Multi-Stage Administrative Linkage Protocol
Observational StudyReview
Tracking student outcomes over multi-year horizons in developing nations often suffers from substantial measurement error and missing records in primary administrative identification numbers.
Picture this
Think of finding a lost library book using a system where ID stickers often fall off. Instead of giving up when the bar code is smudged, the librarian searches by the bar code, then by bar code plus city branch, and finally by bar code plus the author's full name, successfully locating books that would otherwise be declared lost.
What the evidence says
Applying a three-tiered matching protocol (ID alone, ID and city, ID and name) recovered match rates ranging from 18.8% to 25.5% among vocational applicants, establishing a statistically significant 4.9 to 6.1 percentage point increase in examination participation for voucher winners.
- Who was studied
- N = 3,422 human secondary school voucher applicants (1995 cohort) in Bogotá, Colombia.
- How
- Longitudinal administrative record matching connecting randomized lottery baseline data to national college entrance examination (ICFES) databases using multi-stage deterministic algorithms.
What to do
Implement multi-pass record matching algorithms using secondary geographic and demographic identifiers to minimize attrition bias when merging administrative datasets with imperfect primary keys.
From the source
"To improve the accuracy of matching, we used multiple matching strategies - matching by ID alone, matching by ID and city of residence, and matching by ID number and name."
Are Educational Vouchers Only Redistributive?