Palmer Jane E, Winter Samantha C, McMahon Sarah
American University, School of Public Affairs, Department of Justice, Law & Criminology, 4400 Massachusetts Ave NW, Kerwin Hall, Washington, DC, 20016, United States.
Columbia University, School of Social Work, 1255 Amsterdam Ave, Rm 801, New York, NY, 10027, United States.
Eval Program Plann. 2020 Feb 25;80:101794. doi: 10.1016/j.evalprogplan.2020.101794.
The purpose of this study was to examine the final analytic sample of a longitudinal randomized control trial (RCT) evaluation of a sexual violence prevention program at a university after facing challenges with the implementation of a self-generated identification code. The matched and unmatched samples (e.g., all unique surveys across all time periods) included 10,135 surveys. Eighty-eight percent of these surveys were matched into the final longitudinal dataset. Findings suggest that students with certain characteristics were more likely to be matched over time (i.e., students who participated in student government, Latino/a students, and Asian students). In addition, students who did not comply with RCT protocol were less likely to be matched. Student history of victimization or perpetration of sexual violence was not associated with being matched over time. This study provides recommendations for preventing matching problems in longitudinal studies, a process for rectifying matching issues and a critique of studies that do not address issues of matching-related sample bias in their final analytic sample.
本研究的目的是在面临自行生成识别码实施挑战后，对一所大学的性暴力预防计划进行纵向随机对照试验（RCT）评估的最终分析样本进行考察。匹配和未匹配的样本（例如，所有时间段内的所有独特调查）包括10135份调查问卷。其中88%的调查问卷被匹配到最终的纵向数据集中。研究结果表明，具有某些特征的学生随着时间推移更有可能被匹配（即参与学生会的学生、拉丁裔学生和亚洲学生）。此外，未遵守随机对照试验方案的学生被匹配的可能性较小。学生遭受性暴力或实施性暴力的历史与随时间推移被匹配无关。本研究为预防纵向研究中的匹配问题提供了建议，提出了纠正匹配问题的流程，并对在最终分析样本中未解决与匹配相关的样本偏差问题的研究进行了批判。
