With a little help from my friends: can social navigation inform interpersonal privacy preferences?

Reviewer: You Chen

In social media, users have many opportunities to specify their privacy setting preferences, such as whether to show their personal information to all of their contacts or only specific contacts. In many cases, users do not understand these settings, much less know how to specify them. Thus, several social media platforms provide the selections of a user's social circle or friends to guide him or her in choosing preferences. The authors of this paper conducted a series of experiments to verify whether the privacy attitudes of one's social circle can serve as a useful guide when specifying one's privacy preferences. They find that the feature's privacy sensitivity is the most important factor influencing a user's selection of privacy preferences. This does not indicate that one's social circle has no influence. In fact, the paper demonstrates that the privacy attitudes and practices of one's social circle can serve as additional guidance when specifying one's privacy preferences; however, this influence is secondary. To verify how one's social circle influences the selection of privacy preferences, the authors asked 68 people to participate in the experiments. Each participant followed two steps. First, each participant installed instant messaging (IM) software and selected six privacy preferences during the installation. The authors divided the participants into four groups, and then reconciled the percentage of high-privacy and low-privacy recommendations provided by their contacts for each group. All of the participants were supposed to believe that the IM software had information on their contacts, although, in fact, it had none. Groups 1, 2, 3, and 4 had, respectively, 15, 20, 15, and 18 participants. In group 1, four out of six settings had high-privacy recommendations from one's social circle, and two out of six settings had low-privacy recommendations. In group 2, three out of six settings had high-privacy recommendations, and three out of six settings had low-privacy recommendations. In group 3, two out of six settings had high-privacy recommendations, and four out of six settings had low-privacy recommendations. In group 4, nobody had recommendations of high privacy and low privacy. Second, after installing IM software, each participant was invited to conduct a post-study of the installation to verify whether or not the participant had been deceived. The results show that only three people did not believe the assumption. Based on the above experiments, the authors analyze the experimental data from two views: the group aspect and the privacy setting aspect. For each group, the authors use the chi-square test to verify how the ratio of high-privacy and low-privacy recommendations influences a user's privacy preference. The results indicate that participants' choices are not influenced by the ratio of high- and low-privacy recommendations. For each setting, the authors analyze whether it was influenced by the recommendations of users' social circles. The results show that the recommendations of one's social circle have no influence on one's privacy preference in settings 1, 5, and 6. It depends on the features of the setting itself. The recommendations of one's social circle have little influence on one's privacy preference in settings 3 and 4, but the magnitude of the effect is very small. The design and analysis of the experiment are very important. The authors develop several types of experiments and analyze the experimental data; however, the description of these experiments is not direct, which makes it hard for readers to understand. Furthermore, the results of the analysis are not clearly provided. Many privacy settings exist in social media. However, this study only looks at six. Future studies will need to verify the authors' findings. Online Computing Reviews Service