A Comparison of Methods for Collecting Self-Report Data on Sensitive Topics
Rosenbaum, Alan, Rabenhorst, Mandy M., Reddy, Madhavi K., Fleming, Matthew T., Howells, Nicolette L., Violence and Victims
Insufficient attention has been paid to whether disclosure rates of sensitive or stigmatizing information vary as a function of method of inquiry. Methods vary both in terms of the anonymity afforded the participant and the opportunity to make a connection with the researcher, both of which might affect participants' willingness to disclose such information. In this investigation, 215 undergraduate students were randomly assigned to complete identical questionnaires using one of the three most common methods of data collection (in-person interview, telephone interview, and paper-and-pencil questionnaire) or an automated telephonic data collection (ATDC) system. Questions on six topic areas of increasing social sensitivity (study habits, substance use, physical and sexual aggression, victimization and perpetration) were included. The results indicated that there were no differences in disclosure rates due to methods and no method by topic interaction, but the two telephonic methods both produced significantly higher participation rates than the two other methods. The results suggest that, at least for a college student sample, an automated telephonic system produces data comparable to that of more traditional methods, while offering greater convenience, economy, and participation.
Keywords: IVR; disclosure; data technology; sensitive information
Interpersonal aggression has long been a topic of academic interest especially to ethologists, sociologists, and psychologists. However, the study of intimate partner violence, child abuse, and sexual aggression (including rape, incest, and child sexual abuse) is a more recent development. Although there is a substantial body of general aggression research employing experimental designs and aggression analogs (Anderson & Bushman, 1997), much of our specific knowledge of family violence and sexual assault has been derived from research designs relying primarily on data provided by victims and perpetrators. Although self-report is a frequent source of data for psychological research, it entails a number of threats to validity, including self-selection biases, problems with accuracy and ability to recall, as well as the desire of participants to be viewed positively. Socially desirable responding is such a common threat that measures of social desirability are often included in social science research as a check on its occurrence (Sugarman & hotalling, 1997).
The pressure on participants to answer in socially desirable ways is magnified when the information being solicited is embarrassing, stigmatizing, or illegal (which is often the case in research on interpersonal violence). Rubin and Babbie (1993) noted that participants might fear being negatively evaluated by the researcher if they admitted to attitudes or behaviors inconsistent with social norms. Willis (1997) suggested that when respondents feel that their privacy is violated by excessively personal questions, they might be more likely to fabricate responses. Relatedly, Arias and Beach (1987) administered both the Conflict Tactics Scale (Straus, 1979) and the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960) to 172 voluntary participants and reported a significant relationship between socially desirable responding and willingness to report perpetration of physical aggression.
Some evidence exists that socially desirable responding may vary as a function of data collection method. Joinson (1999), for example, found that participants reported lower social desirability scores when completing an internet based assessment than when using paper-based methods. Gano-Phillips and Fincham (1992) suggested that telephone interviews offer a decreased sense of anonymity and might, therefore, increase socially desirable responding, in comparison to a written questionnaire.
On the other hand, Brewer, Hallman, Fiedler, and Kipen (2004) found that participants reported more health symptoms via a mailed survey than via telephone interview. …