February 3, 2012

Why most Facebook users get more than they give

Part 3: The Structure of Friendship

Some information on the use of social networking sites is extremely difficult or impossible to collect as part of a phone survey. For example, information on the structure of people’s online friendship networks, such as the number of friends of friends, or how densely connected are a person’s friends (i.e., if a person’s friends have all friended each other). Such measures, while difficult to collect in a survey, are important in understanding how use of Facebook is related to different social outcomes. For example, measures such as social cohesion (density) in people’s personal network of relations is a strong predictor of things like trust and social support – the ability of people to get support when they are in need or seeking help making decisions [8].

In this section we look at measures of Facebook use that we could only obtain from logs of people’s actual use of Facebook. Specifically, we examine how these measures relate to people’s everyday experiences outside of Facebook in terms of the amount of social support they receive, trust, and political participation.

A friend of a friend is … probably not your friend on Facebook

As the common saying goes, a friend of a friend is a friend.  But on Facebook this is the exception rather than the rule. When we explored the density of people’s friendship networks, we found that people’s friends lists are only modestly interconnected. A fully connected list of friends would have a density of 1 (everyone knows everyone else). The average Facebook user’s friends list has a density of only .12 (SD=.07). There was a maximum density of .42 (see Appendix A: Graph 1).

As an example of what this means, if you have 10 friends, the number of possible friendship ties among everyone in your network is 45 (possible ties=n*(n-1)/2). If you were an average Facebook user from our sample, with 245 friends, there are 29,890 possible friendship ties among those in your network. Our density measure of .12 means that for the average user with 245 friends, 12% of the maximum 29,890 friendship linkages exist between friends.

A network density of .12 is low in comparison to studies of people’s overall personal networks. A 1992 study found a density of .36 between people’s offline social ties [9]. We suspect that Facebook networks are of lower density because of their ability to allow ties that might otherwise have gone dormant to remain persistent over time.    

There are a number of factors that predict how densely connected a friends list is likely to be.

The longer people have been using Facebook1, and the more Facebook friends they have2, the less dense their friendship networks tend to be. This is consistent with research on other social networks which have found an inverse relationship between network size and density [10].

We expect that new Facebook users typically start with a core group of close, interconnected friends, but over time their friends list becomes larger and less intertwined, particularly as they discover (and are discovered by) more distant friends from different parts and different times in their lives. For instance, for most people there is not much chance that Facebook friends from their high school graduating class will know their current work colleagues.   

Your friends have more friends than you

In our sample, the average Facebook user has 245 friends. However, when we look at their friends, the average friend has a mean of 359 Facebook friends.

The vast majority of Facebook users in our sample (84%) have smaller sized networks of friends than their average network size of their Facebook friends.

The finding that your friends have more friends than you is a near universal feature of Facebook use. It is especially likely to be true for people who have the smallest friends lists. In our sample it was only those participants who had among the 10% largest friends lists (more than 780 friends) that had more friends than their average friend.

The difference in size between a person’s Facebook friends list and that of their average friend is not trivial. The average friend of a Facebook user in our sample has 4.3 times as many friends as the person from our sample.  Even the median Facebook user from our sample with a network of 111 friends – the Facebook user who falls in the middle of our sample – sees their average friend as having a friends list that is nearly two and a half times larger than their own (2.4 times larger).

How can it be that people’s friends almost always have more friends than they do? This little known phenomenon of friendship networks was first explained by a sociologist Scott Feld [11]. Not just on Facebook, in general and off of Facebook, people are more likely to be friends with someone who has more friends than with someone who has fewer.

Facebook users can on average reach more than 150,000 other Facebook users through friends.

A related dimension of this analysis is how many people the average person can “reach” through friends of his or her friends. Again, the average Facebook user in our sample has 245 friends, and their average friend has 359 friends. We also know that the average friends list is interconnected such that 12% of a user’s friends friends are already their friends.

An overly simple calculation would lead us to believe that the average Facebook user in our sample can reach 77,400 people through their friends and their friends of friends (calculated as 245 *(359*1-.12)). However, this calculation overestimates the reach of most people’s Facebook networks. The relatively small number of Facebook users who have very large friends lists disproportionately inflates this average, both because their networks tend to be so large and because their networks tend to be less dense on average. In our sample, the reach at 2-degrees of separation is estimated to be as high as 7,821,772 people (for a Facebook user that had a very large friends list that was not very interconnected). Facebook users from our sample on average can reach 156,569 other Facebook users through their friends of friends. The median user can reach 31,170 people through friends of friends.

Group membership and photo tagging is related to knowing more different types of people

In our national phone survey, we measured the diversity of people’s overall social networks (not just the diversity of their friends on Facebook) in terms of the variety of people they know from different social positions – a broad measure of diversity, not specifically a measure of contact with different racial or ethnic groups, or political perspectives.

In our June report on Social Networking Sites and Our Lives [1], we reported that internet users had relationships that tended to be more diverse than non-users. However, we did not find a relationship between frequent use of Facebook and the diversity of people’s overall social networks. That is, we found that frequent and non-users of Facebook knew a similar mix of people as other internet users.  

With this new data on how our sample actually used Facebook, we are revisiting the diversity of people’s overall social networks to explore how more detailed measures of Facebook use might be related to the breadth of people that one is likely to know. That is, we want to know, are people who use Facebook in certain ways more likely to be sheltered in their exposure to people of different backgrounds and experiences?

Our measure of diversity in our phone sample was based on the well-researched insight that people who know a lot of different types of people have better access to information and resources. We asked respondents in our sample if they knew anyone in 22 different occupations that ranged in occupational prestige.3 It is worth restating that our measure of diversity encompasses not just people’s Facebook friends, but all the different types of people they know.

Based on this sample, we found similar patterns to the ones from our phone survey. The same demographic characteristics in our survey that predicted more diverse social networks still predict social mixing when it comes to examination of server logs showing how they used Facebook (see Appendix A: Table 3).

  • Education is a strong predictor of having a diverse social network. The longer people remain in school, the more diverse people they tend to know.
  • Age is a weak but significant predictor. Older people tend to know more people from different backgrounds.

Few Facebook activities are correlated with the diversity of people’s overall social network. Those relationships that we did find are based on statistically significant, but substantively weak correlations, each sharing only about 2% of the variance with network diversity. 

  • Those who were added to a Facebook group by one of their friends during the month of our observation tend to know more diverse people overall.
  • Those that tag their friends in photographs uploaded to Facebook also tend to know more diverse others.

These findings are consistent with our broader knowledge of relationships. Group membership on Facebook, as well as group membership offline, is associated with knowing people from a greater variety of backgrounds. As we have hypothesized elsewhere [12], photo sharing and tagging on Facebook likely has as much to do with increased exposure to diverse others as it does with increased awareness, by seeing just how diverse existing friends already are. In other words, people who share photos online tend to belong to more diverse groups and they enjoy taking pictures of those groups. 

Tagging friends in Facebook photos is related to having more close friends

Sociologists study social networks in a variety of ways. One important dimension is to examine people’s core relationships: those people with whom people discuss important matters. These “core ties” can be people we interact with online, offline, or most likely both. In our June 2011 report, and elsewhere, we have reported that internet users tend to discuss important matters with more people than those who do not use the internet [12, 13]. As part of our report on Social Networking Sites and Our Lives [1], we found that frequent Facebook users tended to have even more close relationships than those who do not use Facebook.

Facebook activity logs provide the unique opportunity to explore what it is about Facebook that might be associated with having more close ties.

As in our June 2011 report, we find that being female and having more years of education is associated with having more close ties. The only specific Facebook activity that we found to be associated with having more core ties was the act of tagging Facebook friends in photos (see Appendix A: Table 4). However, while statistically significant, the correlation is especially weak. Correlated at .12, photo tagging and core network size share only 2% variation in common. We don’t know what it is about tagging that is related to core ties. It may be that those with more close relationships see those relationships more in-person, and thus have photos to upload of these close friends, or it may be that the act of tagging people in photos increases a sense of intimacy or awareness.      

Facebook use in general is associated with higher social trust, but no specific Facebook activities in particular are tied to trust

In our Social Networking Sites and Our Lives [1] report we found that internet users, and heavy Facebook users in particular, were more trusting than other people. That is, in response to the question “Generally speaking, would you say that most people can be trusted, or that you can’t be too careful with people?”, they were more likely to respond that most people can be trusted. Through the use of Facebook logs, we had hoped to narrow down what it was about using Facebook that was related to higher levels of social trust.

As in our June report, we found that those who were older and had higher levels of education were more likely to be trusting. However, data from Facebook on the structure of people’s friends list (e.g., density), and their participation in specific activities, such as frequency of commenting, did not reveal an association with trust. Frequent Facebook users are more trusting, but it does not appear to be related to their everyday activities on the site (see Appendix A: Table 5). It maybe that those people who select to use Facebook frequently are more trusting by nature – to begin with – or it may be that use increases trust, or there maybe something else about how people use Facebook that is related to higher trust that we were not able to capture in our data.

Making friends on Facebook has a weak, but positive relationship to higher levels of social support

People receive a range of different types of support on and off of the internet. They get emotional support, such as advice; companionship, such as spending time with someone; and more tangible support, such as help when they are sick. In our report on Social Network Sites and Our Lives [1], we found that internet users, and Facebook users in particular received more social support – not just online but from all their relationships combined.

In our survey, we measured support using the MOS Social Support Scale [14] which included measures of “total support,” “emotional support,” “companionship,” and “instrumental aid.”

Activity logs of how people actually use Facebook provide further evidence of the positive relationship between Facebook use and social support (see Appendix A: Table 6). Those Facebook users who received more friend requests and accepted more of those friend requests tended to report higher levels of total support. It is interesting to note that sending friend request that were not reciprocated was not associated with more or less support.

Posting status updates is associated with higher levels of emotional support

When we break down our measure of social support into subscales for companionship and instrumental aid, the relationship between friending and support largely disappears. However, the relationship between receiving and approving more friendship requests remained positive, although still weakly correlated, for emotional support. In addition, those people who made more status updates and wall postings also reported higher levels of emotional support (see Appendix A: Table 7). We suspect that the relationship between making frequent status updates and higher levels of emotional support is the result of feedback that people receive from their Facebook friends in response to their posts. It may also be the result of the positive emotional benefit that is often attributed to simply writing about daily problems [15].

Having more friends and being added to groups is associated with attending political meetings

One of the most substantive and perhaps surprising findings from our report on Social Networking Sites and Our Lives [1] was the strong relationship between the use of Facebook and various forms of political participation. We found that heavy Facebook users were much more likely to attend political rallies and meetings, to try to influence someone they know to vote for a specific candidate, and to vote or intend to vote.

Data on use of specific Facebook activities adds further clarification to our original findings.

A wide range of activities on Facebook were found to be associated with attending political meetings (see Appendix A: Table 8). Although the relatively weak relationship, the number of activities associated with attending political meetings is very high.  Those users who have more friends, have more friends of friends, were either added to a Facebook group or added someone else to a group, sent more personal messages, received more wall posts, tagged a friend in a photo, or were tagged themselves in a photo, were more likely to report that they attended a political meeting or rally.

Those added to Facebook groups are more likely to try to persuade someone to vote for a specific candidate

Other political activities, such as voting and trying to influence others to vote for a specific candidate, are associated with a more specific set of Facebook activities.

Participation in Facebook groups, either by being added to a group or adding someone else, was weakly associated with trying to influence someone to vote in a specific way (see Appendix A: Table 9).

While we did find that Facebook users are more likely to vote in general [1],  we did not uncover any specific Facebook activity that was associated with a higher likelihood of voting. Rather, we found that some activities were weakly associated with not voting – such as having a friend request accepted (see Appendix A: Table 10).

We do not have a complete explanation for why Facebook users in general are more likely to vote, but we found that this tendency is slightly lower among those who have more friend requests accepted or post links on the site.

  1. numoffset=”7″ Pearson Correlation -0.205; sig 2-tailed < .01
  2. Pearson Correlation -0.263; sig 2-tailed < .001
  3. This list of occupations is based on the work of Nan Lin, Yang-chih Fu, and Chih-jou Jay Che, at the Institute of Sociology, Academia Sinica.