support, self-concept, and social skills) and academic competence?
Participants included 77 students, in Grades 4 and 8, and 27 teachers from 11 school districts in Wisconsin. The overall student sample was divided in a relatively even manner by sex and grade. Racial and ethnic minority students represented approximately 17% of the sample (see Table 1 for demographic characteristics by participant nomination group).
Teachers were instructed to nominate one student in their entire classroom that best fit each of the three participant descriptions: undeveloped academic competence (UAC), undeveloped behavior competence (UBC), and proficient academic and behavior competence (PABC). A student with a UAC was defined as not performing at grade level in reading-language arts and/or math (i.e., the student performs at a level that is noticeably lower than his or her peers in one or both of these subjects and the student's performance in reading-language arts and/or math does not reflect an expected level of understanding of the given academic subject). A student with a UBC was defined as not acting in a manner that is developmentally appropriate or consistent with his or her peers' behavior (i.e., the student exhibits behaviors that interfere with his or her learning or ability to attend to instruction). However, this student's academic performance is not below grade level in math and reading--language arts. Finally, a student with PABC was characterized as acting in a developmentally appropriate manner consistent with, or better than, his or her peers' behavior. This student also performs at or above grade level in reading-language arts and math. Each of the three participant categories contained three additional criteria: The student may or may not have been classified as having a disability, the student may or may not have been receiving special education services, and the student did not meet the criteria for the other two participant nomination categories.
Two of the teachers were each able to nominate only one student who subsequently participated. Fourth-grade students were nominated by their general classroom teachers and eighth-grade students were nominated by either their math or language arts teachers. This teacher nomination procedure was used for three primary reasons. First, it allowed for a sample that included students with diverse academic and behavior-related skills. Second, the use of distinct groups facilitated the examination of whether students' social adjustment varies according to their general academic and behavior competencies. Finally, it allowed teachers to select students based on local norms, rather than state disability categories or nationally norm-referenced instruments.
The first author contacted administrators at public schools throughout southern Wisconsin to obtain permission to recruit teacher and student participants. Interested teachers were sent recruitment materials and consent forms to be given to their nominated students. Teachers were required to provide written consent for their own participation. In addition to obtaining informed consent from the students' parents, fourth-grade students provided verbal assent and eighth-grade students provided written assent before they were administered the assessments. After receiving written consent from the parents of nominated students, the first author administered the Child and Adolescent Social Support Scale and Student Self-Concept Scale to the students in the school setting. The teachers were given the Social Skills Rating System--Teacher Form and a brief demographic questionnaire to complete for each student. Wisconsin Knowledge and Concepts Examination scores were obtained from classroom teachers and, for one district, from the district office.
Child and Adolescent Social Support Scale (CASSS). The Child and Adolescent Social Support Scale (Malecki, Demaray, & Elliott, 2001) measures 3rd-12th-grade students' levels of perceived support from parents, teachers, classmates, close friends, and the school. This instrument also assesses four types of social support: emotional, instrumental, informational, and appraisal. The five sources of perceived social support are measured through 12-item subscales for a total of 60 items, which are summed for a total support score.
Student Self-Concept Scale (SSCS). The Student Self-Concept Scale (Gresham, Elliott, & Evans-Fernandez, 1993) is a 72-item assessment of children and adolescents' self-perceptions. It is appropriate for use with students in the 3rd-12th grades and is available in two age-appropriate levels: Level 1 for 3rd-6th grades and Level 2 for 7th-12th grades. The composite self-confidence standard score was used to represent students' self-concept.
Social Skills Rating System--Teacher Form (SSRS-T). The Social Skills Rating System--Teacher Form (Gresham & Elliott, 1990) is a norm-referenced behavior rating scale that allows teachers to assess their students in three domains: social skills, problem behaviors, and academic competence. It is available in two developmentally appropriate forms: elementary (K-6) and secondary (7-12). The total social skills standard score was used to represent students' social skills and the academic competence standard score was used as a measure of the students' academic competence.
Standard scores were used for the self-concept and social skills measures described in the preceding paragraphs to control for age-grade and sex, both of which have been shown to be associated with levels of self-concept, social skills, and perceived social support (Gresham & Elliott, 1990; Gresham et al., 1993; Malecki et al., 2001). Although the measure used for perceived social support, the Child and Adolescent Social Support Scale, does not offer published standard score conversions based on a nationally representative standardization study, mean scores and standard deviations are available for each subscale, classified by gender and grade cluster (3-6 and 7-12), from a sample of 1,100 students in five states (Malecki & Demaray, 2002). These data were used to compute social support z scores for the present study. Because of the known differences in perceived social support, self-concept, and social skills by age and sex, composite scores were used to allow for the calculation of standard scores and z scores.
Wisconsin Knowledge and Concepts Examinations (WKCE). The WKCE is a series of standardized achievement tests administered each year to Wisconsin public school students in Grades 4, 8, and 10 (Wisconsin Department of Public Instruction, 2005). The WKCE incorporates multiple-choice and short-answer items along with writing samples to measure students' achievement in five content areas: reading, language arts, math, science, and social studies. Students' performance on the reading and math sections were used as indicators of academic achievement.
It should be noted that the WKCE assessments were administered several months before the data were collected for this project. However, the participating teachers were not aware of their students' performance on the WKCE when they completed the student ratings for this study, as districts had not received the WKCE results.
To examine whether there were significant group differences in social adjustment based on students' teacher-identified academic and behavior competencies, a multivariate analysis of variance was conducted with nomination status (UAC, UBC, and PABC) as the independent variable and perceived social support, self-concept, and social skills as the dependent variables. Wilk's lambda was used to estimate the multivariate F statistic and the a priori alpha level was established at .05. Because the omnibus F statistic was significant, multiple comparisons of the group means were conducted using the Fisher least significant difference procedure. The Fisher least significant difference method was selected because it has been shown to be the most appropriate post-F-test procedure for comparing three groups (Seaman, Levin, & Serlin, 1991). It was predicted that each indicator of social adjustment would be the greatest for the students in the PABC group, followed by those in the UAC group. Students in the UBC group were expected to have the lowest levels of social adjustment.
[FIGURE 1 OMITTED]
The result of the multivariate analysis of variance testing the multivariate effect of participant nomination status on the social adjustment measures was significant (Wilk's lambda = 0.195; df = 3, 65; F = 89.51) at the .05 level. Follow-up multiple comparisons were conducted using the Fisher least significant difference procedure to determine which group means differed on the social adjustment variables. For social skills, each group's mean total score was significantly different from the other groups in the predicted order: PABC > UAC > UBC. For self-concept, students in the PABC group only had significantly higher total scores than students in the UBC group. None of the groups differed statistically on their mean total perceived social support scores.
To determine whether there was empirical support for the conceptual model in Figure 1, SEM was used. Raw data, rather than a covariance matrix, were analyzed using AMOS 5.0, and the model parameters were generated using maximum likelihood estimation. The correlations between each of the variables included in the model appear in Table 2. The conceptual model of social adjustment and academic achievement, represented in Figure 1, includes two latent constructs (social adjustment and academic achievement) and six observed or indicator variables (self-con = self-confidence composite standard score, support = total social support z score, s-skills = social skills standard score, acadcom = academic competence standard score, math = WKCE math z score, and reading = WKCE reading z score). Also included in the model are error terms for each of the observed variables and an error (disturbance) term for the endogenous variable, academic achievement. The hypothesized model was predicted to demonstrate adequate fit with the data.
Sample size is an important consideration when using SEM. The necessary size for reliable results depends on the complexity of the model, the magnitude of the coefficients, and the multivariate normality of the variable distributions. Original studies (e.g., Loehlin, 1992; Lomax, 1983) indicated that input matrices should be based on at least 100 cases. Bentler and Chou (1987), however, provide a general rule of 5-10 cases per parameter estimate. Given that the models tested in the present study consisted of six or seven parameters, the sample size meets this adequacy criterion.
The hypothesized social adjustment and academic achievement model did not demonstrate adequate fit with the sample data, as indicated by the original measures of fit ([chi square] = 25.08, p = .00, [chi square]/df = 3.14, normed fit index = 0.85, comparative fit index = 0.88, nonnormed fit index = 0.78, goodness of fit index = 0.90, and root mean square error of approximation = 0.18), although the predictors in this model accounted for approximately 71.8% of the variance in academic achievement. Figure 2 presents the standardized path coefficients represented in the tested model.
Because the original hypothesized model did not adequately fit the sample data, an exploratory analysis was conducted in which the model was revised and the fit of the new model was assessed. The revised model, which is presented in Figure 3, was adjusted based on the values in the standardized residual covariance matrix of the initial model. Examining these values allows identification of pathways that could be added or deleted to improve the fit of the model (Kline, 1998). Given the absolute values of the standardized residual covariances between support and math and between support and self_con were relatively large, perceived social support was removed from the model. This revision was defensible from a theoretical perspective, as the association between perceived social support and academic performance examined in past research has not been as consistent as the association between academic performance and social skills or self-concept (Demaray & Malecki, 2002; Wenz-Gross & Siperstein, 1998). The revised model demonstrated acceptable fit with the sample data ([chi square] = 1.19, p = .88, [chi square]/df = 0.30, normed fit index = 0.99, comparative fit index = 1.00, nonnormed fit index = 1.06, goodness of fit index = 0.99, and root mean square error of approximation = 0.00). The standardized path coefficients are presented in Figure 3.
[FIGURE 2 OMITTED]
A number of researchers have shown that students' social functioning can influence their academic achievement. This study examined this relationship further by (a) exploring correlational differences among indicators of social adjustment for students with diverse academic and behavior competencies, and (b) testing a model for predicting academic achievement from social adjustment and academic competence. Teacher ratings of students' social skills differed significantly among each of the three participant nomination categories: PABC, UAC, and UBC. Self-concept was shown to be related to classroom behavior, but not directly to academic performance. Perceived social support was not meaningfully associated with classroom behavior or achievement test scores. A revised model of academic achievement, which did not include perceived social support, adequately fit the data. This revised model suggests that students' social skills and self-concept affect academic achievement through their effects on academic competence. However, this model needs to be tested on other samples before any firm conclusions can be drawn.
Implications of Findings
The primary implication of the findings is the direction they provide for future research. The poor fit of the original hypothesized model and the absence of different levels of perceived social support by participant nomination category suggest that perceptions of overall support from various sources may not be good indicators of a student's social and academic competence as they relate to classroom behavior and academic achievement test results. It may be that perceived support from teachers or classmates is more relevant to school-related competencies such as attendance or attitudes about school.
The results of the current study indicate that global self-concept may not be directly related to academic achievement, as the students with relatively undeveloped academic competence did not differ significantly in their levels of self-concept from students with relatively proficient academic and behavior competence as judged by their teachers. It may be the case that self-perceptions only influence academic performance indirectly through their effects on classroom behavior. Perhaps academic-related self-perceptions would have more of a direct effect on achievement than general self-concept.
The findings of this study also have practical implications for school-based intervention. Teachers, school psychologists, and administrators seeking to implement prevention and intervention programs to promote social and academic competence should consider social skills and self-concept as potential intervention targets. Researchers have demonstrated the effectiveness of intervention programs for improving students' social and academic competence that emphasize self-awareness, self-management, and relationship skills (Zins et al., 2004).
[FIGURE 3 OMITTED]
There are several limitations of this study that should be considered when interpreting the results. First, the model of academic achievement examined is limited in focus. There are clearly many variables that affect a student's performance on a standardized achievement test. This model only included several social variables and does not reflect the complexity of the relationships between academic achievement and the many factors that affect it. Second, the sample size was relatively small for the use of SEM, although the relative simplicity of the tested model does not require as large of a sample size as would more complex models. Third, the generalizability of the results may be limited because of the relatively small sample size and the unique sampling procedure used. Fourth, the present study is correlational in nature and thus caution is warranted when interpreting the path coefficients in the tested models. Finally, the adequate fit of the revised model does not mean that it is the "best-fitting" model for the data. Alternative models may fit the data equally well or better.
In spite of these limitations, the results of this study provide direction for future research related to students' social and academic competence and have implications for the selection of intervention targets aimed at improving children's school performance. Future research can test the model of academic achievement presented in this study on other samples, and examine the role of teacher support and academic self-concept or self-efficacy in student's academic performance.
Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural modeling. Sociological Methods and Research, 16, 78-117.
Caprara, G. V., Barbaranelli, C., Pastorelli, C., Bandura, A., & Zimbardo, P. G. (2000). Prosocial foundations of children's academic achievement. Psychological Science, 11(4), 302-306.
Demaray, M. K., & Malecki, C. K. (2002). Critical levels of perceived social support associated with student adjustment. School Psychology Quarterly, 17(3), 213-241.
DiPerna, J. C., & Elliott, S. N. (2002). Promoting academic enablers to improve student achievement: An introduction to the mini-series. School Psychology Review, 31, 293-297.
DiPerna, J. C., Volpe, R. J., & Elliott, S. N. (2002). A model of academic enablers and elementary reading/language arts achievement. School Psychology Review, 31, 298-312.
Gresham, F. M., & Elliott, S. N. (1990). Social Skills Rating System: Manual. Circle Pines, MN: American Guidance Service.
Gresham, F. M., Elliott, S. N., & Evans-Fernandez, S. E. (1993). Student Self-Concept Scale: Manual. Circle Pines, MN: American Guidance Service.
Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford Press.
Loehlin, J. C. (1992). Latent variable models: An introduction to factor, path, and structural analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.
Lomax, R. G. (1983). A guide to multiple-sample structural equation modeling. Behavior Research Methods and Instrumentation, 15, 580-584.
Malecki, C. K., & Demaray, M. K. (2002). Measuring perceived social support: The development of the Child and Adolescent Social Support Scale (CASSS). Psychology in the Schools, 39, 1-17.
Malecki, C. K., Demaray, M. K., & Elliott, S. N. (2001). A working manual on the development of the Child and Adolescent Social Support Scale. Unpublished manuscript.
Merrell, K. W. (1991). Teacher ratings of social competence and behavioral adjustment: Differences between learning-disabled, low-achieving, and typical students. Journal of School Psychology, 29, 207-217.
Seaman, M. A., Levin, J. R., & Serlin, R. C. (1991). New developments in pairwise multiple comparisons: Some powerful and practicable procedures. Psychological Bulletin, 110(3), 577-586.
Wentzel, K. R. (1991). Relations between social competence and academic achievement in early adolescence. Child Development, 62, 1066-1078.
Wenz-Gross, M., & Siperstein, G. N. (1998). Students with learning problems at risk in middle school: Stress, social support, and adjustment. Exceptional Children, 65(1), 91-100.
Wisconsin Department of Public Instruction. (2005). Educational assessment in Wisconsin. Available online at http://www.dpi.state.wi.us/dpi/oea/assessmt.html
Zins, J. E., Bloodworth, M. R., Weissberg, R. P., & Walberg, H. J. (2004) The scientific base linking social and emotional learning to school success. In J. Zins, M. Bloodworth, R. Weissberg, & H. Walberg (Eds.), Building academic success on social and emotional learning: What does the research say? (pp. 3-22). New York: Teachers College Press.
Date Received: July 22, 2005
Date Accepted: June 7, 2006
Action Editor: Shane Jimerson
Corey E. Ray, MS, is a doctoral student in the Department of Educational Psychology at the University of Wisconsin--Madison, with a specialization in school psychology. Her research interests include the relationship between students' social competence and academic performance, assessment and intervention for English language learners, and the effectiveness of school-based mental health programs.
Stephen N. Elliott, PhD, received his doctorate at Arizona State University in 1980 and is a Professor of Special Education and the Dunn Family Chair of Educational and Psychological Assessment in Peabody College at Vanderbilt University. He directs the Center for Assessment and Intervention Research and also heads up the new Interdisciplinary Graduate Program in Educational Psychology: Learning and Measurement at Vanderbilt.
Corey E. Ray
University of Wisconsin--Madison
Stephen N. Elliott
Peabody College at Vanderbilt University
Correspondence regarding this article should be addressed to Corey E. Ray, Department of Educational Psychology, 1025 West Johnson Street, University of Wisconsin--Madison, Madison, WI 53706; E-mail: email@example.com
Table 1 Demographic Characteristics of Student Participants by Nomination Status (n = 77) Proficient Undeveloped Undeveloped Academic and Academic Behavior Behavior Competence Competence Competence (n = 25) (n = 25) (n = 27) Characteristic n % n % n % Sex Female 9 36 5 20 21 78 Male 16 64 20 80 6 22 Grade 4 14 56 14 56 14 52 8 11 44 11 44 13 48 Ethnicity African American 0 0 2 8 0 0 Asian 0 0 1 4 4 15 Caucasian 22 88 21 84 21 77 Latino 1 4 0 0 1 4 Native American 1 4 0 0 0 0 Other 1 4 1 4 1 4 Table 2 Correlations, Means, and Standard Deviations of the Variables Included in the Modeling Analysis Perceived Social Academic Self-Concept Social Support Skills Competence Perceived Social .44* Support Social Skills .35* .12 Academic Competence .13 .08 .66* WKCE math .08 -.21 .43* .61* WKCE reading .17 .05 .49* .72* M 106.54 2.98 99.82 94.59 SD 13.40 1.57 17.93 14.95 Note. WKCE = Wisconsin Knowledge and Concepts Examinations. Cases with missing values were deleted listwise for the modeling analysis (n = 68). Means and standard deviations are reported as standard scores ([mu] = 100, [sigma] = 15) except perceived social support and WKCE scores, which are reported as z scores. *p < .01.…