Self-Reported Social Network Behavior: Accuracy Predictors and Implications for the Privacy Paradox

Self-reported behavioral data is frequently relied upon to understand the population of social network users. These data often consist of self-reported posting, commenting or general engagement frequency within the social network over the last few days or a month. Using a sample of 397 Google+ users, we show that these data can be quite inaccurate when asking users to report on actions that tend to occur infrequently and irregularly (e.g. profile field editing). Indeed, the regularity of the behavior is a strong predictor of self-report accuracy. Users who exhibit a behavior either very frequently or very infrequently are the most accurate in their reporting. For social networks, in which it is often the case that most users are ``lurkers´´ who do not post or comment much, our study suggests that questions should only refer to a very narrow and recent time window to improve response accuracy. Our study also highlights the importance of considering the granularity of privacy concern measurements when investigating the so-called privacy paradox. Within our sample, those users who report that the ability to control post visibility and/or delete posts are more important than other, non-privacy related, features, do take more privacy actions. In particular, this group is less likely to enter profile information, more likely to limit the visibility of their posts and more likely to delete posts.