Article Summary 

This article offers a summary of the research on the subject of social comparison, the act of comparing one's self to others for the purposes of self-evaluation. In this way, we use others similar to ourselves as proxies, and their past and current opinions and behavior to interpret and predict our own inclinations and abilities.

Research has shown that in comparing our abilities with others, both the effort and attributes of our social proxies are important considerations, and in the absense of our knowledge of one, we tend to base our comparisons solely on the other. For instance, if one MSI student was hired for a desirable position, but it is not known how hard that student had to work to get it, that studen'ts peers might conclude that they would be equally likely to secure similar positions if they share the same general attributes of the student.

The authors suggest that there are three basic types of opinions for which we use social comparison to evaluation:

• Current preferences
• Future preferences
• Beliefs

With beliefs, we tend to choose proxies who are share similar attributes to ourselves, such as background, religion, politics or general world views. In fact, although we often base our beliefs on those of experts, our trust in expert opinion can be overriden when an expert does not have these attributes in common. For this reason, researchers have identified the role of similar expert as a highly important one in trend-setting and opinion leadership.

With  preferences, it  seems that similar past behavior is more important than attributes in proxy selection. This is the case with the collaborative filtering mechanism on Amazon.com: users are given recommendations based the buying patterns of users with similar behavior patterns.

(I could go on, but I'm hoping someone will take up the torch of summarizing the second half of the article...)

Social Comparison in the BaWer Community

I see many of the concepts introduced in this article at work in the Becoming a Webhead community, both in ability- and opinion-based social comparions. I imagine that BaWers self-evaluate through social comparison privately as well, but the type I've been able to observe is of a more overt, deliberate nature. The directors and moderators (i.e., the community leaders) position themselves as proxies for the rest of the community (particularly the newcomers), in a way of conveying a kind of "if I can do it, you can do it" attitude towards learning and adopting new technology for teaching. This seems to allay community member's fears about feeing ignorant or unskilled in technical matters, and create an open and supportive atmosphere for learning and informal mentoring.

This is a continuation of Erika's summary of Sul's article of social comparison. Basically, people compare themselves with others for three reasons. 

• self-evaluation - to gain a sense of whether performance is good enough (note: this tends to happen when performance standards are ambiguous or non-existent).

• self-enhancement - to feel better about oneself

• self-improvement - to provide the motivation to do better at a task.

All three of these motives can be seen as regulatory behavior. Self-evaluation tells us when it is time to stop working (e.g. when performance is good enough) self-enhancement helps restore homeostasis in mood and self-improvement may increase the drive to achieve a desired state.

Regardless of the motivation for social comparison, social comparisons necessarily influence both beliefs about performance and affective response towards performance.

Social comparisons can either be upward (against a superior target) or downward (against an inferior target).  Although intuitively it seems like upward comparisons should make people feel worse and downward comparisons this is not necessarily the case. The conclusions people reach depend on whether they contrast themselves with the target or assimilate the target. Assimilation effects occur when people conclude that the target is in some way a part of the current or future self. Contrast effects occur when the target is excluded from a representation of the current or future self. Thus, upward assimilation effects and downward contrast effects produce substantially the same result.

Social Comparison in eCommunities

The main issue for eCommunities seems to be what information people use to determine whether they contrast or assimilate themselves with a target. The vast majority of eCommunity users will have the potential to make both upward and downward comparisons, and will probably engage in both at different times.

• Length of membership in a community may be important. New members may assimilate towards members who have been around for longer. However, this variable will interact with group permeability. New members should be more likely to assimilate in groups that have few barriers to entry (because membership is attainable), whereas excessively high barriers to entry may result in contrast effects for people who believe that they have no hope of becoming a member.
• Similarly, people are likely to assimilate themselves towards targets who have the same "level" of membership (e.g. moderators with other moderators).
• The level of self-construal also seems to matter. Assimilation effects should be more common when the community sees itself as a "we" (perhaps through a membership requirement) while contrast effects should emerge when group identity is low.
• Perceived level of similarity also matters. People assimilate towards similar others even if the similarity is irrelevant (e.g. a shared birthday. I am not sure how this plays out in an environment in which there are relatively few social cues though. On one hand it could be argued that everyone is assumed to be more similar on the net (because there is less info to distinguish them). On the other hand, this could also make them seem more alien.

It is natural for humans to compare themselves to each other; this reflective process may occur so that we can evaluate our current or future standing, or so we can enhance our feelings about ourselves.  Both reasonings are discussed in Suls' article:

Evaluation
Comparison for the purposes of evaluation occurs when one reflects on questions such as:

• Am I capable of doing this?
• Will I like this?
• Is this correct?

In the case of the first question, we use a proxy as a comparison point from which we determine our own capability. The proxy must be someone with similar traits to our own, thus proving capability. For instance, one feels capable of completing a task because someone with similar skills and physical abilities has also proven capable of completing the task.

In the case of the latter two questions, we use a triadic model of similarity, expertise, and past experience to evaluate the answers for ourselves. We determine our own beliefs, opinions and assumptions based on others, because of our need to look up to those who may know better than us, such as experts or those with a shared history.

Enhancement
We compare to enhance our own feelings of self, to feel like we are among "good people" or that we are in a better situation than others. Comparisons may occur in an upward (looking up to those superior to us) or downward (looking down to those inferior to us). We may choose one direction on the other based on our current situation and our motivation for comparing. For instance, if I am in a bad situation (perhaps I am sick) and I want to feel like I am not alone and that there is hope, I am seeking an assimilative outcome that may lead me to compare upward with others to whom I feel a level of psychological closeness and a feeling of cohesiveness as a group. On the other hand, if I want to feel like my illness isn't quite as severe as it could be, I am seeking a contrastive outcome in which I look downward. However, in a moment of hopelessness I could also find myself comparing upward to those who are well and feel worse about my own situation.

Although not directly stated, I think that these ideas apply to online communities in the sense of understanding how groups of disparate people can become a cohesive group. Conversations allow members to evaluate their own capabilities and enhance their own self-image through comparison with  each other, if the community is designed to allow these interactions. Perhaps these differing types of comparisons indicate that online communities should be designed with different areas for each type of comparison-- for a cancer support site, evaluative comparison could be facilitated through boards in which topics revolve around treatment options and opinions of various approaches to treatment, while enhancement could be faciliated through boards in which topics revolve around how the members are doing with their chosen treatment plans.

For a site such as Geocaching, I can certainly see how evaluative comparison comes into play in user profiles. When I look at one's profile I can see the types of caches he has visited, and make assumptions about own own capabilities. If I can see what kind of caches the member likes to explore, his/her physical capabilities (can they handle rough terrain, find the tough-to-see places, etc) then I can determine whether I can handle the types of caches this member has visited and enjoyed. I may also enhance my own self image by comparing my successes to geocachers who are superior or inferior to my skills-- looking up to learn how to get better at locating difficult caches, or looking down to see how much progress I've made since I was a novice cacher.  

So much social science for one week ☺

Notes

People compare themselves to others intentionally or unintentionally all the time. This phenomenon, called social comparison is one source of self-evaluative information among others such as comparison with past selves. The article discusses two motivations for social comparison: self-evaluation and self-enhancement.

Two models are proposed for explaining self-evaluation. The first one is proxy model – social comparison of ability. According to this model, we compare ourselves to people similar to us that are called proxies. People expect to perform at the level of proxy on a new task if one’s history of performance on some initial relevant task is similar to the proxy’s history of performance on that task and the proxy is known to have exerted maximal effort on that preliminary task. The second model for self-evaluation is the triadic model (opinion formation) relying on current preferences, beliefs, and future preferences.

Self-enhancement part of the article discusses the upward and downward comparisons in terms of assimilation and contrast and it concludes that self-evaluations are not intrinsically linked to the direction of the comparison. Comparison can produce positive and negative contrastive and assimilative effects.  

Critique

The proxy model makes perfect sense to me; it’s like a controlled experiment in that you try to keep as many variables constant as you can. For example, you compare your performance in graduate school to people whom you went to college with who have similar backgrounds as yourself.

For self-enhancement, I find it natural for people to do upward comparisons since it is generally good to shoot for higher targets rather than lower ones. I think the results of a downward or upward comparison depends greatly on the mood a person is in and his or her experience on the task.

Application

Relative standing is not as salient in 43T as it is in other online communities such as the activity-based ones. But the number of completed goals, entries and cheers received might be elements in social comparison. Additionally whether a user completed a particular goal or not might also be important for comparison. For example, a person who completed the goal of quitting smoking might make down comparisons and feel himself better as a result of the comparison.

Jesse's point about self-evaluation is well taken. This gets at Ling et al's idea of multiple, overlapping social theories. We'll evaluate against someone else as described in Suls ASSUMING we don't have a pre-existing performance standard. From Jesse's post, it sounds like a clear standard will trump comparison evaluating... (Is this correct?) This doesn't mean we won't still compare for other purposes.

I'm thinking about applications, about leaderboards, about tapping into automatic comparison behaviors. I'm thinking about Nika's annoying animated Family Guy icon hovering above me in the top rated users listing. (Curse you, Quagmire.) So I guess if I'm feeling threatened by Nika, I may start comparing myself favorably against those who are worse off (Brian and Erika).

However, Collins' might suggest something different: I want to perceive myself as having positive qualities, so I take my inclusion in the top rated posters as proof that I'm a good student. My inclusion in the top rated posters becomes proof to me that I'm one of the good ones, even though my placement there may be equally due to luck.

Given that being bested by Nika makes me want to post more, I think there could be some decent applications to the wiki world, if similar leaderboards were set up. But I see this in the same class as CEM tricks to motivate contributions. I have a feeling that if I were more awake, there might be larger applications to online communities. (esp. intrigued by the idea of mutability and whether design choices can lead to certain learned social behaviors at a deeper level than contributing simple things like votes or ratings. if community members are mutable, can the community be designed in such a way to facilitate more high-impact learning?) Now to sleep.

The proxy model sumarized in this short paper suggests that we take a combination of related attributes and expended effort into account when gauging ourselves against others. There are interesting informational issues contained in such a model:

** Visible vs. invisible related attributes. Some aspects of others are readily available for us to "control" over when making comparisons. In the gym, i think i should curl at least as well as someone in visibly similar physical condition, and in the long run push as mush weight as anyone my same height and weight. In many other situations, we may look to appearance or SES to infer the related attributes we would use in weighting the other person as a proxy. The lack of robustness in such attributes, or our inablity to infer from them the underlying attributes we need to make a meaningful comparison may be a problem in some online communities. In MIXI, I personally have trouble relating my knowledge of a topic to others - they may be experts, be studying the topic in school, etc. and we have no "cues" besides their posts (and profile information)

** Guaging other's effort. Unlike certain related attributes, effort is not easily visible. Especially in the online world, my weighting of others' efforts is always heavily influenced by a distrust of other's announced effort. In the real world, we have evolved social conventions that lead us to distort effort expended up and down out of consideration of others. ("Hmm... really, it was mostly luck") Some of these conventions may transfer to the online world, some may not.

Suls' comment that people are quite capable of drawing assimilative evaluations from upward comparisons reminds me of social histories I've read about early modern Europe, in which nouveau riche merchants buy, or otherwise force, their way into the ranks of the traditional aristocracy, and, at the same time, look on the lower classes with stark horror, for fear that they might instead lose their privileges and fall down to the bottom. (That dynamic also seems relevant to a discussion of our own 'classless' society, but I'll opt against starting that off-topic flame war.)

In terms of Milliways, I'm sure at least some individual players use some private, more-or-less subconscious form of the Proxy Model, when they compare other players' theads with their own. Certainly there is no shortage of players who mourn their own lack of talent (sometimes just out of false modesty, perhaps, but surely sometimes out of sincere self-doubt as well).

Examples of public downward comparisons are harder to find. People might make comparisons at their own expense, but (fortunately) it's not socially acceptable in Milliways to praise your own talent at the expense of others. (Self-praise is possible and, within sharp limits, acceptable, but not if it's done in a relative way.)

Suls' discussion of the three types of opinion is interesting (particularly for his point that only a 'similar expert' will do), but I'm not sure I would be able to drill down to that level of player interaction without moving farther into players' personal lives than I would prefer for this study. Questions of self-evaluation, especially when they may include feelings of superiority or inferiority, should probably be left to experienced researchers.

I could see that the homogeneity of the culture may blur the difference of the predicted performance of a person relative to effort ambiguous and effort maximum of the proxy. What it means is that people with more homogenous culture may tend to less believe efforts shown if it is deviant from what they believe is “supposed to be” according to the attributes. Thus the length of the black bar may be less different with the white bar than shown in figure 1, in homogenous cultures.

Korean society is a very homogeneous society due to many reasons: our closed policies about foreign trades during the 17 and 18^th centuries (we basically thought China was the best and no other countries were trustworthy), unified race, leftovers from what we were forced to do by Japan in the first half of 1900’s (unified and organized groups are always easy to rule), religious / philosophical belief, etc etc.

I may have to press a little bit about my preconception about the Korean society here. People with certain attributes behave certain ways, or at least are expected to behave certain ways. (although we are getting better) Thus we are not used to coping with a person who is difficult to anticipate based on the attributes. The culture is nicely fitting in with the nice symmetrical bell shaped curve, and anything that is outside the curve is supposed to be deviant, and somehow will be assessed by the standards that are inside the bell curve. The standards in the bell curve are the assessments with the attributes in the proxy model, and the proxy is always a sample of the population that is ideally graphed as the bell curve. Thus again the effort maximum may not win the belief toward perceptions around attributes, if the effort maximum contradicts what the proxy should do according to the attributes.

Erika and Jesse have already done an excellent job summarizing this article, so I'll sum it up in one sentence for my own benefit: people frequently compare themselves to others in order to evaluate their current state or make some prediction about their aptitude for a different activity.

Application
While much of the comparison occurs out of my view offline (or privately, internally), I feel pretty confident that comparison with others is a common activity on Cool Running. The upward and downward comparisons Suls describes are readily available in this community. In fact, the criteria by which I could compare myself are made readily available in the thread titles/topics. The Couch-to-5K program, the 30-somethings threads, the "pre-dawn patrol" runners, all offer me easy access to upward or downward comparisons described by Suls.

The "self-improvement motive" Suls described (he cites Wood (1989)) provides a very plausible description of the types of and reasons for comparison in Cool Running. Motivation can be drawn from comparison to proxies, and can be either positive or negative in Cool Running.