Statistical Procedures for Determining the Validity and Reliability of
a Multidimensional Survey Instrument: The School Success Profile
Gary Bowen
School of Social Work
The University of North Carolina at Chapel Hill
301 Pittsboro Street
Chapel Hill NC 27599-3550
USA
Phone: 919-962-6542
FAX: 919-962-3689
Email: glbowen@email.unc.edu
Roderick Rose
School of Social Work
The University of North Carolina at Chapel Hill
301 Pittsboro Street
Chapel Hill NC 27599-3550
USA
Phone: 919-962-8826
Email: rarose@email.unc.edu
Natasha Bowen
School of Social Work
The University of North Carolina at Chapel Hill
301 Pittsboro Street
Chapel Hill NC 27599-3550
USA
Phone: 919-843-2434
FAX: 919-843-8715
Email: nbowen@email.unc.edu
Objective: Valid and reliable assessment of students is a starting
point for evidence-based practice in schools. This presentation reviews
both the steps and the findings of a comprehensive psychometric evaluation
of the School Success Profile (SSP). The SSP is a 220 question self-report
survey that examines middle and high school students' beliefs about their
neighborhood, schools, families, peer groups, and themselves. The SSP,
which is widely used for student-level intervention planning by school
systems across the United States, results in one-page summary profile for
each student respondent that includes information on 22 dimensions related
to the student's social environment and individual adaptation. Information
about the validity and reliability of the SSP is critical for school professionals
who depend on quality assessment data for informing individual and group
interventions.
Methods: Data were collected between 2001 and 2003, and included
SSP results for16,636 middle and high school students from nearly 400 schools
in multiple states. Each scale dimension (14 social environment dimensions,
8 individual adaptation dimensions) was examined for normality, internal
consistency, standard deviation of error, factor structure, measurement
invariance (sex, race/ethnicity, grade level), construct validity, and
either predictive validity (social environment dimensions) or clinical
validity (individual adaptation dimensions).
Results: Results support the psychometric integrity of the SSP
as an assessment tool across race, gender and grade level. The factor and
construct validity analyses support the composition of the scales from
the SSP items. Social environment dimensions predict individual adaptation
dimensions, which subsequently demonstrate clinical validity across socioeconomic
status.
Discussion: Psychometric data is lacking on many comprehensive
youth surveys used in schools and communities. The analysis reported provides
guidelines for evaluating multidimensional survey instruments, and demonstrates
the quality of the SSP as a practice tool for schools and communities that
want to improve adolescent outcomes.