*2.6. Social Comparison*

The Active2Gether system uses social comparison, as one of the core ideas behind the system is that a healthy lifestyle can be maintained and achieved in the presence of social support network. Social comparison is implemented in two ways in the Active2Gether system, namely on a group level and on an individual level. On the group level, social comparison is implemented by showing group averages adjacent to the user's physical activity data in the graphs on the dashboard. This allows the user to compare his/her daily performance to other Active2Gether users anonymously.

On the individual level, social comparison is implemented as a ranking of the user's performance within a list of other users, which is shown in one of the panels in the app/website. The ranking automatically updates every time a user visits the website or opens the app. Overall physical activity is used as basis for the comparison, which is determined by the number of steps taken by an individual in the last seven days, but implementations based on activity data for one of the coaching domains are also conceivable.

In order to increase the relevance of the comparative data, the system tries to show actual friends of the user in the ranking. To do so, it extracts friendship relations from Facebook. Facebook is an obvious choice, since it is one of the most popular social media websites globally and also very popular among Dutch young adults. A connection to the Facebook Graph API is established through an open authentication mechanism. As a new user logs in for the first time on the system, it asks the user whether he/she wants to grant access to Active2Gether to check for friendships with other Active2Gether users. If permission is granted and a match is found, the friendship connection is also registered in the Active2Gether system. It is not mandatory for users to grant access, but individuals who do not opt for such explicit social comparison can only observe the activities of other users anonymously, which will probably make the social comparison less effective.

Social comparison can be either *upward* or *downward*, depending on whether an individual compares to targets that perform better or worse. Both variants address different underlying motivational processes. Upward comparison can be beneficial if individuals use the target as a role model and motivation to self-improve, but it can have a discouraging effect if the target's performance seems unattainable [24,25]. Downward comparison can boost an individual's self-esteem and thereby lay the groundwork for self-improvement [26]. However, downward comparison could also have an adverse effect, since it leads to relatively low goals and since it does not challenge an individual to minimize the discrepancy with a better performing individual. An experiment testing the effects of presenting people with their preferred or opposite direction of social comparison in the domain of physical activity showed that it is important to take personal preferences into account [27]. Participants who were shown the type of social comparison opposite to their preference showed a decrease in overall physical activity. Showing the preferred direction of social comparison typically resulted in an upward trend, which was not statistically significant however. Even though the preferences were based on a simple question (whether the participants prefer to compare themselves to individuals who perform better or worse) and the sample size was small, the results demonstrated that these preferences matter: if not to enhance the motivational effects of social comparison, then at least to avoid the potential adverse effects of social comparison [27].

When creating the ranking, the Active2Gether system takes the social comparison preference (upward or downward) of the users into account. The comparison preference of users was determined in the intake questionnaire with the same question as in the experiment described above [27]. The system first tries to find up to six friends whose activity level is in line with the preferred comparison direction. If there are more than six friends that match the preferred direction, the six friends that are closest to the current user in terms of step total of the last seven days are selected. If there are less than six friends in the preferred direction, the system selects other users in the preferred direction. If not sufficient people are found to create the ranking, then the system searches for users in the opposite direction, since it is not appropriate to show an empty ranking list to the user. This could happen, for example, if an individual's preference for social comparison is upward, but his performance is among the best of the users. The data of befriended users are shown with their first name, but data of other users are shown with their initials only, to maintain a level of anonymity. Figure 6 shows the step-by-step process of selecting the friends or other users to show in the ranking.

**Figure 6.** Algorithm for selecting users for the social comparison.
