Self-generated identification codes (SGICs) successfully matched approximately 72.65% of a longitudinal study sample of secondary school students over a one-year period, while revealing that male students and younger students were less likely to be perfectly matched.
The overall matching rate using SGICs was approximately 72.65%, with varying levels of match quality.
Male students were significantly less likely to be perfectly matched compared to female students.
Younger students (Year 1) were significantly less likely to be perfectly matched compared to older students (Year 3).
Matched participants (both perfectly and partially matched) were less likely to have missing values compared to unmatched participants.
Matched participants were more likely to exhibit positive attitudes toward the sexual health program and related topics compared to unmatched participants.
The SGIC comprised a structured 5-element code used to link baseline and follow-up data collected approximately one year apart.
The authors concluded that further refinement of SGIC generation processes and matching algorithms is needed to minimize data wastage.
This research investigated a method for tracking anonymous participants across different time points in a school-based sexual health study in Hong Kong. Instead of using names or other identifying information that could compromise privacy, students created their own personal identification codes (called self-generated identification codes, or SGICs) — a combination of letters and numbers based on personal information they would remember, such as birthdate digits or initials. Researchers then tried to match each student's answers from the beginning of the study to their answers one year later using these codes. The study found that about 73% of student records could be successfully linked across the two time points — roughly half were matched perfectly, and about a quarter were matched partially using a flexible algorithm. However, about 27% of records could not be matched at all, representing lost data. Male students and younger students (first-year secondary school students) were less successful at generating consistent codes, making their data harder to match. Importantly, students whose records could not be matched tended to have more missing data and less positive attitudes toward the sexual health program, which means the unmatched group was not just randomly missing — it was systematically different, potentially skewing study results. This research suggests that SGICs are a practical tool for protecting student privacy in longitudinal research while still allowing researchers to track changes over time. However, the meaningful proportion of unmatched data — and the fact that unmatched students appear to differ from matched students — highlights the importance of designing clearer code-generation instructions and better matching algorithms. These improvements could help researchers capture a more complete and representative picture of outcomes in future studies involving anonymous participants.
Choi E, Andres E, Fan H, Ho L, Fung A, Lau K, et al.. (2025). Using self-generated identification codes to match anonymous longitudinal data in a sexual health study of secondary school students: a cohort study.. BMC medical informatics and decision making. https://doi.org/10.1186/s12911-025-03028-1