The Health and Retirement Study 2022 Social Security Permission Experiment

Principal Investigator: Helen Levy, PhD
Research Professor, University of Michigan

Background:
Administrative data linkages are widely viewed as a promising strategy to improve data quality in light of the increasing difficulty and cost of conducting surveys (National Academies of Sciences and Medicine 2023). The Health and Retirement Study (HRS) asks respondents for consent to link their survey responses to administrative data on earnings and benefits from the Social Security Administration. In 2022, the team conducted an experiment to see if the placement of this consent request within the survey matters. Specifically, they randomized the approximately 6,000 respondents scheduled to get this request in 2022 into two groups. Half were asked for consent at the end of the survey, according to usual practice, while the other half were asked midway through the interview in the context of questions about retirement and Social Security benefits. Early analysis of the data suggests a significant increase in consent for those asked midway through the survey, consistent with earlier work on this topic (Sala, Knies, and Burton 2014; J. Sakshaug, Tutz, and Kreuter 2013; J. W. Sakshaug et al. 2019).

A higher consent rate may mean more linkage consent bias or it may mean less, depending on whether question placement exacerbates or offsets existing differences in consent rates across groups. Results in the literature are mixed on how demographic characteristics affect the likelihood of consent. The studies of question placement cited above have relatively small samples (about 1,200 to 2,400 respondents) and very limited data on respondents’ personal characteristics. The HRS experiment, in contrast, has a much larger sample and a rich set of characteristics, including race and ethnicity, for a nationally representative sample of older Americans.

Current Proposal:
The team proposes to analyze the underlying patterns of consent for groups defined by age (55-64 and 65+), education, race/ethnicity (White non-Hispanic, Black non-Hispanic, Hispanic [any race]), gender, cognitive ability, and self-reported health status, as well as the increase in consent associated with question placement for each subgroup. The results will shed light on the nature of linkage consent bias and how it is affected by question placement.

References:
National Academies of Sciences and Medicine. 2023. “Toward a 21st Century National Data Infrastructure: Enhancing Survey Programs by Using Multiple Data Sources.”
Sakshaug, Joseph, Valerie Tutz, and Frauke Kreuter. 2013. “Placement, Wording, and Interviewers: Identifying Correlates of Consent to Link Survey and Administrative Data.” In Survey Research Methods, 7:133–44. https://ojs.ub.uni-konstanz.de/srm/article/view/5395.
Sakshaug, Joseph W., Alexandra Schmucker, Frauke Kreuter, Mick P. Couper, and Eleanor Singer. 2019. “The Effect of Framing and Placement on Linkage Consent.” Public Opinion Quarterly 83 (S1): 289–308.
Sala, Emanuela, Gundi Knies, and Jonathan Burton. 2014. “Propensity to Consent to Data Linkage: Experimental Evidence on the Role of Three Survey Design Features in a UK Longitudinal Panel.” International Journal of Social Research Methodology 17 (5): 455–73. https://doi.org/10.1080/13645579.2014.899101.