Using self-generated identification codes to match questionnaires in panel studies of adolescent substance use

Area of Science:

• Adolescent Health
• Psychology
• Sociology

Background:

• Panel studies are crucial for tracking adolescent substance use over time.
• Maintaining anonymity and confidentiality is vital in sensitive research involving minors.
• Self-generated codes offer a potential method for anonymizing longitudinal data.

Purpose of the Study:

• To evaluate the effectiveness of self-generated codes for linking data in adolescent substance use panel studies.
• To assess potential sample bias introduced by using self-generated codes.
• To determine the optimal criteria for code matching to balance accuracy and completeness.

Main Methods:

• A seven-element self-generated code system was implemented with Irish post-primary students.
• Data linkage was assessed using exact code matches and matches allowing one element difference.
• Respondent characteristics and substance use behaviors were compared between matched and unmatched groups.

Main Results:

• Exact code matching achieved 71% linkage, rising to 88% when allowing one element variation.
• Unmatched respondents showed slight differences (male, lower SES, higher spending money, weaker school/religious bonds, increased substance use).
• Despite minor differences, the combined matched sample closely represented the total sample, with similar substance use prediction models.

Conclusions:

• Self-generated codes are a useful tool for maintaining anonymity and confidentiality in adolescent panel studies.
• Allowing for minor respondent errors in coding enhances data linkage rates without significant sample bias.
• This method supports robust longitudinal research on adolescent behaviors, including substance use.