The Influence of Social Capital On Test Scores: How Much Do Families, Schools & Communities Matter? Glenn D. Israel, University of Florida Lionel J. Beaulieu, Southern Rural Development Center February 2002
Why Study Test Scores? Common measure for assessing students’ educational outcomes (e.g., FCAT, SAT) Increasingly used to measure performance of schools (e.g., Florida’s School Performance Grades) Key factor in setting educational policies, including resource allocations (e.g., President Bush’s ‘No Child Left Behind Act of 2001’)
Simplified Model of Educational Achievement School Social Capital Community Social Capital Educational Outcomes Test scores Grade average High school completion Structural attributes Process attributes Structural attributes Process attributes Family Social Capital Structural attributes Process attributes Background Attributes Student’s ability Parent/student attributes School resources Community resources
Elements of Family Social Capital Structure Parental structure, number of siblings, and climate (sibling who dropped out) affect opportunities for interaction Process Nurturing activities, e.g., assisting with homework, discussing school plans, encouraging college Monitoring activities, e.g., limiting TV time, checking on homework, present after school
Elements of Community Social Capital Structure Local capacity, inequality; isolation; and instability affect opportunities for structural differentiation and integration Process Linkages between youth and others in the community, e.g., member of a church group, involved in other youth groups, number of moves Closure in networks, e.g., parents know friends parents
Elements of School Social Capital Structure Composition of student body, school size and resources, quality of teachers and staff, and school climate influence opportunities for relationships among students, teachers and parents Process Nurturing activities, e.g., student and teachers talk, student perceive teachers to be nurturing Monitoring activities, e.g., amount that student’s parents contact the school
Data & Analysis Methods Data Source National Educational Longitudinal Survey (NELS) surveyed 8 th grade students, parents, teachers, and school administrators in 1988 Follow-up surveys conducted in 1990 & 1992 NELS data supplemented with data from School District Data Book, UDSA-ERS & ICPSR Analysis Methods Descriptive statistics and factor analysis used for selecting variables and data reduction Hierarchal Linear Models using student-level and school/community-level variables used to estimate parameters for 8,756 students and 577 public schools
R 2 by Community Type
Family Structural Attributes: Predicted Change in Test Scores 1P 2P Ot Parent Structure Number of Siblings No of Sib Dropouts
Family Process Attributes: Predicted Change in Test Scores 0x 1-2x 3x+ Discuss school w/ parents No of parents expecting college N R S O How often TV time limited
Family Process Attributes: Predicted Change in Test Scores Non-adjacent nonmetro N R S O Parents check homework Adjacent nonmetro N R S O Metro other N R S O Metro core N R S O
Family Process Attributes: Predicted Change in Test Scores Time alone after school Metro core N 3 Metro other N 3 Non-adjacent nonmetro N 3 Adjacent nonmetro
Community Structural Attributes: Predicted Change in Test Scores Socio-economic capacity: metro other Percent of voter Participation, ‘88 Percent living in the same county
Community Process Attributes: Predicted Change in Test Scores No Yes Involved in a religious group Number of moves No of Non- religious groups
School Structural Attributes: Predicted Change in Test Scores Expenditures Per student Percent of free & reduced price lunch: Non-adj. Nonmetro Percent of free & reduced price lunch: Metro core
School Structural Attributes: Predicted Change in Test Scores Extent of school problems Positive school atmosphere School emphasis on academics
School Process Attributes: Predicted Change in Test Scores Number of sch. clubs student in How much teachers nurture Teachers talk w/ outside of class
School Process Attributes: Predicted Change in Test Scores No Yes Involved in other school organizations Student’s parent contact school N 1 2 >4x How much parents do in PTO
Conclusions & Implications Background and family variables account for most of the explained variance in test scores – between 19 and 24% Process social capital variables for families, communities and schools have significant influences on test scores & these are malleable Reliance on standardize test scores as the performance measure for schools is not supported by research