Average Effect Size (AES) Quality Indexing
Observational StudyReview
Evaluators evaluating school quality across small sample sizes often face low statistical power and false-positive risks when analyzing multiple correlated indicators individually.
Picture this
Think of inspecting a used car's health by looking at ten individual parts like tires, battery, and brakes separately—each test might not be conclusive on its own. Combining all ten tests into a single standardized average overall index provides a clear, statistically reliable rating of vehicle quality.
What the evidence says
Combining multiple peer quality indicators (mean college entrance exam scores, dropout rates, remedial programs, matriculation fees) into an AES index revealed that voucher winners attended schools with statistically significant lower observable peer quality across academic indicators.
- Who was studied
- N = 300 target secondary schools serving voucher applicants in Bogotá, Colombia.
- How
- Joint estimation technique scaling multiple peer quality indicators into standardized deviation units to compute aggregated Average Effect Sizes while accounting for within-subject outcome correlations.
What to do
Aggregate correlated school quality metrics into a standardized Average Effect Size index to increase statistical power when evaluating small institutional sample sizes.
From the source
"To estimate the Average Effect Size we first scale outcome variables in terms of standard deviation units, and so that positive numbers indicate more desirable peers. We jointly estimate the effects of the voucher on observable measures of peer quality and report these "average effect sizes" for vocational schools..."
Are Educational Vouchers Only Redistributive?