Sample Size and Design Considerations for School-Based Assessments

• In Event: Measurement and Methodological Challenges and Treatments in Large-Scale Database Analyses

Abstract

Traditional sample designs for large-scale educational assessments typically involve quite large sample sizes within schools. These sample sizes add to the assessment burden and may lead to concerns about monitoring of individual teachers or schools. These concerns add to the risk that the sampled school does not participate, which increases the risk of potential non-response bias. Reducing the within-school sample size may be effective for improving school participation in an assessment and reducing this risk. Following a brief overview of factors that contribute to sample design decisions, this paper reports on an investigation of outcomes from a study using data from USA’s participation in TIMSS 2011, comparing full sample outcomes with outcomes from sub-samples of schools and students.

Author

• Martin Murphy, Australian Council for Educational Research