Sexual Assault Research in the Military: Is Oversampling Necessary for Representativeness?

INTRODUCTION

Sexual assault in the U.S. Military is a serious concern. Recruiting representative samples of service members to participate in sexual assault research is essential for understanding the scope of the problem and generating data that can inform prevention and intervention efforts. Accordingly, the current study aims to examine response and completion rates of an anonymous survey of sexual assault and alcohol use among active duty sailors aged 18 to 24 with the overarching goal of achieving a representative sample and informing future recruitment efforts.

MATERIALS AND METHODS

This study was approved by the Institutional Review Board at the Naval Health Research Center. The study involved an anonymous survey of sexual assault and alcohol use among 612 active duty sailors aged 18 to 24. Since 79.6% of Navy service members are men and 20.4% are women, women were oversampled to achieve sufficient representation. Survey invitations were emailed to 12,031 active duty sailors: 64.3% (n = 7,738) men and 35.7% (n = 4,293) women.

RESULTS

Response rates were disproportionate, with 3.0% (n = 234) of male and 8.8% (n = 377) of female sailors responding to the study invitation. Survey completion rates, however, were similar between male and female sailors (81.2% and 80.1% for male and female personnel, respectively).

CONCLUSION

Results demonstrated that female sailors were significantly more likely than male sailors to participate in a study of sexual assault and alcohol use. However, once enrolled in the study, male and female sailors completed the 234-item questionnaire at a similar rate. Study findings highlight the challenges of engaging male service members in sexual assault-related research. Despite the disproportionately high representation of men in the military, sexual assault researchers may need to sample according to the overall distribution of gender in the military or perhaps even oversample men to achieve a representative sample.