A simulation-based workflow and web app for matching self-generated identification codes in longitudinal surveys
Longitudinal research links participants’ responses across waves, yet sensitive topics often require anonymity. Self-generated identification codes (SGICs) reconcile both, but participants may sometimes mistype, skip, or swap code parts. Most studies handle such mismatches with ad hoc rules of unknown accuracy. We present a simulation-based workflow and introduce the web app wavelink for choosing a customized matching procedure. To link waves, researchers must describe their design, SGICs, and expected errors; wavelink then simulates studies with known true links, compares exact, rule-based (tiered), and probabilistic procedures, recommends the procedure that misses the fewest links within a set false-link limit, and exports an R script for linking the researcher’s study data. To illustrate, we report a three-wave school panel of adolescents, for which error-tolerant procedures found 91–92% of true links versus 73% for exact matching. Simulations across conditions show that code length limits accuracy and that auxiliary variables compensate for short codes.
Authors
- Felix Dietrich (ORCID: https://orcid.org/0000-0002-0696-3380)
- Dora Weubel (ORCID: https://orcid.org/0009-0004-1652-667X)
- Sophie Mayen
Publication Details
- Journal
- PsyArXiv (OSF Preprints)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.31234/osf.io/26ua5_v1
- Primary Topic
- Survey Methodology and Nonresponse
- Type
- preprint