Professional communities and student achievement – a meta-analysis

Abstract

In the past 3 decades, the concept of professional community has gained considerable momentum in the theoretical and empirical studies in this field. At the same time, the concept has faced conceptual and methodological difficulties in that as yet no universal definition has been formulated and that its operationalization differs in the various empirical studies conducted on the subject. This study presents a comprehensive synthesis of the theories currently available and their implications for the conceptualization and operationalization of the professional community concept including a meta-analysis of the studies that investigated the effect of professional community on student achievement. Our meta-analysis reported a small but significant summary effect (d = .25, p < .05), indicating that within a school environment professional community could enhance student achievement. Furthermore, the need for the conceptual and empirical validation of the concept's key dimension was discussed.

Keywords:

• professional community
• meta-analysis
• secondary education
• student achievement

Notes

1. For a detailed understanding of the comprehensive school reforms and their effects on student achievement, we recommend the study of Borman, Hewes, Overman, and Brown, (2003).

2. PiCarta is a meta-catalogue of OCLC PICA, which contains titles of books, illustrated magazines, and articles (see http://nl.wikipedia.org/wiki/PiCarta).

3. Authentic achievement is a special type of measurement of student achievement, which wasdeveloped by researchers at the Center on Organization and Restructuring Schools (CORS) in the United States during the comprehensive reforms. Authentic achievement contains multidisciplinary standards, measuring knowledge construction, disciplined inquiry, and value beyond school. It is based on the assumption that achievement in one domain says something about a pupil's intellectual capacities in other domains (Louis & Marks, 1998).

4. We started our analysis with the study of Lee and Smith (1996). As presented in Table 3, for each study we specifically focused on the effects, standard errors, p values, and the number of schools. Lee and Smith (1996) used multilevel modeling to take the nested structure of the data into account. In addition, their study reported on the relationship of professional community implemented for a number of subjects, namely mathematics, reading, history, and science. Moreover, considering that teacher control is not a common indicator of professional community, we included in the analysis only the coefficients for responsibility for learning (.33, p < .001) and cooperation (.15,nD). Bearing in mind that the characteristics of professional community were introduced as separate items in the analysis, only the effects for one subject, specifically mathematics, could be covered for the two items. Specific effect sizes were obtained, as presented in Table 3, with a mean effect size value of .24. Given that, the authors reported the results in effect sizes, specific computations were applied using the standard deviation at school level, and a value of student-level correlation of .11 was obtained.

  Louis and Marks (1998) conducted a multilevel modeling approach. Their research sample consisted of 24 schools, and they found a positive and significant effect of professional community on student achievement (β = .26, p < .001), while controlling for elementary level and student background variables. Taking into account these relevant characteristics, the student-level correlation coefficient of .26 was kept.

  Supovitz (2002) operationalized the professional community concept as group instructional practices and applied multilevel modeling to investigate its relationship with student achievement in 79 schools. He found a small and significant effect on student achievement. Supovitz (2002) investigated the results of five subjects: writing, reading, mathematics, science, and citizenship in five grades: Grade 4 to Grade 8. Considering that the age range of the students in these grades was similar to those in the other four studies, the scores for Grade 6 (a middle grade) were used for our analysis (see Table 3). Next, the mean variable (Borenstein et al., 2009) of the five outcomes for Grade 6 was calculated on the basis of 23 teams applying group instructional practices, with a student-level correlation coefficient mean value of .15 (r = .15).

  Visscher and Witziers (2004) investigated the effect of professional mathematics community on achievement of students in mathematics in 39 schools, using multilevel data analysis. They found a significant effect. In this case, the authors reportedthat 6% of the 20% school-level variance (p < .10) was explained by the professional community variables. In sum, professional community accounted for 1.2% of the total explained variance among students (see Appendix 1, Formula 1). After performing other calculations required (see Appendix 1, Formula 2), the student-level correlation coefficient obtained between professional community and student achievement was .11.

  Finally, Bolam and colleagues (2005) performed a correlation analysis at school level toinvestigate the relationship of professional community with student achievement in 153 secondary schools, correcting the student achievement measures by relevant student and school covariates. We call this measure of student achievement overachievement. The authors found a school-level correlation of .16 (p < .01) for the professional and student learning ethos factor and a correlation of .18 (p < .01) for the within school policy, management and support for professional learning factor, both representing the professional community concept with a correlation mean value of .17. Using specific computations (see Appendix 1, Formulas 1 and 2) and assuming that the sign and the direction of the relationship is the same at student level, a student-level correlation coefficient of .18 was obtained.