An evaluation of a self‐generated identification code

Emory University, Atlanta, Georgia, United States
Research in Nursing & Health (Impact Factor: 1.27). 04/2000; 23(2):167 - 174. DOI: 10.1002/(SICI)1098-240X(200004)23:2<167::AID-NUR9>3.0.CO;2-K


We describe a self-generated coding form used in a study of HIV prevention practices of college students and provide information on the success rate of matching questionnaires over a 3-year period using the form. The data for this study were from a longitudinal study of HIV risk-reduction practices of college students. In order to match questionnaires over the 3-year study period while maintaining anonymity, participants were asked to complete a self-generated identification form at each data collection point. In the second year of the project, we were able to successfully match 74.3% of the questionnaires to those returned during the first year using 6 to 8 of the code elements on the form, and in the third year, we were able to match 73% of questionnaires to those returned in the second year. Participants for whom questionnaires matched were more likely than participants with unmatched questionnaires to be white students enrolled as underclassmen. © 2000 John Wiley & Sons, Inc. Res Nurs Health 23:167–174, 2000

6 Reads
  • Source
    • "Although the utility of SGICs has been improved somewhat recently, for example by using a set of identity questions that are more convenient to form individually specific SGICs, problems associated with such practice still remain. Studies with adolescents have reported a sizable difference in health risk behavioral outcome measures between the matched and unmatched groups (e.g., DiIorio et al. 2000; Kearney et al. 1984). Recently, Schnell et al. (2010) highlighted concerns about the high proportion of non-matched subjects in studies that use SGIC to link data across multiple time points and make suggestions for improvements. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Self-generated identification codes (SGICs) are an increasingly utilized methodological feature of longitudinal prevention research among adolescents. This study sought to test the differences between the matched and unmatched groups at baseline on a number of background, health, and well-being and risk behavior measures in a prevention study among 13- to 16-year-old Icelandic adolescents where a SGIC was constructed and used to link individual-level respondent data over two data collection points one year apart. We use pilot data from two Reykjavik city secondary schools collected as part of the population study Youth in Iceland in February 2010 and 2011 (N = 366, SGIC matching rate 61 %). Baseline results for the matched and unmatched participants are compared. Findings indicate that the unmatched subjects are both more likely to be substance users than their matched counterparts as well as being more likely to be boys and/or from disrupted families. Five out of the seven scaled measures for risk and protective factors and personality indicators reveal no difference between the matched and unmatched subjects and the significantly different measures reveal small effect sizes between the two groups. However, the effect sizes for substance use are significantly different between the matched and unmatched groups for all seven substance use measures with effect sizes from 0.52 to 1.32. These findings therefore indicate that the measurement validity of adolescent risk behaviors such as substance use may be put in jeopardy when using SGIC and that unmatched subjects may be more likely to distrust the SGIC process.
    Full-text · Article · Feb 2013 · Prevention Science
  • [Show abstract] [Hide abstract]
    ABSTRACT: Programs to prevent substance use among high risk youth can pose numerous challenges for program evaluators including FEASIBILITY ISSUES such as Participant Recruitment and Retention, Identifying High Risk Youth, and Obtaining a Control/Comparison Group; MEASUREMENT ISSUES such as Social Desirability Bias and Instrument Reliability; METHODOLOGIC ISSUES such as Attrition (both selective and differential), Inadequate Implementation and Variable Dose, Low Statistical Power, Contamination of Comparison Groups, and Low Literacy Skills of Participants. For each of these challenges specific solutions are offered for researchers and practitioners. In general the solutions relate to three themes; 1) incorporate evaluation early into program design; 2) develop a strong partnership with program staff; and 3) be flexible. Also addressed are the need to acknowledge possible bias toward documenting positive outcomes as well as the need to customize evaluation designs for different settings and populations.
    No preview · Article · Feb 2001 · The Journal of Primary Prevention
  • [Show abstract] [Hide abstract]
    ABSTRACT: In panel studies on sensitive topics, respondent-generated identification codes are often used to link records across surveys. However, usually a substantial number of cases are lost due to the codes. These losses may cause biased estimates. Using more components and linking the codes by the Levenshtein string distance function will reduce the losses. In a simulation study and two field experiments, the proposed procedure outperforms the methods previously applied.
    No preview · Article · Oct 2010 · Evaluation Review
Show more