Hybrid Administrative Record Linkage Optimization
Evaluating social interventions via administrative datasets often lacks universal unique identifiers (such as Social Security Numbers), introducing linking errors that bias experimental treatment effect estimates.
Picture this
Think of matching name tags at a huge reunion where some people left off their middle initial or misspelled their last name; requiring 100% perfect spelling misses many real people, but combining exact matches with a smart fuzzy-matching algorithm catches almost everyone without mixing up strangers.
What the evidence says
The hybrid four-tiered matching protocol achieved a 1.97% false positive error rate and a 2.46% false negative error rate overall (compared to exact matching alone which yielded a 0.8% false positive rate but an 8.0% false negative error rate), minimizing overall attenuation bias in binary outcome estimation.
- Who
- N = 163,447 program applicant records linked to adult criminal records from the New York State Division of Criminal Justice Services (14.26% total match rate).
- How
- Evaluation of a four-tiered linkage protocol combining pairwise exact matching (name, date of birth, social security number) with a probabilistic matching tier validated against a benchmark sample (Tahamont et al. 2020).
What to do
1. Combine exact deterministic matching tiers with validated probabilistic matching algorithms when linking administrative records lacking universal identifiers.
From the source
"The four-tiered matching protocol that we used was validated by DCJS against a known sample and has very low false positive and false negative error rates (1.97 percent false positive error rate and a 2.46 percent false negative error rate)."
The_Effects_of_Youth_Employment_Evidence_from_New_York_City_Summer.pdf