In this study we assessed the use of pre-recorded video clips of interviewers reading survey questions to respondents in a Web survey environment. In a randomized fieldexperimental survey among 880 participants recruited from an online access panel we compared the video-enhanced Web survey mode to a traditional text-based Web survey using identical questions. Based on social interface theory we predicted higher levels of social presence, social desirability and underreporting of sensitive information in the video-enhanced Web survey. Also, based on the human computer interaction approach we predicted an enriched communication with the respondent when using video clips in Web surveys, which was assumed to increase question understanding and hence yield higher prevalence rates of sensitive responses. Results indicate that the video-enhanced Web mode is not prone to higher levels of social presence, social desirability and underreporting of sensitive information. In fact, both survey modes yielded identical estimates for sensitive behaviors. However, results from structural equation modeling indicate that cognitive processing of survey questions differs tremendously across the two modes. Interestingly, social presence seems to facilitate the reporting of sensitive information in the video-enhanced mode, while it leads to underreporting of sensitive behaviors in the traditional text-based survey mode.