Examining Factors Contributing to Variation in Effect Size Estimates of Teacher Outcomes from Studies of Science Teacher Professional Development

Abstract

Developing and maintaining a high-quality science teaching corps has become increasingly urgent with standards that require students to move beyond mastering facts to reasoning and arguing from evidence. Effective professional development (PD) for science teachers enhances teacher outcomes and, in turn, enhances primary and secondary student outcomes. What constitutes effective PD? Although proposed features of effective PD have emerged, they have little empirical support. Even when impact studies of PD interventions are conducted, factors unrelated to PD design characteristics can influence effect size estimates. Can we statistically isolate PD design characteristics from factors unrelated to PD design that nevertheless influence effect size estimates? We conducted a meta-regression with robust variance estimation (RVE) of 162 science PD intervention studies. We investigated how effect size estimates for teacher outcomes vary by research design, study context, PD intervention characteristics, and outcomes of interest. Intensity and duration of PD, research design decisions, and researchers’ choice of outcomes of interest significantly influence effect size estimates.

KEYWORDS:

• Science
• professional development
• teaching/teacher
• experimental and quasi-experimental
• meta-analysis
• robust variance estimation

Open Scholarship

This article has earned the Center for Open Science badge for Open Data. The data are openly accessible at https://doi.org/10.3886/E109274V4.

Additional information

Funding

This work was supported by the National Science Foundation [1544236].