Usability testing and user satisfaction are cornerstones of electronic survey development. There are nos clear recommendations regarding how self-reported user satisfaction questions (e.g., Chin, Diehl, & Norman, 1988) should be implemented. Human-computer interaction research and theory suggest that the method matters, and that overly-positive ratings are likely when they are collected on the same computer that previously ran the software being evaluated by the user (Nass & Moon, 2000). We use a 2x2 between-subjects randomized experiment to evaluate the effect of two components of social presence on user satisfaction ratings: data collection mode (paper-and-pencil v. computerized, where the computer used for satisfaction ratings is sometimes the computer that ran the software that was usability tested), and physical proximity of the respondent to the computer on which the usability test was conducted (i.e., in the same room as the usability test or a different room). We present data from about 200 respondents, across 5 usability tests of websites and web surveys. Our initial analyses do not support past research findings that participants rate satisfaction highest when answering on the computer that was used for the usability test. We find the lowest ratings when the satisfaction questionnaire was completed on paper in the same room as the usability test (social presence on the proximity dimension, but not the mode dimension). Individual satisfaction scale items asking about information arrangement, clarity, and navigation within the website or survey showed this difference. This difference was also found in a simple summative satisfaction scale. Our findings bring into question whether the robust social presence findings of Nass and colleagues and their theoretical implications (e.g., ethopoeia) are of concern in all human-computer interaction settings, and may provide guidance for usability testers‘ decisions about how and where to collect user satisfaction ratings.