Self-report sexual risk behaviors are a cornerstone of reproductive health-related research, particularly when related to assessing risk-related outcomes such as pregnancy or acquisition of sexually transmitted diseases (STDs) such as HIV. Despite their frequency of use, the utility of self-report measures to provide an accurate account of actual behavior are questioned,[1] and methods of enhancing their accuracy should be a critical focus when administering such measures.[2][3] Self-reported assessments of sexual behavior are prone to a number of measurement concerns which may affect the reliability and validity of a measure, ranging from a participant's literacy level and comprehension of behavioral terminology to recall biases and self-presentation (or confidentiality concerns resulting from stigmatization of the behavior in question).[2][3][4]
Therefore, the decision to incorporate a self-report measure of sexual risk behaviors is often one of practicality. Self-report measures are both inexpensive and more feasible than behavioral observation given the private nature of most sexual risk behaviors. To this end, the validity of self-reported sexual risk measures can be strengthened by the level of concordant answers obtained from sexual partners or through more objective measures of risk (such as the incidence of pregnancy, HIV or other STDs.[2][3]
It is important to consider the way in which measures of self-reported sexual risk behaviors will be collected during the research development phase. Frequently self-report measures are self-administered but may also be elicited from an interviewer, either face-to-face or by telephone; such modalities help address literacy and comprehension confounds but may increase the potential for self-presentation bias. The delivery of risk behavior assessments via the internet or computer can increase a sense of privacy, and may reduce self-presentation biases.[5]
Similarly, it is important to select a self-report sexual risk measure which meets the study's assessment needs. Dichotomous (yes/no) evaluation of engagement in risk behaviors (risk screening), assessing the level of risk via frequency of engagement in risk behaviors (risk assessments) and detailed event-level data related to the co-occurrence of other factors (e.g., alcohol use or primary versus secondary partners) that may facilitate engaging in risk behaviors (risk-event data) are unequivocal, serving very distinct functions in evaluating self-reported risk behavior.[3]
Attention should also be paid to the period of time over which self-reported risk measures ask individuals to recall the occurrence and frequency of engaging in risk behavior; generally, recalling frequency of risk behaviors over a period of approximately three months supports recall accuracy.[2]