Locating Study Respondents After 50 Years

Celeste Stone, Leslie Scott, and Danielle Battle, AIR
Patricia Maher, Institute for Social Research, University of Michigan

Many longitudinal and follow-up studies face a common challenge in locating participants over time. U.S. residents move as often as eleven times in a lifetime—mostly between ages 18 to 45—and about half of those aged 55 and older live in a state other than the one in which they were born, according to the U.S. Census.

1950s computersThe 2011–12 Project Talent Follow-up Pilot Study conducted by AIR survey research experts examined the extent to which a geographically dispersed subsample of participants can be located again after decades with no contact, using relatively low-cost methods. Relying mostly on commercially available databases and administrative records, the follow-up study located nearly 85 percent of the original sample members—many of whom had not participated in the study since 1960.

Study data collected in the base year was used to examine which subpopulations were the hardest to locate. For example, females were located at significantly lower rates than males, and sample members with lower scores on a test of cognitive ability were among those hardest to locate. If certain subgroups are located at disproportionately low rates, this can lead to biased survey results.

The Project Talent study suggests that when differential tracking rates are expected to vary greatly across subgroups, studies attempting to locate individuals after a long hiatus should not rely solely on statistical adjustments to reduce and remove biases. Longitudinal studies can make good use of existing data, and prioritize cases with lower tracking propensities so they can be subject to more intensive tracking methods. Variables correlated with tracking propensity can be incorporated in the sample design to inform tracking-loss adjustments.

Read more about the study in an article published by the Journal of Official Statistics.

Learn more about Project Talent.