Computers in Human Behavior 27 (2011) 1223–1230

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier .com/locate /comphumbeh

The effect of uniform virtual appearance on conformity intention: Socialidentity model of deindividuation effects and optimal distinctiveness theory

Junghyun Kim a,⇑, Hee Sun Park b

a School of Communication Studies, Kent State University, PO Box 5190, Kent, OH 44242, United Statesb Department of Communication, 481 Communication Arts & Sciences Building, Michigan State University, East Lansing, MI 48824

a r t i c l e i n f o

Article history:Available online 28 January 2011

Keywords:Virtual appearanceVirtual group identityComputer-mediated communicationUniquenessSimilarity

0747-5632/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.chb.2011.01.002

⇑ Corresponding author. Tel.: +1 330 672 0285; faxE-mail addresses: jkim23@kent.edu (J. Kim), heesu

a b s t r a c t

This study integrates social identity model of deindividuation effects (SIDE) and optimal distinctivenesstheory (ODT) in investigating the effect of uniform virtual appearance on individuals’ willingness to con-form to a majority opinion in computer-mediated groups. SIDE posits that sharing the same visual cuecan promote group identification process and eventually induce stronger conformity. Meanwhile, ODTindicates that too much visual similarity rather concerns individuals about their deprived uniqueness,so they would be reluctant to conform to a majority opinion as a way to restore their uniqueness. Thisstudy concurs with previous research based on SIDE by showing that group identification induced by uni-form appearance increases conformity intention. It also showed that perceived deindividuation, anothervariable that is induced by a high level of visual similarity relative to others, decreases conformity inten-tion. As a result, the current study shows that the effect of virtual uniform appearance on conformityintention is inconsistently mediated by group identification and perceived deindividuation.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction mote a specific political opinion or create a congenial group for

Much research on social influence has focused on conformityrather than nonconformity, which is contrary to the value of indi-viduals over groups manifested by Western societies (Hornsey &Jetten, 2004). This can be explained by one of the suppositions ofsocial identity theory (Tajfel, 1981): individuals naturally try tobe acceptable to others in the same group (Allen, 1965; Santee &Maslach, 1982). Even if individuals are grouped on a random basis,they still adopt norms and values of the group by not expressingtheir deviant opinions for fear of social sanction (Deutsch & Gerard,1955). Postmes, Spears, and Lea (1999) have applied this perspec-tive to computer-mediated communication (CMC) contexts, suchas virtual communities, and found that reduced social cues inCMC can accentuate group identification and adoption of groupnorms more than in face-to-face interaction.

Intensified group identification might or might not be wel-comed depending on virtual communities’ diverse goals or mo-tives. If a goal of a virtual group is to create a space forunrestricted discussions, ‘‘conformity to a majority opinion’’ pres-sure coming from strong group identification might prevent indi-viduals from expressing themselves freely (Hiltz & Turoff, 1978;Janis, 1982; Nunamaker, Briggs, Mittleman, Vogel, & Balthazard,1997). On the other hand, if a goal of a virtual community is to pro-

ll rights reserved.

: +1 330 672 3510.n@msu.edu (H.S. Park).

individuals of diverse demographic backgrounds, strong groupidentification and conformity to a majority opinion should beencouraged.

Pertaining to virtual communities promoting congeniality, ahigh level of visual similarity among members is known to beone of the ways to intensify their identification with other mem-bers, such as being represented by the same visual cue (Kim,2009; Lee, 2004). Yet, can a high level of visual similarity (i.e., uni-form appearance) also make individuals conform to a majorityopinion more than a low level of visual similarity (i.e., different vi-sual cues)? Research investigating the effect of uniform appearanceon conformity to a majority opinion in offline settings has pro-duced mixed findings. A group of studies has shown a positiveassociation between similarity in appearance and conformity to amajority opinion (Terry & Hogg, 2001; Turner, 1984). Sharing thesame visual cue with others can make one feel that he/she belongsto a majority group, and research on social influence has shownthat majority sources are highly influential (Imhoff & Erb, 2009;Mackie, 1987; Wood, Lundgren, Ouellette, Busceme, & Blackstone,1994). Following the logic of social influence research, one wouldbe likely to agree with a group of others who look like him/hermore than with a group of individuals represented by idiosyncraticvisual cues, because conforming to the opinion of the former givesa more sense of belonging compared to the latter (Asch, 1956).

On the other hand, another group of studies finds the opposite– individuals who look highly similar to others are hesitant to con-form to a majority opinion (Codol, 1984; Cooper & Jones, 1969;

1224 J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230

Fromkin, 1970). Being too similar to others can make individualsfeel deprived of their distinctiveness as unique individuals, whichmight hinder them from thinking or saying what others do (Imhoff& Erb, 2009). Perceiving diminished distinctiveness, one might bemotivated to do something to compensate his/her reduced unique-ness such as voicing a different opinion from others’.

Instead of arguing which one makes more sense than the other,we presently propose that the two aforementioned effects of highvisual similarity (i.e., uniform appearance) on conformity to amajority opinion coexist in computer-mediated group settings.Two theoretical frameworks emerge that can explain these twotypes of effects of high visual similarity: social identity model ofdeindividuation effects (SIDE) (Lea & Spears, 1991) and optimaldistinctiveness theory (ODT) (Brewer, 1991). Following the logicof SIDE, uniform appearance can expedite group identification pro-cess by making group members pay less attention to each other’sidiosyncratic qualities while making them focus on the shared vi-sual cue (Kim, 2009; Lee, 2004). This, in turn, leads to members’attachment to their group and conformity to a majority opinionpromoted by the group members (Postmes et al., 1999; Reicher,1987). On the other hand, ODT posits that individuals want to bemoderately similar or unique in any given situations, so too muchsimilarity relative to others would work against their willingnessto conform to a majority opinion (Brewer, 1991). In short, the cur-rent study investigates whether these two theoretical frameworkscan explain the disparate effects of high visual similarity (i.e., uni-form appearance) on computer-mediated group members’ willing-ness to conform to a majority opinion.

1.1. Social identity model of deindividuation effects (SIDE)

SIDE posits that the paucity of individuating cues (e.g. facialexpressions, gestures, or eye contacts) in computer-mediatedinteractions can make group membership (i.e., social identity) sali-ent and induce adoption of group norms even more compared toface-to-face interactions (Lea & Spears, 1991; Postmes, Spears, &Lea, 1998; Postmes et al., 1999). Such individuating cue-deprivedCMC renders extremely limited information with which individu-als make sense of each other. When individuals have little informa-tion of their anonymous conversation partners of zero interactionhistory, any given cues in CMC can be magnified as one of the mostimportant characteristics of one another. For instance, if socio-demographic information such as gender or ethnicity is availablewithout any other individuating information, individuals mightperceive and treat each other as a prototypical member of certaingender or ethnic groups (Postmes et al., 1998). Such reduced focuson individual differences (i.e., personal identity) and increasedattention to group membership (i.e., social identity) is called‘‘depersonalization’’ (Postmes et al., 1999).

In that sense, if one is represented by a visual image withoutany other personal information available, it can serve as the onlyindicator to which others relate. When the other’s only availablevisual cue happens to be the same as mine, such uniform appear-ance serves as a basis for psychological group formation as a cueto a common social category (Turner, 1984). Thus, sharing thesame image can intensify group identification process, even if theimage does not contain any socio-categorical cues (e.g., gender orethnicity). Uniform appearance has been used to induce a high le-vel of depersonalization in both CMC research (Kim, 2009; Lee,2004) and face-to-face context-based research (Maslach, 1974;Zimbardo, 1969). Functioning as a cue to a common group mem-bership, uniform appearance fosters identification with othermembers while diluting individual differences. On the contrary, ifindividuals are represented by cues that are different from others’,they focus on visual differences without any cue for similarity.Different visual cues interfere with group identification process

by highlighting individual differences among group members(Lee & Nass, 2002). In this condition, personal identity would beemphasized more than social identity. As suggested by SIDE, themore individuals identify with their group and group members,the more they are willing to conform to a majority opinion pro-moted by other group members (Terry & Hogg, 2001).

1.2. Optimal distinctiveness theory (ODT)

Alternatively, some researchers have brought up a possibilitythat too much similarity can deter individuals’ willingness to con-form to a majority opinion. This supposition was supported by ear-lier experiments by Fromkin (1970) as well as by Cooper and Jones(1969) showing that individuals who were made look highly sim-ilar to one another tried to promote their uniqueness either by giv-ing unusual answers or by changing their opinions into somethingdifferent from those of others. In CMC contexts, Lee (2004) foundthat uniform appearance elicited less conformity compared to dif-ferent visual representations. As a possible account for this result,Lee (2004) suggested optimal distinctiveness theory (ODT)(Brewer, 1991, 1993).

ODT posits that individuals always look for equilibrium be-tween extreme similarity to and extreme distinctiveness from oth-ers. As a way to explain this propensity, ODT relies on anassumption that individuals have two competing motivations innature – a motivation to belong to others (Baumeister & Leary,1995; Maslow, 1968) and a motivation to be different from others(Snyder & Fromkin, 1980). Because of such conflicting nature ofthese two motivations, individuals cannot gratify one motivationmuch without sacrificing the other. Therefore, the best way of sat-isfying these two competing motivations at the same time is tomaintain both of them at moderate levels.

Following this logic, individuals would not feel comfortable ifthey perceive that they are too deindividuated, since their motiva-tion to be unique is forsaken. Thus, ODT presumes that individualswant to see themselves moderately unique and do not value a highdegree of similarity relative to others (Snyder & Fromkin, 1977).This proposition leads to a hidden assumption of ODT and unique-ness theory – individuals have a capability to sense how deindivid-uated they are in any given situations and are conscious of externalfactors lessening their individuality. That is, a higher level of visualsimilarity shared among group members would make them per-ceive that their uniqueness as distinctive individuals diminishedgreatly compared to a lower level of visual similarity.

The fact that individuals are conscious of maintaining a moder-ate level of uniqueness in any given conditions can help us explainan unexplored process connecting uniform appearance and indi-viduals’ reluctance to conform to a majority opinion. Conformity,including conforming to social norms, pleasing others, getting oth-ers’ approval and avoiding rejection, can be a reflection of one’s de-sire to be similar to others. In a state of deprived uniqueness,however, individuals would avoid further deprivation by agreeingwith a high consensus opinion and rather agree with a minorityposition as a way to reduce the feeling of impoverished uniqueness(Imhoff & Erb, 2009). In that sense, voicing different opinions fromthe majority opinion can be used as one of the behavioral strategiesfor regaining some distinctiveness from others (Baumeister, 1991;Snyder & Fromkin, 1977). Nonconformity can be seen as a behaviorreflecting one’s desire to be different, because deviating from amajority opinion or a group norm promotes independence andautonomy of self (Simonson & Nowlis, 2000). Thus, if individualsfeel that their uniqueness is deprived from assigned (or even un-wanted) similarity relative to others, they would feel reluctant toconform to a majority opinion, but rather bring up deviant opin-ions to restore their uniqueness.

J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230 1225

1.3. Inconsistent mediation

Integrating the aforementioned two theoretical frameworks –SIDE and ODT – we presently propose that a high level of visualsimilarity induced by uniform appearance increases both individu-als’ identification with their group members and perceived deindi-viduation at the same time. Perceived deindividuation is derivedfrom awareness that their individual differences are oppressedby assigned uniform appearance among group members. Thesetwo have inconsistent associations with conformity to a majorityopinion: while group identification increases individuals’ willing-ness to conform to a majority opinion raised by other members,perceived deindividuation deters individuals’ willingness to con-form to it. Overall, SIDE and ODT provide competing predictionson the effect of uniform appearance on individuals’ willingnessto conform to a majority opinion through two different mediators– identification with group members and perceived deindividua-tion. In other words, the indirect effect of uniform appearance onconformity intention through group identification has an oppositesign (positive) to that of the indirect effect through perceived dein-dividuation (negative). When there is at least one indirect effectwith a different sign from other indirect effects or the direct effectbetween an independent and a dependent variable, it is calledinconsistent mediation (Davis, 1985; MacKinnon, Fairchild, & Fritz,2007; MacKinnon, Krull, & Lockwood, 2000). Thus, the currentstudy hypothesizes that the effect of uniform appearance on will-ingness to conform to a majority opinion is inconsistently medi-ated by group identification and perceived deindividuation (Fig. 1).

H1. Group identification and perceived deindividuation will incon-sistently mediate the effect of uniform appearance on individuals’willingness to conform to a majority opinion.

2. Method

2.1. Participants

A total of 345 undergraduate students (61.6% women) rangingfrom freshmen to seniors from a variety of majors enrolled in anadvanced Communication course from a large Midwestern univer-sity in the United States of America voluntarily participated in thestudy. Age of participants ranged from 18 to 39 years (M = 19.63,SD = 1.75). Participants were required to use the Internet and tendto use a variety of Internet functions.

Fig. 1. The hypothesized incon

2.2. Procedure

The experiment was introduced as a research project on a com-puter-mediated group discussion to participants. All participantswere directed to visit an online experiment website and to partic-ipate in a discussion resolving a controversial issue with other dis-cussants. Once participants logged into the online discussionwebsite, they were randomly assigned to one of the two condi-tions: the uniform-appearance condition where all participantsshared the same visual cue, or the different-appearance conditionwhere each member had a unique visual cue. Of the 345 partici-pants, 173 were assigned to the uniform-appearance condition,and 172 were assigned to the different-appearance condition. Per-taining to visual cue designs, two computer graphic designers wereasked to design images that would not contain any negative cul-tural or social connotations. This process resulted in a set of imagesconsisting of animals and cartoon characters. A focus group of fivestudents was asked to evaluate these images, and they did notindicate any negative reaction toward those images.

Even though participants were told that they were to partici-pate in a real-time online discussion with four other discussants,those four discussants were programmed ones. This deceptionwas necessary to make sure that all participants were exposed tothe same transcript across the two different types of appearanceconditions and that they focused on the discussion topic withoutbeing distracted by talking about different subjects. Moreover,using the programmed discussion manuscript made it possible tocreate the same conformity pressure for every participant stem-ming from all four other discussants’ agreeing on one opinion. Inorder to increase the reality of the programmed conversations,all the programmed conversation texts were taken from two focusgroup sessions that had taken place before the online experiment.

Before participants moved to the real-time discussion page,they were asked to choose one image, which would represent themduring the discussion, out of four. In the uniform-appearance con-dition, whatever images were chosen by participants, the otherfour discussants were programmed to be represented by the sameimage (Fig. 2). Meanwhile, four other discussants were pro-grammed to have different images from one another’s in the differ-ent-appearance condition no matter what images were chosen byparticipants (Fig. 3). The instruction clearly noted that every dis-cussant was asked to choose his/her image freely, so all partici-pants believed that sharing the same image as others’ or beingrepresented by different images from others’ were the results ofother discussants’ voluntary choices. The discussion instruction

sistent mediation model.

Fig. 2. Snapshot of the uniform-appearance condition.

Fig. 3. Snapshot of the different-appearance condition.

1226 J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230

also clearly noted that discussants were from diverse ethnicity andage groups, so participants could not assume whether other discus-sants were students or not, from the same gender group or not, orfrom the same ethnicity group or not. We intentionally left otherdiscussants’ demographic information unknown to participants,so that participants could not pick up any other nuanced similari-ties or differences than other discussants’ visual cues. Seeing theway they were visually represented, participants were asked toindicate how much they felt their uniqueness suppressed (i.e., per-ceived deindividuation), and how much they could identify withother discussants as members of the same group (i.e., groupidentification).

After moving into the discussion page, participants were askedto read a hypothetical scenario (Lee, 2004) in which a person faceda dilemma situation and had to choose one option between goingto a music school or a medical school (Appendix B). About the timeparticipants were to be finished with reading the scenario (i.e.,about one minute), four programmed discussants presented theiropinions one after another in order to give an impression that eachdiscussant expressed his/her opinion after reading his/her priordiscussant’s opinion. Four other discussants were programmed to

share the same opinion, which implied the existence of a consen-sus norm or a majority opinion within the discussion group. Partic-ipants were asked to choose how much they agreed with thedominant opinion shared by other discussants. Then, participantswere told that the first discussion would be followed by more dis-cussion sessions. Given that all the other discussants were to agreeon one opinion in those discussions as in the first one, participantswere asked whether they would be willing to conform to theirgroup members in coming discussions (i.e., conformity intention).After the discussion was done, participants were asked to fill out aseries of questionnaires.

2.3. Measures

2.3.1. Perceived deindividuationBecause there has not been any previous research that specifi-

cally measured individuals’ perception on how much their unique-ness is suppressed by being extremely similar to others, four newlycreated items were used to measure perceived deindividuation(Cronbach’s a = .90, M = 3.78, SD = 1.52). Items consisted of suchstatements as ‘‘I think members in this group were not represented

J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230 1227

as unique individuals in this group,’’ and ‘‘I think I was not consid-ered as a distinctive individual in this group.’’.

2.3.2. Group identificationGroup identification was measured by three items (Cronbach’s

a = .88, M = 4.58, SD = 1.12) used in Lee’s (2004) study, which wereadapted from Cheney’s (1983) organizational identification scale.

2.3.3. Conformity intentionConformity intention was measured by four items concerning

participants’ intentions to agree with any dominant opinion pro-moted by four other discussants in upcoming discussions (Cron-bach’s a = .88, M = 4.67, SD = 1.19). Example items were ‘‘I amwilling to agree with them,’’ and ‘‘I am willing to follow their opin-ion.’’ Conformity intention was used as the dependent variable forthe data analysis in this study instead of actual conformity deci-sion, since generalizing the conformity decision regarding one sce-nario to other decision-making cases can threaten external validityof the study.

A total of 11 items were used to measure these three variableson 7-point Likert-type scales (1 = Strongly Disagree, 7 = StronglyAgree). Confirmatory factor analysis (CFA) was conducted to checkif these 11 items were indicators of three separate factors. The CFAmodel was tested using AMOS 6.0 structural equation modeling(SEM) program (Arbuckle, 2005). According to Hu and Bentler’srecommendations on model fit indexes (Hu & Bentler, 1999), .06or less means a good fit for root mean squared error of approxima-tion (RMSEA), .95 or more for incremental fit index (IFI), and .95 ormore for comparative fit index (CFI). Pertaining to RMSEA, while avalue below .06 means a good fit of the model with the data, valuesbetween .08 and .10 indicate a mediocre fit and values over .10 apoor fit (Hu & Bentler, 1999). The CFA result showed a good fit,RMSEA = .065, IFI = .980, CFI = .980.

3. Results

3.1. Main effects of manipulation

There was a significant main effect of the types of visual appear-ance conditions on two mediating factors: perceived deindividua-tion and group identification. Participants in the uniform-appearance condition perceived a higher level of deindividuation(M = 4.74, SD = 1.58) compared to those in the different-appear-ance condition (M = 2.93, SD = 1.01), F(1, 343) = 81.84, p < .001. Vi-sual appearance manipulation also showed a significant main

Fig. 4. The standardized coefficient values of the hypothes

effect on group identification. Participants in the uniform-appear-ance condition showed a higher level of group identification(M = 4.96, SD = 1.13) compared to those in the different-appear-ance condition (M = 4.07, SD = 1.08), F(1, 343) = 32.67, p < .001.On the other hand, participants in the uniform-appearance condi-tion did not show a significantly different level of conformityintention (M = 4.66, SD = 1.32) compared to those in the differ-ent-appearance condition (M = 4.60, SD = 1.13), F(1, 343) = .83,p = .44.

3.2. Inconsistent mediation model testing

In order to test the overall fit of the hypothesized inconsistentmediation model in Fig. 1, structural equation modeling (SEM)was used. The hypothesized path model showed a good fit withthe data: v2 (1, N = 345) = .498, p = .480, RMSEA = .00, IFI = 1.000,CFI = 1.000. Uniform appearance increased group identification(b = .37, p < .001) as well as perceived deindividuation (b = .56,p < .001) more than different appearances. While group identifica-tion increased conformity intention (b = .24, p < .001), perceiveddeindividuation decreased conformity intention (b = �.24,p < .001). These results are consistent with what SIDE and ODT pre-dict (Fig. 4).

AMOS does not allow us to test the significance of indirect ef-fects nor to contrast different indirect effects in their strengths.Thus, the current study chose a multiple mediation analysis(Preacher & Hayes, 2008), an analytic strategy appropriate for test-ing hypotheses with multiple mediators, to test Hypothesis 1 andResearch Question 1. This method was chosen based on the factthat the current study’s hypothesized model might not make a caseof the mediation test if following Baron and Kenny’s criteria formediation. According to Baron and Kenny (1986), a significantmain effect of an independent variable on a dependent variableis one of the prerequisite criteria for mediation. Following this lo-gic, the current study’s indirect effects through perceived deindi-viduation and group identification cannot be tested, since therewas no significant main effect of the differentiated levels of visualsimilarity on conformity intention (b = �.065, p = .624). However,the possibility that mediation can exist even without a significantmain effect has been recognized by a group of researchers, such asthe case of inconsistent mediation (Collins, Graham, & Flaherty,1998; Kenny, Kashy, & Bolger, 1998; Preacher & Hayes, 2004;Shrout & Bolger, 2002). To detect an inconsistent mediation with-out any significant main effect, a multiple mediation model can beuseful, because it can test more than one mediator at the same

ized inconsistent mediation model analysis, ��p < .01.

1228 J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230

time. As suggested by Preacher and Hayes (2008), specifying andtesting a multiple mediation model might be better than testingseparate simple mediation models when there are more than onemediator. Including several mediators in one model increases thepracticality of the hypothesized model by testing to what extentone factor mediates the impact of an independent variable on adependent variable when other mediators are present. Multiplemediation analysis is also useful for theory comparison (i.e., SIDEvs. ODT) by directly contrasting the magnitudes of the same-signedindirect effects associated with different theoretical perspectives.

The aforementioned hypothesized model was tested with SPSSmacro program, which was developed and updated by Preacherand Hayes (2008). The indirect effect of uniform appearance onconformity intention through group identification was signifi-cantly positive (b = .22, Z = 3.75, p < .001). Meanwhile, the indirecteffect of uniform appearance on conformity intention through per-ceived deindividuation was significantly negative (b = �.33,Z = �3.68, p < .001). Such opposite signs of indirect effects supporta case of an inconsistent mediation in the data obtained for thecurrent study.

4. Discussion

In this study, we propose that high visual similarity among indi-viduals (i.e., uniform virtual appearance) would have two possibleeffects on willingness to conform to a majority opinion in com-puter-mediated group settings. This study concurs with previousresearch based on SIDE (Lee, 2004) by showing that group identifi-cation induced by uniform appearance increases conformity inten-tion. This study also showed that perceived deindividuation,another variable that is induced by a high level of visual similarityrelative to others, decreases conformity intention. The existence ofthese two mediators might provide some explanations for theinconsistency in the association between uniform appearanceand conformity to a majority opinion in previous research.

The current study has some theoretical implications for explain-ing the impact of a high level of visual similarity on conformity ordissent in computer-mediated groups by showing that ODT, inaddition to SIDE, can be used as another theoretical framework.Both SIDE and ODT attend to the effect of intensified similarity rel-ative to others (i.e., deprived individuality) on individuals ingroups, but focus on different reactions of individuals. While SIDEposits that increased similarity makes individuals identify easilywith others and pay less attention to individual differences, ODTsuggests that the same attribute rather makes them focus on theirimpoverished uniqueness as distinctive individuals. The currentstudy showed that a high level of visual similarity induced by uni-form appearance makes individuals go through disparate processesat the same time: it intensifies group identification process, but atthe same time, makes individuals feel their uniqueness diminished.Subsequently, group identification is positively associated withconformity intention, while perceived deindividuation makes indi-viduals reluctant to agree with others.

While previous studies rely solely on social identity perspectivesuch as SIDE, or uniqueness theory perspective (i.e., ODT), the cur-rent study uses both frameworks in explaining individuals’ behav-iors in computer-mediated groups. This study predicts thatindividuals experience two simultaneous, but opposing processeswhen they decide whether they would conform to a consensusnorm in their groups or not (Davis, 1985; MacKinnon et al., 2000,2007). Such inconsistent processes negates each other’s effect onconformity intention, and this might provide an explanation forwhy uniform appearance did not have a significant effect on con-formity intention in the current study (M = 4.66) compared to thedifferent-appearance condition (M = 4.59), F = .834, p = .435(MacKinnon et al., 2000; Shrout & Bolger, 2002).

This indicates that too much similarity (i.e., looking exactly likeothers) might concern individuals about their deprived uniquenessas much as make them perceive similarity as a basis for a commongroup membership. This explains why some individuals lookingexactly like one another tried to voice different opinions from oth-ers’ in both face-to-face contexts (Codol, 1984; Cooper & Jones,1969; Fromkin, 1970) and CMC contexts (Kim, 2009; Lee, 2004).Still, these results lead to more questions: how much similarityis too much similarity that makes individuals feel reluctant to con-form to a dominant opinion of others? Future studies should inves-tigate how diverse levels of visual similarity relative to othersaffect individuals differently in computer-mediated groups, andwhere is the equilibrium between similarity and uniqueness thatmakes individuals feel most comfortable.

In addition to the aforementioned theoretical implications, thisstudy also has some practical implications especially for onlinecommunities or network games that use virtual self-representa-tions. Following the logic of SIDE, wearing the same visual cue asothers can intensify individuals’ identification with others incyberspace. In that sense, online game players or virtual commu-nity members can identify with one another beyond their national,cultural, ethnic, or even gender differences by sharing uniformappearance. This is in line with the early Internet researchers’ opti-mistic prediction that using the Internet can enable individuals toovercome socio-categorical boundaries and increase equality insociety by enhancing their opportunities to participate in socialactivities (Dubrovsky, Kiesler, & Sethna, 1991; Kiesler, Siegel, &McGuire, 1984).

However, the current study also shows that sharing the samevirtual appearance does not always lead to congeniality in com-puter-mediated groups. Too much similarity shared among indi-viduals can interfere with group identification process bythreatening their individual uniqueness, and even induce deviantbehaviors as a way to restore their deprived uniqueness. Such devi-ant behaviors as a backlash to enforced visual similarity can be aproblem if virtual community members have to make an importantdecision as a group. For instance, a member of an online commu-nity that make every member look alike might disagree with oth-ers as a way to refurbish his/her uniqueness among communitymembers or receive attention. If the member’s desire to restorehis/her uniqueness or stand out overshadows expressing whathe/she truly believes right, it will discourage sound and rationaldecision-making process. In order to reduce such unintended sideeffects from uniform or standardized visual representation, devel-opers and administrators of virtual communities should be carefulin deciding how much space for individuation is allowed for theirmembers even when they want to use uniform appearance as away to promote congeniality within the communities.

There are some limitations in this study. Although it was a nec-essary measure for holding consensus norm constant across differ-ent conditions and controlling for unexpected interaction amongdiscussants, using a programmed discussion might have reducedecological validity of the online discussion. In future studies, par-ticipants should have actual discussions and be asked to reach adecision within a limited time. Secondly, the present study onlylooked at the situation in which there is one dominant opinionshared by other discussants. This might have brought a confound-ing effect in the uniform-appearance condition: a participant couldhave shown willingness to agree with other discussants, either be-cause he/she wanted to be accepted by a majority group whoshared the same image as his/hers or because he/she compliedwith the consensus pressure coming from the majority opinion.In a future study, these two potentially confounding influencesshould be disentangled.

The findings of the current study – sharing uniform appearancewith others can make individuals perceive that their uniqueness as

J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230 1229

individuals is threatened and hesitant to conform to a majorityopinion – suggests a major assumption on human motivationand behavior: once individuals perceive a discrepancy between adesired state and an existing state, they do something to eliminatethe discrepancy (Higgins, 1987; Kruglanski, 1996). Given that‘‘states’’ are not permanent and that the desired state can beachieved through diverse means (Heider, 1958), it is possible to as-sume that the negative effect induced by individuals’ perceiveddeindividuation on majority influence should be weakened if theywere given opportunities to regain their uniqueness (Imhoff & Erb,2009). Based on this preposition, a future study can investigatemore dynamic changes of desired and satisfied states of unique-ness, for example, by testing whether individuals who regainuniqueness after being shorn of uniqueness would be more willingto conform to a majority opinion compared to those who are leftdeprived of their uniqueness without any remedy.

Appendix A

A.1. Scales

�Perceived deindividuation

(1) I think I could be who I am in this group (reversed item).(2) I think I saw myself predominantly as an individual in this

group (reversed item).(3) I think I was not considered as a distinctive individual in this

group.(4) I think members in this group were not represented as

unique individuals in this group.

�Group identification

(1) I have a lot in common with other group members.(2) I find it easy to identify with this group.(3) I find that my values and the values of other group members

are very similar.

�Conformity intention

(1) I am willing to agree with them.(2) I am willing to follow their opinion.(3) I am willing to join them as the same group member.(4) I am willing to share the same opinion as the same group

member.

Appendix B

B.1. Discussion topic

Ms. E, a college senior, has studied the piano since childhood.She has won amateur prizes and given small recitals, suggestingthat she has considerable musical talent. As graduation ap-proaches, she has the choice of taking a medical school scholarshipto become a physician, a profession which would bring certainfinancial rewards, or entering a conservatory of music for advancedtraining with a well-known pianist. She realizes that even uponcompletion of her piano studies, success as a concert pianist wouldnot be assured.

A: Ms. E should enter a conservatory of music for advancedtrainingB: Ms. E should choose to take a medial school scholarship tobecome a physician

References

Allen, V. L. (1965). Situational factors in conformity. In L. Berkowitz (Ed.), Advancesexperimental social psychology (pp. 133–176). Orlando: Academic Press.

Arbuckle, J. L. (2005). Amos 6.0. Spring house. PA: Amos Development Corporation.Asch, S. E. (1956). Studies of intendance and conformity: A minority of one against a

unanimous majority. Psychological Monographs, 70, 1–70.Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in

social psychological research: Conceptual, strategic and statisticalconsiderations. Journal of Personality and Social Psychology, 51, 1173–1182.

Baumeister, R. F. (1991). Meanings of life. New York: Guilford.Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal

attachments as a fundamental human motivation. Psychological Bulletin, 117,497–529.

Brewer, M. (1991). The social self: On being the same and different at the sametime. Personality and Social Psychology Bulletin, 17, 475–482.

Brewer, M. (1993). The role of distinctiveness in social identity and group behavior.In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychologicalperspectives (pp. 1–16). New York: Harvester-Wheatsheaf.

Cheney, G. (1983). On the various and changing meanings of organizationalmembership: Field study of organizational identification. CommunicationMonographs, 50, 342–362.

Codol, J. P. (1984). Social differentiation and non-differentiation. In H. Tajfel (Ed.),The social dimension: European developments in social psychology (pp. 314–337).Cambridge: Cambridge University Press.

Collins, L. M., Graham, J. W., & Flaherty, B. P. (1998). An alternative framework fordefining mediation. Multivariate Behavioral Research, 33, 295–312.

Cooper, J., & Jones, E. E. (1969). Opinion divergence as a strategy to avoid beingmiscast. Journal of Personality and Social Psychology, 13, 23–30.

Davis, M. D. (1985). The logic of casual order. In J. L. Sullivan & R. G. Niemi (Eds.),Sage university paper series on quantitative applications in the social sciences.Beverly Hills: Sage Publications.

Deutsch, M., & Gerard, H. (1955). A study of normative and informational socialinfluences upon individual judgment. Journal of Personality and SocialPsychology, 51, 629–636.

Dubrovsky, V. J., Kiesler, S., & Sethna, B. N. (1991). The equalization phenomenon:Status effects in computer-mediated and face-to-face decision-making groups.Human Computer Interaction, 6, 119–146.

Fromkin, H. L. (1970). Effects of experimentally aroused feelings of indistinctivenessupon valuation of scarce and novel experiences. Journal of Personality and SocialPsychology, 16, 521–529.

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.Higgins, E. T. (1987). Seld-discrepancy: A theory relating self and affect.

Psychological Review, 94, 319–340.Hiltz, S. R., & Turoff, M. (1978). The network nation. Cambridge: MIT Press.Hornsey, M. J., & Jetten, J. (2004). The individual within the group: Balancing the

need to belong with the need to be different. Personality and Social PsychologyReview, 8, 248–264.

Hu, L.-T., & Bentler, P. (1999). Cutoff criteria for fit indexes in covariance structureanalysis: Conventional criteria versus new alternatives. Structural EquationModeling, 6, 1–55.

Imhoff, R., & Erb, H.-P. (2009). What motivates nonconformity? Uniqueness seekingblocks majority influence. Personality and Social Psychology Bulletin, 35,309–320.

Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and fiascoes(2nd ed.). Boston: Houghton Mifflin.

Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social psychology. InD. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed.,pp. 233–265). New York: McGraw-Hill.

Kiesler, S., Siegel, J., & McGuire, T. (1984). Social psychological aspects of computer-mediated communication. American Psychologist, 39, 1123–1134.

Kim, J.-H. (2009). ‘‘I want to be different from others in cyberspace’’ The role ofvisual similarity in virtual group identity. Computers in Human Behavior, 25,88–95.

Kruglanski, A. W. (1996). Motivated social cognition: Principles of the interface. In E.T. Higgins & A. W. Kruglanski (Eds.), Social psychology: A handbook of basicprinciples (pp. 493–522). New York: Guilford Press.

Lea, M., & Spears, R. (1991). Computer-mediated communication: Deindividuationand group decision-making. International Journal of Man–Machine Studies, 34,283–301.

Lee, E. (2004). Effects of visual representation on social influence in computer-mediated communication: Experimental tests of the social identity model ofdeindividuation effects. Human Communication Research, 30, 234–259.

Lee, E., & Nass, C. (2002). An experimental test of normative group influence andrepresentation effects in computer-mediated communication: Wheninteracting via computers differs from interaction with computers. HumanCommunication Research, 28, 349–381.

Mackie, D. M. (1987). Systematic and nonsystematic processing of majority andminority persuasive communications. Journal of Personality and SocialPsychology, 53, 41–52.

MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. AnnualReview of Psychology, 58, 593–614.

MacKinnon, D. P., Krull, J. L., & Lockwood, C. M. (2000). Equivalence of themediation, confounding, and suppression effect. Prevention Science, 1, 173–181.

1230 J. Kim, H.S. Park / Computers in Human Behavior 27 (2011) 1223–1230

Maslach, C. (1974). Social and personal bases of individuation. Journal of Personalityand Social Psychology, 29, 411–425.

Maslow, A. H. (1968). Toward a psychology of being. New York: Van Nostrand.Nunamaker, J. F., Briggs, R. O., Mittleman, D. D., Vogel, D. R., & Balthazard, P. A.

(1997). Lessons from dozen years of group support systems research: Adiscussion of lab and field findings. Journal of Management Information Systems,13, 163–207.

Postmes, T., Spears, R., & Lea, M. (1998). Breaching or building social boundaries?SIDE-effects of computer-mediated communication. Communication Research,25, 689–715.

Postmes, T., Spears, R., & Lea, M. (1999). Social identity, normative content and‘‘deindividuation’’ in computer-mediated groups. In N. Ellemers, R. Spears, & B.Doosje (Eds.), Social identity: Context, Context, Commitment, Content(pp. 174–183). Oxford: Blackwell.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirecteffects in simple mediation models. Behavior Research Methods, Instruments, andComputers, 36, 717–731.

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies forassessing and comparing indirect effects in multiple mediator models. BehaviorResearch Methods, 40, 879–891.

Reicher, S. (1987). Crowd behavior as social action. In J. C. Turner, M. A. Hogg, P. J.Oakes, S. D. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group: Aself-categorization theory (pp. 171–202). Oxford: Blackwell.

Santee, R. T., & Maslach, C. (1982). To agree or not to agree: Personal dissent amidsocial pressure to conform. Journal of Personality and Social Psychology, 42,690–700.

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimentalstudies: New procedures and recommendations. Psychological Methods, 7,422–445.

Simonson, I., & Nowlis, S. M. (2000). The role of explanations and need foruniqueness in consumer decision making: Unconventional choices based onreasons. Journal of Consumer Research, 27, 49–68.

Snyder, C. R., & Fromkin, H. L. (1977). Abnormality as a positive characteristic: Thedevelopment and validation of a scale measuring need for uniqueness. Journal ofAbnormal Psychology, 86, 518–527.

Snyder, C. R., & Fromkin, H. L. (1980). Uniqueness: The human pursuit of difference.New York: Plenum.

Tajfel, H. (1981). Human groups and social categories. Cambridge: CambridgeUniversity Press.

Terry, D., & Hogg, M. (2001). Attitudes, behavior and social context: The role ofnorms and group membership in social influence processes. In J. Forgas & K.Williams (Eds.), Social influence: Direct & indirect processes (pp. 253–270).Psychology Press.

Turner, J. C. (1984). Social identification and psychological group formation. In H.Tajfel (Ed.), The social dimension: European developments in social psychology(pp. 518–538). Cambridge: Cambridge University Press.

Wood, W., Lundgren, S., Ouellette, J. A., Busceme, S., & Blackstone, T. (1994).Minority influence: A meta-analytic review of social influence processes.Psychological Bulletin, 115, 323–345.

Zimbardo, P. (1969). The human choice: Individuation, reason, and order versusdeindividuation, impulse, and chaos. In W. Arnold & D. Levine (Eds.), Nebraskasymposium on motivation (pp. 237–307). Lincoln: University of Nebraska Press.