Cummings, J., Butler, B., & Kraut, R. (2002). The quality of online social relationships. Communications of the ACM, 45(7), 103-108.

Key Points

Understanding the impact of the Internet on human social relationships requires two types of evidence. First, we need to know how computer-mediated communication affects the quality of particular social interactions and relationships. Are the online ones better, the same, or worse than those sustained by other means? Second, we need to know how computer- mediated communication affects one’s mix of social interactions and relationships. The impact of the Internet is likely to be very different if it supplements communication with existing friends and family, or if instead it substitutes for more traditional communication and social ties.

In terms of membership size and change, communication volume and structure, and participation levels, Internet listservs do not appear to be intimate social groups. . . . "our results imply that while interactivity can occur in these contexts, it is the exception, not the rule, when it occurs."

However, in one-to-one comparisons, an email message is not as useful as a phone call or a face-to-face meeting for developing and sustaining social relationships. Listservs are not as valuable as small groups for establishing a sense of identity and belonging and for gaining social support. Relationships sustained primarily over the Internet are not as close as those sustained by other means.

Critique

This all seems quite in keeping with personal experience. While electronic communication is fantastic for conveying information, it is difficult to convey all the subtlties of face-to-face contact through a mediated environment. You can't, for instance, tell that I'm a dog from this. If you could, you'd take me much less seriously.

CathyLu : I personally feel this article is very weak in its argument that Internet only supplements but not substitutes traditional social relationships. I know Xiaomu raised many questions on the McKenna2002 article, which stands in the opposite of this one. Nonetheless I think that one is actually a lot more convincing in terms of study sample selection, type of computer communication means chosen, and the completeness of data analysis.

First of all, for the Cummings2002 surveys, the samples the researchers chose are problematic in that coworkers do not typically hold strong personal relationships and college students should not be used as a sample pool in this study. We probably all agree that close cooperative relationship at work does not necessarily extend to closeness in personal life, especially not typically in a multinational bank where competition could be fierce among employees. From our personal experiences, college students are not among the groups in the society which are in great need of building social relationships: networking with mentors and potential employers maybe, but not necessarily lacking people to make friends with or establishing other types of relationships with. I also could not find the sample size (CaRichardson, n = 39 for this study, p 104 under the graph) of the study on the college student group in the paper and this greatly cut down my confidence in their data. Secondly, I am not sure that email group (listserv) is a typical tool for people to make social relationships online. Email group is one of the most passive ways of online communication and you can hardly say it is a type of person-to-person communication. There is no real time experience or strong interpersonal involvement that other online communication forms provide, such as chat rooms, forums, and messengers.

LaurieBuis: I agree with CathyLu in many of her above stated points in regard to this article. I feel particularly that the authors of this article had a weak argument that in many cases was expounded by operationalization problems. Like CathyLu, I don't know if I agree that a listserv is an appropriate type of community to be basing these findings on. Not only do listservs serve a potentially enormous audience (that may lead to breakdowns in communication as discussed in the previous class readings), but there is not as much of a shared interdependence of listserv community members as there are on other types of community. This article makes the point that there is little social support fostered online, but the Walters article directly contradicts this claim. Just thinking about all of the online support groups I've looked at, I tend to side with Walters on this one. (PaulResnick: Walther2002 is takling about social support groups, whereas this article is talking about listservs more generally. That might explain the difference.)

PaulResnick: I agree with the concern about inferences made with respect to listservs. First of all, selecting only open listservs where anyone could download the membership list introduces an important selection bias. Second, the only statistic they have about each listserv that seems relevant to whether they support social behavior is "interactivity". (If member participation is low, that might just mean that there's a small core group that are having a lot of social activity). Third, I'm not sure it matters what the typical listserv is like, only the most social listserv that a typical person is a member of. I am on many email lists; even if most are not social, or even are inactive, that wouldn't mean that I'm not getting much social from my lists as a whole.

I also think there's something odd about the basic question they've formulated in this paper. They are comparing a unit of f2f communication to a unit of electronic communication, and finding the latter wanting. But since there isn't any reason to believe that one is substituting for the other, it seems to me the question should be about the marginal impact of the electronic communication, not about a comparison between the two.

YongKim: In discussion of psychological closeness from frequency of communication for the nonInternert partner and Internet partner, it seems that the subjects actually use email to communicate with their “nonInternet partner” as nearly much as they do with their Internet partner (although there is statistical difference). Therefore, it may not be correct to interpret the result as the effect of computer-mediated communication (CMC) vs. non-CMC. I also think the partners that the subjects never physically met before should be excluded from the analysis because the initial physical contact is known to make a big difference in trust and emotional closeness, thus the result may have been confounded. In addition, it would be interesting to see how the use of instant messaging and web blog possibly change the result.

Connections

Oh, they're harder to make in electronic groups--I thought that was the whole point.

Class Discussion

CaRichardson: I agree with all of the points raised in the above discussion. The main take home message is that
the research in this article (and in the opposing McKenna2002 article) is weak and unconvincing. I think the major flaws in the paper can be classified into two categories:

1) Over-generalization: It is absurd to imply that this limited, weak sequence of studies of a particular mode of computer mediated communication conducted during the early days of the Internet (even 1999 is 6 years ago, before eBay, Amazon.com, yahoo groups, meetup etc became widely known and used) can support a conclusion like "Relationships sustained primarily over the Internet are not as close as those sustained by other means." (a quote from Cummings).

2) Error in causal attribution: The studies discussed in this paper are cross sectional, corelational studies. There is no doubt that many relationships are dynamic and not static. Relationships evolve and the strength of the relationship is a critical variable for predicting the path of evolution. In particular, it is possible that the stronger a relationship becomes the more likely it is that there will be some FTF communication. Thus, cross sectional studies are inherently flawed in that the baseline strength of the relationship likely results in profound selection bias. There are two possible causal pathways for the associations found in this paper. The authors imply a and I think b is likely to be a stronger effect.

a) Internet communication --> low relationship strength

b) low relationship strength --> Internet communication more likely than FTF

The only way to untangle the causality would be to conduct a longitudinal experiment, tracking relationships over time, and limiting mode of communication to either Internet or Face to Face in the two arms of the study. As the communication and interaction tools on the Internet are rapidly becoming more sophisticated, the ideal way to conduct this type of research would be to recruit and follow participants in annual cohorts over time. This both allows comparison of earlier Internet applications to more recent Internet applications and ideally it also allows more rapid analysis and dissemination of research results as the study does not have to be designed and implemented from scratch every few years.

Of course, this presumes that the correct question to ask is "Which mode of communication is better for sustaining relationships?" As PaulResnick mentioned above, this is probably not the most interesting question. The really interesting question is how does adding the Internet as a communication channel affect relationship formation and maintenance in a world in which Face-to-Face communication is frequently but not always an option.

One interesting study might be to assess the value of Internet communication in maintaining relationships between soldiers in Iraq and their families back home. Does access to the Internet decrease the divorce rate after long overseas tours?

Along these lines, internt communication is not normally a substitute for Ftf, but usually a complement. I think the question should be, how would current relationships be different in the absence of internet communication? [ErikWJ]