ESRA 2027 sessions by theme

Back to Overview of Sessions

Session Details

Survey researchers and the organizations they represent are ethically and often legally bound to prevent allowing individual respondents to be identified in any statistical estimates or in any data sets they release. Members of the public expect this “disclosure avoidance,” and they expect that information they provide will not be used against them. The range of protections varies in different kinds of organizations and in different countries, with full population censuses facing different concerns than sample surveys. But the spectrum of options ranges in the same way, from treating all data as proprietary (i.e., never releasing data at all, which fully protects respondent confidentiality) to fully releasing original data sets for potential reanalysis (promoting open science and reuse of data but potentially compromising confidentiality).

Protecting respondent privacy is becoming increasingly challenging as new ways to algorithmically reconstruct respondent identities become possible. Individual respondent identities can be reconstructed from multiple released statistics or crosstabs from a single data set, or from augmenting what is released in a single data set with available information from a different data source. Figuring out what to do about what Hawes et al. (2025) call the “triple tradeoff” between data availability, accuracy and confidentiality (increasing one or two of these necessitates decreasing another) is thus harder than ever, even as the demand for trustable data and aggregate statistics continues to grow.

This session is intended to promote discussion about new challenges in balancing legitimate competing needs and demands for data while meeting ethical requirements–addressing the triple tradeoff. Suitable submissions could include concrete discussion about specific surveys, data sets, and data use cases in official statistics or in the private sector, as well as about principles for respondent protection that