 And welcome to Nicole Ellison. Nicole is a professor at MSU in the Department of Telecommunication, Information Studies, and Media. She's also a visitor this summer at MSR in working with Dana Boyd and a couple of other folks who are also in the room, I believe. She's here to talk to us about social capital on Facebook. Welcome. Sure. Thank you. Thank you. And thank you again to the Berkman Center for allowing me to share my work. I am a, my training is a communication scholar. As some of you here are, and the real focus of my work has been for many years how these tools enable us to form and maintain social relationships. I want to point out that all the work that I'm presenting today is co-authored. And the names of my collaborators are on the screen. And as I said, the question kind of animating this particular piece of work is thinking about the social and technical affordances of social network sites, specifically Facebook. And really trying to understand what are the motivations behind its use. So what do people get from using Facebook? So here, for instance, is a small capture of some of my Facebook friends. What meaning do they have for me in my life? And the kind of theoretical perspective that we have used to express some of these benefits is social capital. So social capital, the original friends with benefits. This is an established theoretical perspective in the sociological literature. And it essentially describes the resources that can be deployed and activated through your social relationships. So factors that may influence this is kind of the kinds of people that you have in your networks, the size of your network, for instance. And so if you think about the people that you know, you probably have different associations and different benefits from them. So for instance, with close friends, we might look to them as a source for social support. I'm actually not going to speak about that form of social capital in this presentation, although we've have published on that topic elsewhere. The focus of today is bridging social capital. And bridging social capital is typically associated with weak ties. So Granaveter's work on the strength of weak ties speaks to the kinds of diverse information that we are more likely to receive from bridging ties, which are more likely to be weak ties. And so Granaveter's work specifically looked at the found that his subjects were more likely to find out about valuable job opportunities from people that they didn't see as often. And the idea is that we spend a lot of time with people that we're close with due to homophily pressures. We tend to be attracted to people who are like us. And so people who are less like us are more likely to be different in productive ways and useful ways. And Judith Donoth and Dana Boyd pointed this out back in 2004 that technologies, and I'd insert here such as Facebook, that expand one social network will primarily result in an increase in available information and opportunities. So thinking about what is a particular kind of instance in which we see this happening. And question asking in Facebook or social network sites is a very nice example of a specific instance where people are trying to mobilize these informational network, informational resources in their network. And this is really the focus of some of our more recent work. So for instance, here's an example. I ask what should I do with the kids before leaving Boston? I get a variety of responses from my social network. Some of them are more useful than others. And a couple of things are happening here. So first of all, thinking about norms of reciprocity. And this was mentioned in one of the lens definition of social capital, which I kind of skipped through. But let me just pull it out. So he's talking about this notion that social capital is an investment in social relationships with this expected returns in the marketplace. So if we think about this instance, the fact that these people are answering my question will probably make me more likely to, in the future, reciprocate. And the other thing to point out is that there's a signaling story that's happening here. So essentially, in a site like Facebook, we send out status updates. And we really don't have a clear sense of who actually sees them. The algorithm decides who gets to see what. And unless someone explicitly responds to a comment or a status update, you really have no sense of whether they've even seen it, whether they're paying attention, whether they care. So there's this kind of valuable signaling process here that's happening when people explicitly answer questions. Our early work, so we have a piece in the Journal of Computer Media and Communication in 2007 called The Benefits of Facebook Friends. And in that piece, we used an undergraduate sample. And we found a relationship between Facebook use, measured by Facebook intensity, and social capital. And really, what we've been doing since then is trying to refine our measures and get a better understanding of what is the mechanism behind this relationship. So why is this happening? And that's the work that I'm presenting today, which isn't, it's not published yet. I'm still very receptive to feedback about how we're thinking about and interpreting our analyses. So thinking about why we think Facebook might influence social capital. And I tend to be attracted to the notion of social and technical affordances. So the idea here is that not that Facebook causes X, but rather that it makes possible a set of opportunities that human agents can then choose to take up in various ways. So it's moving away from a technologically deterministic perspective. So when we think about it with that lens and thinking about what are the affordances that Facebook has, we know that it supports the maintenance of a larger network of weak ties. So many of us have friendship networks in the hundreds. And this is probably the barriers to kind of maintaining that are lowered. It becomes much easier to kind of keep in touch with this wider network. It also, the information in the profile and user activity is very useful for enabling people to kind of connect with latent ties. And so there's a piece that we have in press and new media in society that talks about social information seeking as a particular strategy that Facebook users employ, whereby they find out information about someone that they have some kind of offline connection with. They don't necessarily know this person. Maybe they just met them very quickly at a party. Maybe in the case of undergraduates, this is someone that is in their dorm. They've never really spoken to them. And the idea is that by engaging in what could otherwise be called Facebook stalking, you're finding out about this person. You're finding common ground. You have topics of conversation. You might be more willing to engage them in a face-to-face interaction. And then finally, just from a kind of transaction cost perspective, that Facebook enables users to broadcast these requests for information, advice, and recommendations. And it makes it very easy to respond to those requests. So when I say a request for information or social support, this doesn't necessarily need to be in the form of, I'm having a really bad day. People, please tell me you love me. It could be just I'm having a bad day. And this is one of the mechanisms by which people receive social support. So in kind of the broader picture, some of the work that we're trying to address with this particular data collection effort is determining whether the patterns between social capital and Facebook use that we've identified in undergraduate samples hold true for adults. And we're also really thinking about what are the specific behaviors that are associated with social capital accrual. So if you think about research looking at internet use and internet effects, it becomes very clear that it doesn't necessarily make sense to just look at whether someone is online or not, because there's such a range of activities that they could be doing, which would affect the way in which they're accessing the social capital in their network. And so the same, I think, is true of Facebook. And if you think about the wide range of activities, it becomes very clear that just kind of measuring this kind of global measure of Facebook intensity as we did in our earlier work doesn't necessarily make sense. It doesn't necessarily give us the kind of specificity that we need to kind of start to unpack which behaviors are going to be more productive from a social capital perspective. And then finally, we're thinking about this notion of the network on Facebook. So who are these friends? Are all friends equal? Are all Facebook friends equal? And so thinking about the kind of proposition that I put forward that Facebook enables this larger network of weak ties and that this increased this more heterogeneous kind of network enables access to diverse information, you would think that it's kind of a linear relationship. The more friends, the better. But that's actually not true. And I'll speak to this in a minute. But this was, again, something that we were interested in kind of parsing out is the relationship between number of friends and social capital. So the data set that I'll be that was used in the analysis that I'll be sharing today was collected fairly recently last fall and spring. We are, as I said, attempting to kind of move beyond the student sample. So we are with this case where we've actually our survey sample was MSU staff. So these are individuals who are non-academics. They're not teachers, they're not professors. They're secretaries, administrators, those kinds of positions. And you can see that the average age, for instance, 45 years old. So definitely a different demographic than we've looked at before. And we employed a multi-method approach. So we have a survey instrument that we asked our participants to complete. And everything I'm presenting today is just based on the survey data. But we also have a whole set of other methods that we're still in the process of working through. So for instance, we used Bernie Hogan's Facebook application that basically captures the actual Facebook network of friends. And so this will enable us to do some interesting things. We're looking at what are the kind of network characteristics of that friends network that are more predictive of social capital, for instance. We have a series of other kind of pseudo-experimental techniques that we've used. We did an interview with our participants. And we also asked them to request a favor of their network. And then we were captured who of their network was responded to that favor. So in that sense, we're really trying to think about, OK, well, what is a particular instance in which we can capture an actual mobilization of the network? And we didn't want to use a question because a question by its very nature is some people can answer, some people can't. We thought that would bias it. So this was supposed to be kind of the lowest common denominator of things that people could do. OK, so the survey measures, and I'll talk about each of these individually, but just to kind of give you a sense of where we're going. We're predicting bridging social capital. And then we have a set of IVs that we use to be able to understand those relationships. So this is our dependent measure. This is Facebook-specific bridging social capital. This is based on adapted from the Williams JCMC piece, which actually is 2006. Sorry about that. And we've kind of gone back to his original scale. So we have all 10 items. The difference here is that instead of the online offline, which we felt didn't necessarily make sense in this context, we have specifically tried to kind of narrow in on the benefits specific to the site and the connections that are articulated in that space. So you'll see the items are things like interacting with people in my Facebook network, making me want to try new things, interacting with people in my Facebook network makes me interested in what people unlike me are thinking. So really trying to capture this notion that these are people who are different from you. OK. So moving along to our IVs, in our early work, we asked how many total Facebook friends do you have at MSU, in this case, or elsewhere. And this was combined with minutes on Facebook per day and a set of attitudinal items. And we kind of wrapped that into this Facebook intensity measure. Now we're trying to kind of take it apart to be able to see specifically a little better what is the kind of effect of these individual items. And so we still ask how many total friends do you have, Facebook friends. But we've also asked in recent work, approximately, how many of your total friends do you consider actual friends? And this is the in-press piece, Ellison et al. in New Media and Society, actually uses this measure to predict social capital. And so essentially what we found is that the total number is not a significant predictor, but the number of actual friends is. And just to give you a sense of the descriptives around this. So from our 2008 undergraduate data set, that's the one that was published in the in-press piece, the median total number of friends was 300. But the number of friends that they reported as actual friends was 75. So about 25% of undergraduates' Facebook network would be considered by them actual friends. We did not give them any definition of what we meant by actual friends. We left that for that to interpret. So with our adult data set that I'm presenting today, you can see that the median, there's fewer friends, but roughly the same percentage are considered actual friends. And when people say, well, what do you think they mean by actual friends? My sense of that is that these are people who you're actually interacting with. There's a lot of contacts on these sites that are just not activated. Maybe we've hidden them. Maybe they've hidden us. We just don't necessarily engage with them. And when we look actually at data from Facebook itself, this is a 2009 memo from the Facebook data team, they actually show that the percentage of people that you're having this kind of reciprocal communication with is a fairly small percentage. So here I just want to kind of lay out what we know about the nature of Facebook friends. So as I said, we know that actual friends are more important, at least in terms of predicting social capital than total friends, that we have these large networks but we only interact with a small number of them, a small number of friends. There's a very nice piece that I would recommend that Mora Burke, Bob Crout, and Cameron Marlow has published earlier this year, where they're actually using server data from Facebook itself predicting social capital. And so they can really get at a level of granularity that we can't with survey instruments. And what they found was that directed communication with individual friends was predictive of social capital gains, but not passive consumption, so kind of lurking, reading, but not posting back, or broadcasting. So the things that they are looking at, I think are things like sending messages, liking something, posting on someone's wall, directed is the key here. But then I think kind of coloring all of this is this notion that we don't necessarily know who is seeing our content when we're posting it, and that this is the newsfeed algorithm. It's very obscured unless someone is specifically commenting back. So we developed a measure that we call cultivation of social resources. And this captures kind of instances in which you are, so a sample item is when I see a friend asking for advice on Facebook, I try to respond. So the idea is that you specifically, you are reporting specifically responding to requests that you see among your Facebook network. And this was an extremely powerful predictor. This is kind of a behavior that we should all do if we're trying to kind of harness the social capital benefits of social networks that use. And essentially, when we're thinking about why this might be, there's a few things going on. So one is when you specifically respond, you're kind of creating this expectation about reciprocal behavior, going back to Lynn's concept of social capital as expected returns in the marketplace. There's also a social grooming function that we think is happening. So thinking about social grooming, this idea that these are activities that affirm relationships. They display bonds and alliances. And they also, in this case, signal attention. So the idea here is that you are kind of telling people you're important to me. I see you. The third, which is kind of a technical consideration, is this idea that engaging in these behaviors can kind of train the newsfeed. So for instance, if I'm responding to a lot of what Fred is posting and he then kind of responds back with a thank you or whatnot, I can expect that I'm showing up probably more predominantly in his newsfeed because Facebook interprets me as a relevant source of information for him. And then last but not least, these kind of behaviors, I think, can kind of activate these dormant ties within the network. So these are people who are just out there. And by responding to them, you can kind of bring them back into the fold of your actual friends. So this is the actual scale. So I think you can all probably read that. But again, the notion here is that these are behaviors that probe the respondents' willingness to try to respond when they see a request from someone in their Facebook feed. They're activating ties that essentially through this norm of reciprocity. I want to point out that there is one that is slightly different, which is when a Facebook friend has a birthday, I try to post something on their wall that's slightly different in that it's not kind of responding to a request in the same way. But again, there's a very kind of strong social grooming story I think there that you're kind of paying attention to someone where you want to signal that. OK, we also included an item that measures, I'm sorry, a scale that measures information seeking behavior. And this is attempting to kind of measure activities, the extent to which people are employing Facebook in a specific way as a channel for seeking information. So they're engaging in this instrumental engagement with the network for information related goals, such as asking questions about health issues, getting questions to specific answers, getting product recommendations, business referrals. So these are people who are actually specifically looking to Facebook to kind of information for solving information needs. So when we do, so essentially what we do is we have a series of models here. These are nested OLS regressions. So again, we're predicting bridging social capital. So this is just with the kind of controls, if you will. So information seeking and cultivation of social resources aren't in there yet. I'll show this in a moment. But what I want to point out is that actual friends on Facebook is significant, but not the total friends. And also, Facebook minutes per day is as well as gender. When we introduce the information seeking behavior, I use Facebook to get questions from my network. You can see that that becomes significant, and actual friends is still significant. But total time on Facebook is not. And also, the R square increases here to 31 from 14, something, 14. So this is the third model where we've introduced the cultivation of social resources. So again, that's that kind of reciprocity measure that I talked about. And here, you can see that the R square goes up now to 0.45, and that that is a fairly powerful predictor of these self-reported assessments of bridging social capital. And I'm not going to show it, but we did test for an interaction effect between cultivation of social resources and actual friends, which was significant. And so this is work that we're still trying to figure out. But I'll present it anyway. So essentially, we did a simple slopes calculation to try to figure out what is this kind of interaction between actual friends and cultivation of social resources. And you can see that at all, so this is cultivation. These are people who report high levels of these cultivation of social resources behaviors. And you can see that all of these are higher than the people who are low. What's interesting is that for the blue line is low actual friends. And you can see that for these individuals, there's actually an interesting effect where they are actually reporting higher levels of bridging social capital than people with high numbers of friends who are not engaging in these behaviors. So essentially, this is a very powerful way of even if you have a few actual friends, if you're engaging with them in these way, in these CSR behaviors, you're actually activating them and going to report higher social capital from it. So essentially, I'm going to close up here. I just wanted to frame this as what we've been trying to do over the last, I would say, five years now is really focusing on specific behaviors, and moving away from global measures of use to better understand what are the mechanisms behind social capital and social network site use. The information seeking variable is one that we have identified. And then I think more interesting, at least for me, is this cultivation of social resources. And the idea that these are particular activities that thinking about the social and technical affordances of the newsfeed and of Facebook use that these are activities that actually have a very clear story in terms of communication, so as a social grooming in terms of social capital, in terms of these norms of reciprocity, and then just also from a technical perspective, the idea that you are training the newsfeed. So this is what we will be spending time thinking about and that I would welcome thoughts and feedback on. Just in terms of future work, I think I mentioned that we had the network data. And so now we're starting to see the relationship between the specific network, right? So who's friends with who among your Facebook friends, those kinds of social network variables, and our social capital measures. We're also moving through our data set. So looking, for instance, at those requests to see who responds, who doesn't. And then also spending a bit more time looking at privacy issues, which was not an original focus of the work. But we've really come to understand that most of these social capital benefits are predicated on the notion that you are disclosing something, right? If you are not going to tell your network that you're sick or that you're looking for a new job, it's much more unlikely that you'll be able to extract that information and social support from your network. And so the extent to which these disclosures then are going to kind of unanticipated audiences or are kind of going to different parts of your life that you want to keep separate really make it clear that controlling who the audience is is kind of the key to be able to enable people to disclose in meaningful ways while still avoiding privacy concerns, context collapse, a lot of these kind of the downfalls of unintended audiences. So that's it. All our papers, including the impressed one, is on my website. And I wanted to point out also this is work supported by the National Science Foundation, much to the chagrin of conservative senators everywhere. Thank you. So I'd love feedback or questions or? Uh-huh. In your previous slide, you mentioned that broadcasting does not build as much social capital as direct interaction and interpersonal communications. But I mean, this probably touches on privacy and the way that Twitter works. So from my own personal experience is that my posting, I received, I did not realize how many people read that. And it's actually significantly higher than I realized. And how do you think this effect of other people knowing more about you than you realize changes the dynamics of people's relationship with each other, especially in the university environment? Sure. So let me just point out that that finding is actually from the Burke et al paper. So it's BURK is the last name, Mora. And so I would encourage you to go to that piece. But let me address your other question about this notion that we have information, I think, about more information about others. So I think the real concern is when we are not necessarily aware of the audience. So in the kind of anecdote that you gave, you said that you were getting feedback of some sort that kind of made you realize, oh, these people are in my audience. Offline feedback. I don't want to talk you into that. OK. So that's actually, I think, a system where that has worked in some way. So now you know. And of course, a better design would be that you have better awareness of that from the beginning. But I think the more concerning action would be when you have audiences that you're not aware of or that you are where there is a real need to not disclose information to them. So for instance, if I find that I have a disease, this may be something that's important for me to share with my close friends. I don't necessarily want on my work colleagues. I don't want the HR person at my organization knowing this. And so with Facebook specifically, especially as we see it adopted by so many more adults, there is much more of an opportunity for context collapse where people have segments of their identity or their network that they are able to compartmentalize in offline settings. And here they are all mushed up into one big audience. And so I think that's the concern. Alice Marwix, who's here, is written on this as well. Uh-huh. I sometimes will have a separate Facebook list and actually will make status updates viewable only by certain groups of people. Is that a common thing? Or is that the only type of thing that a computer geek like myself would actually bother with and understand? I actually defer to Fred who's done more work specifically looking at privacy settings. My sense is that it's not all that common. But we have a chapter coming out that looks at privacy issues. And basically we're arguing that using lists like that is actually a very productive strategy because it enables you to still disclose but to avoid the kind of unwanted audiences. So I think that's actually a productive strategy. It does not seem to make it pleasant to do with why but you can do it. Sure, sure. Yeah, and I think my sense is that people are not doing that to the extent that you might expect. Just to add on to that, there's been some recent news stories that younger kids now are, they're figuring out an app. I think there's an app out there that can say this information can't go to such and such people meeting their parents. So I think it's starting to become more. Sure, sure. So there's probably specific audiences that we can all think of that might be motivating to take the trouble to do that. So in the case of a child, perhaps a parent. Sure. Oh, sorry. Thank you. Could you remind me what your dependent variables were and if they were all self-reports? So it's bridging social capital. This is a perceptual assessment of the extent to which you are accessing different parts of your being exposed to new information. This is something that we're thinking about and working on. So for instance, in another piece, we use a social provision scale that's kind of validated out there in the social support literature as a way of moving away from the bonding social capital and trying to look a bit more specifically. But yes, they're all self-reported. Do you have the desire to travel always for your measures? Correct. That was the DB. Yeah. So the favor asking is kind of our attempt to get a better sense of so that we're not just asking about perceptions, but we can actually see who did this favor for these people when they requested it. The limitation there is that we don't necessarily we can't assume that everyone in the network saw that favor because of the algorithm. So that's a challenge that we're going to have to think about. What's your significant hypothesis for the gender difference that you found? Well, I think, let's see. So we're women, well, although it falls out in the cultivation of social resources, in the final model when that's, so it may be that women are engaging in those behaviors more responding to requests. What other social networks they're on? Because I know a lot of people will use, let's say, LinkedIn for one motivation, Twitter for another, if it's a health related issue. Let's say if they have a wellness or caring page or another issue. So did you ask something that got to how many other social networks there may be using? We did. I don't think it's, I mean we would probably look at that independently. And I think it's an interesting question because when you do divide up, when you have different sites for these to address different components of your life in a sense, what I think that enables is deeper self-disclosures and more subject specific interactions. The kind of downside is that when you're thinking about the diversity of a network, if you've got it kind of parceled out like that, you're not necessarily going to have the kind of cross fertilization that might happen. So for instance, when I post a message, a status update, and two friends who don't know one another are engaging in a dialogue, I mean I think for me that's a very kind of exciting instance because these are people who are from different networks. I mean this is kind of the Bert's kind of structural whole. But they are interacting in a forum that I think is going to, I would speculate, would be have a different kind of tenor than if these were two people who had different political beliefs and were in an anonymous chat room, for instance. Like I would expect that this, because they are visible to their networks, because they have me as a friend in common because of the kind of norms of Facebook, that that interaction would be kind of qualitatively different between two people who didn't necessarily agree on something. So I guess to kind of get back to your point is that I think that when you have these kind of segments, which I agree people do, and LinkedIn is probably the best known example of a professional network, that you lose the opportunity for those kinds of cross fertilization from different parts of your network. I have a question about the interaction effects. Oh, OK. I mean that's something, yeah, OK. So just kind of thinking about the relationship between high, medium, and low, and the outcome variable. I'm wondering if you considered rescaling that as a proportion, because I assume you're just using means. And so somebody could have very few actual friends, but lots of not actual friends. And somebody could have very few actual friends, and that'd be a big proportion of their network. And it seems like there might be some relationship to the type of social capital you could accumulate. And so I'm wondering. OK, so looking at kind of percentages. Yeah, exactly. As opposed to just raw numbers. As opposed to raw means. Yeah, it seems like the distributions would be different. Yeah, I mean we can certainly look at that. I guess my kind of understanding of who these non-actual friends are is that they're essentially invisible. I mean they're there, if you go to your friends list, you see them, you're not really engaging with them. And so one, if you look at the numbers of how many actual friends, I mean they're like 100. So this isn't just close friends. These are people who I think you have some kind of engagement with on the site. And so I guess I would be less inclined to think that it would necessarily matter because my sense is that those people are just kind of invisible. And perhaps until you post something that you don't want them to see. So I don't know. I mean is your, so what is the kind of rationale for thinking that that percentage would matter? Well I guess I'm trying to use that relationship between how non-actual friends don't figure into the estimation of bridging social capital. Because through the news feed and things like that, we're definitely going to be seeing people that we don't consider to be our actual friends, which I see could be part of or contribute to an understanding of bridging social capital. And so the ratio that I'm talking about is kind of thinking about, OK, so somebody could have like five actual friends, actual friends, but hundreds of non-actual friends. And I think they would be sort of a different type of person in the data set than somebody who has five actual friends and 10 total friends. And especially considering the type of population that you're studying. And so treating as a ratio might produce a different effect. Yeah, no I can see that. And I think really maybe this is where our network data may be interesting to try to kind of suss out who are those non-actual friends. I guess my theory of the ratio is the interaction effect would go away. OK, OK. Yeah, I guess my, so I think different assumptions. My assumption would be that people, if you're seeing them in your news feed, you would consider them in that actual friends. Yeah, I mean that's that I'm thinking of all these people who are just like basically invisible. OK, yeah, we have different assumptions. Yeah, yeah. So, yeah. Uh-huh. Yeah, I was thinking about this. It seems like there might be an interesting middle category there of people who are acquaintances, but not close friends. And so if you have a high ratio of people who you consider to be, I mean if you have a low ratio of people you consider to be close friends versus your net overall Facebook friends. I'm curious if you would see a different sort of pattern of behaviors. And that was the question I was going to ask is if you saw with different numbers of friends people had different patterns of what uses they made, like did they use private messaging versus broadcasting, for example. Because it seems to me like broadcasting often isn't done in an anonymous way. It's done with an assumed audience. Sure. And so I wonder whether people who broadcast a lot tend to be people who have a high ratio of close friends to their overall. And I also wanted to ask how you're thinking of using the network structure data in the bigger project. Sure, sure. So the network data, one, so looking at kind of the network characteristics. And I'm not a social network analysis person. So this is Bernie Hogan at OII is actually the one who's doing a lot of this work. But the idea is to kind of think about what is the thinking what we know about social network theory and what we would expect about, for instance, structural holes and clusters and things like that. How does that kind of map on to our findings around social capital? But I think perhaps even more interestingly would be thinking about this. So we have the self-report survey data and the network data from the same participant. So even though it's a smaller subset, we can look at the way in which those kind of social behaviors moderate the network characteristics. And I think there will be some interesting effects that we'll be able to see that we're starting to look at. And that's really kind of in a similar fashion showing that engaging in these cultivation of social resources is actually a way of kind of compensating for a network that on the face of it would be less productive from a social capital perspective. And again, I think it has to do with this idea that you're kind of activating these ties. Remind me what your other question was. Whether or not you see distinctions in the affinity for different types of behaviors on Facebook, different types of communication practices by those who fit in the high versus low friends categories. So some using more private method messaging versus others broadcasting and things like that. Sure, so we didn't ask specifically about that kind of granularity of behavior. What I can tell you is that in the in-press piece, we've identified three kind of suites of communication behaviors. So one of them is interacting with close friends on Facebook. And the mean for that is very, very high. So if you have a friend, a close friend on Facebook, you've friended them, you've interacted with them on Facebook, et cetera. On the other extreme, we have a kind of scale that measures the extent to which you are kind of looking for strangers to friend on Facebook. And this, as you would imagine, it's not a normative behavior. It's very low. We have a third kind of set of behaviors that we call social information seeking. And this describes the extent to which people are using the site to find out more information about someone that they have some kind of offline connection with. It's not a total stranger. But they haven't necessarily kind of activated that tie yet. And so we hypothesize that people are using the kind of identity information that they discover in Facebook to do things like find out, what do we have in common? What's our common ground? What can we talk about? If I'm sitting next to you in a large class. So for instance, one of those items was, I use Facebook to find out information about someone in a large class, something like that. And so the idea is that we would speculate that you find out you have something in common. And it kind of lowers the barriers to a face-to-face interaction. It makes it easier to talk about something that you already know you have in common. I think the interesting question is, how exactly do you introduce that information without it seeming creepy? So that's kind of a whole nother set of questions. Facebook stuff. Yeah, yeah, exactly. So anyway, to kind of tie it back to the social capital story, of those three suites of communication behaviors, connection strategies we call them, only that social information seeking was predictive of actually both forms of social capital. And I think that, so the idea is that if you've already have these close friends, you're already interacting with them in other ways. Facebook isn't really going to have a meaningful impact on how you kind of exchange social support with a close friend. And then on the other extreme, the kind of looking for complete strangers is not very normative. I mean, I think that's a clear kind of descent into creepiness. But this middle ground, that's where I think there's some very interesting dynamics around using online information to facilitate a meaningful connection in some way. And whether that is online interaction or offline interaction, it doesn't really matter to me. Mm-hmm. Oh, I'm sorry. You had your hand up before. Yes, to you. Yeah. I was just curious, I know, if the people who have a lot of social capital through Facebook were also the same type of people who had social capital in the offline world, or if suddenly they serve a new group of people who are shy, normally but are suddenly empowered by all these online tools. Sure, sure. So actually, in the past, we've used a measure of social capital that kind of specifically references MSU for our MSU undergraduate samples. In this data collection effort, we've done something different, which is we actually ask the same set of measures twice. One is asking about interactions with their Facebook network, and that's what I presented here. We have another set of scales that is the same essential question, but we've asked them to think about everyone in their social network, kind of online and offline, and then answer the questions thinking about just your broader social network, not constrained to Facebook in any way. And we find the same patterns, but the effect is less. Like essentially, the R squared or lower, it's not. Facebook use isn't making as much of a difference, but it's still significant. But that's the kind of thing that we're definitely trying to suss out. I'm just wondering, you mentioned audience a couple of times. That really brings it back to communication for me, too. Mary and Dana, I believe, had an article in New Media Society about imagined audience. And I'm wondering, in the data that you had in the interview data, if anybody had talked about who they envisioned as their audience for maybe their network, did they talk about that at all? It came up a couple of times. And so, for instance, one of the things that we were interested in looking at is the extent to which people were using it for professional purposes, which is, of course, one of those instances where there's some kind of high stakes there in terms of what you're disclosing to people you work with. So I can't speak to the interview. It's still being transcribed. Yeah, so I can't say anything meaningful. Although, if anyone else wants to, no. OK. I think that there's a couple of things that it might depend on. I'd be very curious if in six months or whatever, you find some of this stuff. I think that from what we got, what we got a general sense, and we don't say this in the paper because we didn't have firm, clear data for it, was that on Twitter or Facebook, people tend to think of the people who pop up in their news feed or in their Twitter feed as the people that they're talking to, which is exactly what Nicole is talking about when she talks about actual friends, like the people that you see. So you think of those people as people you're talking to, and that's who you imagine speaking to when you're writing a post for that particular audience. And I think that that speaks to this sort of, if you're lurking, people don't know you're there. I mean, obviously, that's what it means to be a lurker. But also, you don't get to become part of that audience. And then I think there's other groups of people who have specifically segmented their audience in particular ways. And whether or not they've been successful with that or not, I think then they imagine that they're speaking to like, OK, well, this is my Twitter account that I only use for professional purposes, right? Or this is my Facebook account that I only use to communicate with my family. Whether or not those are actually the people that they're speaking to or not, I think. It's a very different case. I mean, and so to think carry that, so I would imagine that the disclosures you make when you've constrained that audience may be more productive for kind of bonding social capital. But that segmentation is limiting the extent to which you have that kind of more heterogeneous network. Yeah. We interviewed so after they tried to make the request. When were they interviewed? I believe they were done after they tried to make the request. But we gave the friend network two days. So they didn't necessarily. And the person wouldn't have necessarily known who had done the favor or not. Oh, I see. Yeah. And the other question I have that comes off of that is how reflective were the people that were in the sample, the 666 people that were in the sample, how reflective were they about maybe how them just participating in this study changed their Facebook habits? I can't speak to that, but I think that the act of asking them to ask their network, in terms of any of our measures or any, I mean, I would point to that as probably being the most kind of perhaps disconcerting or enlightening, depending on how you want to frame it. And we actually had a couple of people who refused who said, I don't want to put this out there in my status update, because they know that somebody, in this case you or your fellow researchers are watching them, maybe are getting more friends or are posting just a little bit more than, let's say, their norm. Because suddenly they know that they're part of this Facebook study. So we weren't really gathering data at that level about behavior. So I don't think they would have any sense that we were kind of following them in any way. I'm sorry. Was there a question back here before? All your friends on Facebook would be equal. So someone who's been a lifelong friend and someone who's just in a class with you can have become equal in this group of Facebook. You can be a broadcast, a broadcast to everybody. And that kind of with the person that you just met in a class, you kind of skipped the natural process of becoming someone's friend. So I want to know what you thought about if Facebook has at all kind of redefined the notion of what a friend is. So OK, so just to clarify, I don't think all friends are equal. Right? Yeah, yeah, yeah. But in terms of audience, when you broadcast, you do kind of broadcast to everybody in the same way, whether you're thinking of them or not. Sure, sure. So I think your question about Facebook kind of redefining friendship is very interesting. And I'm sure we've all kind of seen popular press stories that kind of decry the death of friendship because no one can have 500 friends. And that's a kind of meaningless concept. And it's a violation of friendship. And I actually think that our findings actually point out the fact that people are cognizant of the fact that 75% of their network is not an actual friend. So I think that we're kind of doing a disservice to users if we assume that they kind of have this uniform understanding of friendship. I think that people are much more savvy than that and that they have a very clear sense of friendship is a term that Facebook used to describe a connection on the site. So my sense is that I think we are probably, as humans, aware of that and that I don't necessarily think that we need to be worried. And not that you were doing this, but there's been a fair amount of kind of coverage of this idea that somehow Facebook is killing the notion of friendship. I think it's definitely redefining the ways that we're connecting with people that we're maintaining these weak ties, that we're kind of finding common ground with people. So I think there's a lot of changes that are happening, but I don't think we need to be worried about kind of redefining friendship in that way. This would be a small point, but I think it's still worth mentioning. Are you at all worried about like a recency bias in that people that you recently added or recently met are more likely to appear in your newsfeed than old, long-standing friends? Particularly if people are accumulating friends relatively quickly, maybe with the size of the networks we're looking at, it doesn't really matter. But for some people, I think it could be an issue. Yeah, sure. I mean, I think, again, I think that just thinking about the one item, I don't know. I mean, again, I don't think we're getting at that level of specificity. I think people probably, first they're like, what do you mean by actual friends? And then they just try to kind of take their best kind of understanding of what that means. And yeah. Well, I went on my phone and looked at my newsfeed, and a lot of those people I wouldn't consider actual friends. Whereas many people I consider actual friends don't often appear on my newsfeed, particularly if I live in the same city with them and interact with them sort of outside of Facebook. Sure, sure. And of course, there's different platforms. So for instance, when I'm on my laptop, I'm seeing a very different newsfeed than when I'm on my phone. So yeah, I mean, I think that the interesting and the very, very challenging thing for a researcher is kind of how to measure and kind of conceive of all these offline interactions that are kind of a black box as far as we're concerned. I mean, we're very, very careful in our measures to, so for instance, with the ones about try to respond. That response may be, I see you face to face, and I answer that question. Or I call you up because I see that you just announced that you were sick. So we're very, very careful about trying to not limit ourselves to only online interactions. Because I think what these two kind of online and offline, we've been kind of thinking of them as separate worlds, and they're so fluid. And there's just kind of a greater communication ecology in which information we get in one channel is affecting the way we interact in others. So I think that's a great question. So this whole New Media and Society piece that I've been talking about, what we were trying to do with those scales is kind of unpack a finding that we had in 2007, which was the extent to which people are, is it online to offline or offline to online? And we realize that's probably not the most useful way to think about it. And so again, but I can't, for me to ask you, did you, is this an online friend or an offline friend? Or where did you, I mean, maybe our first interaction was face-to-face, but it was only because I saw information about you online. How do you capture that in a survey question? So I mean, these are the kinds of things that keep me up at night. Just thinking back to differentiating actual friends and leaving that amorphous, is there any more maybe specific typology that emerged from your interviews? Because again, reflecting on personal experience, my closest or my actual friends, I don't interact with them at all on Facebook. Whereas the people I interact with tons on Facebook, they're sort of Facebook friends that we have these great discussions, but it's just limited to that. And we don't really interact outside of that. So I'm wondering if you saw anything similar to that. Well, so again, and I'm sorry to keep saying this, but we haven't looked at that data yet. But so for instance, what we did do is we asked our participants to show us an example from their Facebook feed of an instance where they had asked a question, and then we had each of them rate the closeness of each of the people who responded. So for instance, that would get us at some of that. The network data I think also would give us kind of a vocabulary for talking about that. So yeah, I think there's just a lot of interesting work that can be done to really kind of untangle these different levels of closeness and utility and interaction and everything that's going on. Uh-huh, Fred. Oh, and sorry, I didn't see you. Yeah. So I think this information search variable is really cool and very important. This one. Yeah, I'm kind of wondering, maybe asking to speculate a little bit. Sure. Do you think that this variable is different in this population as compared to undergraduates? And do you think that it's changing over time so is Facebook getting more important as an information search location? I, yes, to the later ladder question. And I think in terms of, and that's I think really, there's a lot of attention right now being paid to kind of social search and the idea that people are going to kind of harness their network because they will give, because that's the place where they think they will get more relevant information. So for instance, in our interview data from another sample, we had a woman who was talking about how when her child was sick, she went to Facebook and asked what medicine she should give her child and instead of going to Google. And when we asked her about this, she said, well, these are people who've actually used the medicine and they can tell me more about it. It's not just like going off to Google and finding some random person. And I mean, as a parent, I was somewhat horrified by this, but I mean, I think that that was, this notion that those answers are seen as more kind of directed at me. However, I think you lose this kind of the neutral kind of information that might be provided by a kind of more open search engine. But to answer your question about the difference in samples. So let's see. So I think that probably the items would look different about music. Might be something that we wouldn't see with the adults, but maybe would with an undergraduate sample. And I guess in terms of the extent to which they're kind of specifically doing that, I guess I couldn't say. I was curious sort of to tell you a step back from, I think you did a really nice job doing a lot of the kind of hard details about how to operationalize all this stuff. But if you could sort of simplify it a little bit and what would be the, if you woke up tomorrow and you were a social science, you know, deity and you could have the universe work however you wanted. What would be the experiment for this? I mean, what would you, what's the thing that what's the kind of big underlying thing that you're trying to see if Facebook does? Or is it something that Facebook does? Is it, I mean, what's the, it's a very fair question. So yeah, and I guess, yeah, and I think that's a great question. So I guess I, and I probably should have said this kind of at the beginning, that I don't necessarily consider, I mean, so Facebook is for me a particular technology that captures, you know, that has a lot of these interesting affordances. So for someone who has kind of studied social impacts of new media for many years, this is, you know, kind of a beautiful, you know, context in which to explore a lot of those questions because of a whole bunch of stuff I've talked about. So I guess in the grand scheme of things, I don't mean, I don't mean to, I mean, 10 years from now, I don't think I'll be studying Facebook. I mean, I think this is a particular instance in which, yeah, yeah, so, but I do think that it is reshaping the extent to which we can't, you know, the way in which we form and maintain social relationships and, you know, in all the benefits of communication that we know about, right? So I mean, this is all happening because we're communicating on the site, all of these kind of social capital benefits. So that's really kind of the true motivation is how is this affecting, you know, the communication, are there communication theories that we can apply? Can we extend, you know, theory through some of this work? I think in terms of if I woke up and I could have anything, yeah, I mean, I think one, you know, I think one missing piece, and this is kind of brought up in a lot of the questions, is that, you know, we're limited in a survey to how many questions we can ask, and also there's, you know, self-report and recall challenges, and so, for instance, being able to ask at a more granular level about the specific relationships or the specific behaviors, you know, are you liking something? I mean, that's just- And to know that they were telling you- Yeah, exactly. So I think the kind of behavioral data that Facebook itself has access to and have been, you know, working with and publishing, and, you know, I'm happy to see that work, yeah. So I think that, you know, they are in a position of being able to kind of answer some of those questions and because it's behavioral and because it's nuanced in a way that we can't get at. Yeah. I was wondering if you've worked at all with the same students repetitively so that you could see how they're sort of structured, their network changes over time, like creating panel data or something like that through repeated surveys of their perception of their social capital. Because I kept wondering how you differentiate, say, between these patterns being a reflection of already their social habits. That's a great question, kind of the chicken and the egg. Is it just that people who have more kind of social capital to begin with are going to then use the site more? So we do have a piece. It's Steinfeld Ellison et al. 2008 and it's on that pub's page right there. And it's in the Journal of Applied Developmental Psychology and we have three years of panel data and what we do in that piece is we look at social capital at time one, Facebook use at time one, social capital at time two, Facebook use at time two and we look at the strength of the relationship between Facebook use at time one and social capital at time two and social capital at time one and Facebook at time two and we show that there's indeed a stronger relationship between Facebook use at time one and social capital at time two than the other way around. So again, we can't necessarily make causal relationships and I apologize if I've kind of insinuated that in any way but I think that piece with the panel data does address at least some of those concerns about directionality, which I think is an excellent point. I think we have time for one more question. Even that you had, oh wait, Facebook's stalking the descent into craziness should be a journal title. Okay. But I won't pull out the notes. I just provided the title of the subject. My question, given that you have the social networks, why are you all so reporting for the total number of Facebook friends? Sure. And actually in the, so we, so because this was all self-report, we kind of wanted to keep that consistent. We do have the number of nodes and that's what we're using in the social network analysis and I think they're pretty highly correlated, although not exactly. Yeah, but great question. Okay. Thank you very much. This was really interesting.