The purpose of the study was to investigate the effects of survey mode, gender, and perceived sensitivity on the quality of data gathered on sensitive health behaviors. The study adopted a stratified random sampling method in order to select 600 males and 600 females from all undergraduate students who were enrolled in a large Mid-western university. Both male and female students were randomly assigned either to the paper-and-pencil mail survey or to the web survey. The participants were classified as the sensitive group or the not-sensitive group according to their responses on perceived sensitivity of questions about substance use, body weight, and sexual behavior. The quality of data was measured on response rates, number of unanswered items, and mean score differences on reporting of various health behaviors.
Response rates between the mail survey (38.3%) and the web survey (35.2%) were not significantly different. However, the web survey had more unanswered items than the mail survey (p < .001). Regardless of survey mode, there were more female participants (45.1%) than male participants (28.2%) and there was no significant difference on number of unanswered items between males and females. Also, there were no significant differences on number of unanswered items between the sensitive group and the not-sensitive group. However, when the numbers were broken down by survey mode and gender, there were somewhat significant differences. Results from 2 x 2 x 2 analysis of variance revealed that there were no interaction effects between the survey mode, gender, and the perceived sensitivity except on body weight and number of sexual partners. Also, no main effect of the survey mode was found concerning various health behaviors. However, main effects of gender and the perceived sensitivity revealed that male students and those who perceived the questions to be sensitive were more likely to report engagement in sensitive health behaviors. These findings support the notion that the survey mode, gender, and the perceived sensitivity play significant roles in the quality of data pertaining to sensitive health information. Researchers should consider potential effects of these factors when collecting information on sensitive health behaviors in order to improve the quality of data.