Methods for Linking Administrative Records When Unique Identifiers Are Not Available: Using Project Talent and the Death Master File

• Activity Number: 282
• Type: Topic Contributed
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Social Statistics Section
• Author(s): Christopher Plotts and Amber Noel*+ and Ashley Kaiser and Shaun Posey and Samantha Neiman
• Companies: American Institutes for Research and American Institutes for Research and American Institutes for Research and American Institutes for Research
• Keywords: record processing ; longitudinal study ; administrative data ; social security number ; mortality ; record linkage

Abstract

Use of administrative records is an increasingly important part of social and behavioral research because it offers a relatively inexpensive and quick alternative to a full scale data collection. Previous studies have established that unique identifiers such as social security numbers (SSNs) are vital for linking survey data to administrative records. However, few have offered alternative approaches for linking survey data to such records when unique identifiers are not available. Without unique identifiers, administrative data linkages are prone to reductions in accuracy and increases in erroneous linkages. Erroneous linkages are particularly problematic to large-scale studies because they require more manual review, thereby increasing study costs and introducing more coder-specific errors. Using survey data from Project Talent and data from the Social Security Administration's Death Master File, this study uses a seeded sample of records with SSNs to evaluate and compare three methods for linking to administrative mortality records when SSN is not used. We will evaluate each method on the proportion of false negatives and false negatives produced, as well as overall efficiency.