Measuring and Adjusting for Longitudinal Attrition in the Survey of Program Dynamics (SPD)

• Activity Number: 74
• Type: Contributed
• Date/Time: Monday, August 12, 2002 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods*
• Author(s): Denton Vaughan*+
• Affiliation(s): U.S. Census Bureau
• Keywords: Nonresponse bias ; family and person reweighting. ; Social Security earnings records ; propensity scoring ; social security number (SSN) reporting

Abstract

Longitudinal surveys like the Survey of Income and Participation (SIPP) and the Survey of Program Dynamics (SPD) suffer from attrition that grows as the survey process is extended out in time. Until recently there has been little research examining directly the effect of attrition on longitudinal change measures. This paper attempts to use social security longitudinal earnings records for that purpose.

Three sets of SSA longitudinal earnings information will be followed: two sets of matched records from the SPD sample, those who continue to cooperate with the survey, and those who attrit, and a third from aggregate administrative data representing the general population calibrated to be similar in coverage to those that are SPD eligible. Many challenging details will need addressing, including adjusting SIPP/SPD for missing social security numbers and how to adjust both famil