Self-Presentation in IRC

Julie Wolf

Cara Kuperman

Michèle Munoz

(Draft: Do Not Cite)

Abstract

The nature of self-presentation and impression formation was contrasted between Internet Relay Chat (IRC) and face-to-face (FtF) environments. Groups of previously unacquainted persons completed the NEO-FFI personality measurement for "how they are" and "how they wish to be". After interacting FtF or via IRC, subjects judged their group members' personalities using the same measurement. Difference scores between S's self-rated, S's ideal, and judged personalities were compared across FtF, novice IRC and expert IRC conditions. Interactions were coded and compared to measure differential use of self-presentational strategies. Across conditions, Ss were judged more in line with how they were than how they wished to be. IRC experts self-disclosed and employed creative self-presentational strategies. Results partially supported the hypothesis that IRC experts might be seen more like how they wish to be within a reduced cues medium.

Introduction

As we move further into the "Computer Age", ever increasing numbers of social and business interactions are occuring in online forums. Such computer-mediated communication (CMC) is devoid of the nonverbal cues that have been shown to be integral to both the way that people naturally present themselves and how they are perceived by others. It is therefore beneficial to examine the way in which CMC interactions may differ from everyday face-to-face (FTF) interactions.

Nonverbal cues play an important role in daily social interaction. Individuals form impressions of one another through use of stereotypes. "By categorising people into 'types' we are able to systematise our accumulated knowledge about people and greatly simplify the task of person perception". The most efficient way to form a reasonably accurate first impression of another individual is to decide the category, or stereotype, to which that person belongs, and judge his characteristics accordingly. (Forgas, 1985) The stereotypes which we form of others are often based on nonverbal cues AND physical characteristics, such as race, dress and attractiveness, and thus these cues play an important part in impression formation. Nonverbal cues are equally important in self-presentation, in that individuals make use of behaviors such as smiling, nodding and dressing in a particular fashion, in order to convey a favorable image (DePaulo, 1992).

CMC, however, is typically purely text based, and such nonverbal cues are therefore absent. (Reid, 1991) One would therefore expect there to be differences in social interactions between FTF and CMC environments. Such differences are posited by the "cues filtered out" approach. Specifically, the cues filtered out approach suggests that CMC interactions, because of the lack of nonverbal cues, are by nature less personal than FTF interactions. (Walther, 1992 {move this reference to the end of the preceding sentence})

Recent research (e.g., Reid, 1991), however, has observed greater levels of socio-emotional behavior in CMC than in FTF settings, suggesting that the cues filtered out approach is not complete. To account for these findings, Walther proposes the Social Information Processing perspective. Because textual transmission of information is by nature slower than spoken transmission, the rate of social information exchange is slower in CMC than in FTF settings. The Social Information Processing (Why arethese words capitalized? If this usage is borrowed from Walther, then the parenthatical reference should probably be here.) approach suggests that CMC interactions are less personal than FTF interactions because insufficient time has been allowed for comparable socio-emotional content to develop. (Walther, 1992; Walther, 1996)

How is it that, over time, CMC users are able to compensate for the lack of channels through which social information can be assessed? The answer is [{Seems to be?} that users actively create {text-based?} forms of expressive communication, which results in a unique culture of the virtual world. Thus, cues which are normally provided through sight, sound, and touch are now provided through description. (Baym, 1995; Reid, 1995) There are many examples of such strategies for tackling the lack of nonverbal socio-emotional cues in CMC. One is to describe physical actions and reactions in a text based form, as in "<nickname> winks". Another convention is the use of "emoticons", such as the smiley face [:-)], which are used to indicate the intended affect of the person making the statement.

One effect of the lack of nonverbal cues in CMC settings is that it creates a sense of anonymity. Reid (1995) suggests that this anonymity allows CMC users to become disinhibited in their behavior: the anonymity provides users with a feeling of safety, which in turn allows them to feel secure enough to "come out of their shell". This disinhibition can manifest itself in {How about rephrasing this "either through increased ... or through ... To"?} either of two ways: through increased displays of intimacy or through increased displays of aggression.

In addition, the anonymity provided by CMC allows individuals to experiment with different persona{"personas," unless we want to go for the erudite "personae"} -- the aspects of their personalities which they suppress in FTF settings due to "real life pressures" (Turkle, 1995; Reid, 1991). An IRC user nicknamed Blitz (personal communication, September, 1996) described this phenomenon, indicating that IRC is different from FTF settings "in that here I can be whomever and whatever I wish you to think I am ... And you've no way to see differently as I'm just text on a screen".

This creation of the ideal self is aided through the self-presentational strategies utilized by CMC users. Erving {first name unnecessary} Goffman described social behavior as a performance, whose purpose is to allow the individual to present himself in a favorable manner so as to influence the impressions which others form of him: "In a sense, and in so far as this mask represents the conception we have formed of ourselves -- the role we are striving to live up to -- this mask is our truer self, the self we would like to be" (Goffman, 1959.

Walther (1996) terms this performance "selective self-presentation". Selective self-presentation is limited in FTF settings by nonverbal cues, such as physical appearance or emotional displays, which are unalterable or difficult to suppress. In CMC, however, where nonverbal cues are not present, individuals have greater control over which of their characteristics (real or invented) they make known to others, allowing CMC users to present themselves in a socially ideal manner (Reid, 1991).

Because the ability for selective self-presentation is enhanced in CMC, one would expect that impressions formed in the CMC medium should differ from those formed in FTF settings. A study on impression formation in CMC by Fuller (1994) investigated the accuracy of personality judgments made in CMC versus FTF environments. Individuals were asked to complete the Keirsey personality type indicator (Keirsey, _) imagining that they were a particular target individual. For one group, this target individual was a co-worker with whom the subject had only previously communicated via electronic mail (e-mail); for the other group, the target was a co-worker with whom the subject had only communicated face-to-face. All target individuals were then contacted and asked to fill out the personality measure for themselves. Results showed that within the e-mail condition, subjects rated their co-workers higher on both the "thinking" and "judging" sub-scales than the coworkers rated themselves. No significant results were found for the face-to-face group. This suggests that individuals form mis-perceptions about others with whom they communicate only through a CMC medium.

The Present Study

Other than Fuller's study, little empirical research has investigated impression formation and self-presentation in a CMC setting. In particular, no research has investigated these phenomena in a synchronous CMC setting (one in which interactions occur in "real time"). Such a medium for communication is unique in that users can engage in "live" conversation as in FTF settings, yet the effects of the anonymity and of the absence of nonverbal cues still apply. The CMC medium which will be {delete: we've done the study} investigated in the present study is [WAS] Internet Relay Chat (IRC).

The Internet Relay Chat program was written by Jarkko Oikarinen in 1988 {Unnecessary sentence}. IRC is a "network of conversational channels that can be accessed by connecting, through the Internet, to one of many networked servers across the country and world" (Serpentelli, 1992). {How about a sentence here to the effect that "IRC has been available to Internet users since 1988" and add a citation of Oikarinen here?} Users convene in "channels", or chat rooms, and communicate through typed textual messages, which are received instantaneously by other IRC users within the same channel. (Reid, 1991) The setting is therefore analogous to a conference phone call (Serpentelli, 1992). In addition to the general public discussion within a channel, it is possible to send a private message to another IRC user, so that the message will be visible to only that user. (Reid, 1991) On average {We'd better have a date here, since these numbers are probably exponentially increasing.}, 33,984 users access IRC's two most popular servers (Efnet and Undernet) over the course of a day. The typical IRC user is young (as a large percentage of its users are college students), male, and likely of middle or upper class standing (Computational Mechanics, 1996).

The present study used the Five Factor model to examine personality judgments made on IRC versus FTF. The purpose of the present study was twofold: to determine whether impressions formed on IRC differ from those formed in FTF settings, and to ascertain the extent to which any differences in impressions formed are due to the use of differential self-presentational strategies in IRC versus FTF settings. Subjects completed the NEO-FFI {refs?} personality measure for "how they {"you"?} are" and "how they {"you"?} wish to be". They then interacted in one of three conditions: a FTF setting, an IRC setting in which all subjects were novice IRC users, or an IRC setting in which all subjects were expert IRC users. Subjects' NEO-FFI personality judgments of one another were measured {How about "collected immediately following each session"?}, and {logs of verbal} interactions were analyzed for use of self-presentational strategies.

The following hypotheses were posited:

1a) Because FTF interactions are constrained by uncontrollable nonverbal cues, subjects in the FTF group should present themselves as more in line with their actual self ratings than subjects in the IRC groups.

1b) Because IRC is characterized by an absense of such constraints of nonverbal cues, subjects in the IRC groups should present themselves as more in line with their ideal self ratings than subjects in the FTF group.

As a result of these proposed differential self-presentational strategies,

2a) Subjects in the FTF group should be perceived as more in line with their actual self ratings than subjects in the IRC groups.

2b) Subjects in the IRC groups should be perceived as more in line with their ideal self ratings than subjects in the FTF group.

3a) Within the IRC groups, subjects should be judged as more like their ideal than their actual self ratings.

3b) Within the FTF group, subjects should be judged as more like their actual than their ideal self ratings.

4) Subjects in the IRC groups should display more disinhibited behavior patterns than subjects in the FTF group. This disinhibition should be manifested in greater use of behaviors such as "smiles" and "hugs", a greater degree of self-disclosure/ intimacy as compared to the FTF group, more equalized participation in conversation amongst group members, and perhaps a greater display of aggression.

Method

Design. The study was a 3x2x2 mixed design. The independent variables were: communication condition, subject sex, and counterbalanced questionnaire order.

Subjects. There were 56 subjects (50% female). Thirty-eight IRC-naive subjects were recruited from Haverford College with the incentive of gaining experimental credit for their participation; these subjects were randomly assigned either to the IRC novice or FTF condition. Eighteen IRC-experienced subjects were recruited directly from IRC with a financial incentive; these subjects comprised the IRC expert condition.

Materials. A "Background Questionnaire" was devised by the authors to collect data on age, gender, experience with computers, the Internet, and IRC.

The NEO-FFI Form S was employed to tap into five dimensions of personality and to allow for comparisons in personality judgments across both judges (self vs. observer; observer vs. observer) and conditions (FTF, novice IRC, and expert IRC).

Procedure. Subjects in the novice IRC and FTF conditions were informed that this was a study about the way in which people form impressions of one another. Subjects in the expert IRC condition were told specifically that the study was interested in how people form impressions of one another on IRC (in order to justify why it was necessary to recruit them from IRC).

All subjects first completed the Background Questionnaire. They then completed the NEO-FFI for "how they are" (actual self) and for "how they wish to be" (ideal self). An experimenter assured subjects that all data will be held in the strictest confidence, and that names would be replaced with codes prior to analysis.

After completing the above measures, subjects were escorted to a physical room (FTF conditions) or a designated IRC channel (IRC novice and expert conditions) which were videotaped or logged via computers, respectively. Within each condition, subjects had been divided into five groups of four individuals (two male and two female). Because certain subjects did not report for the experiment, four of the groups (one novice IRC, one FTF, and two expert IRC) comprised only three individuals. The groups were assigned such that no members of a particular group had any prior acquaintance.

After the hour had elapsed, subjects completed the NEO-FFI for each member of their group according to the impressions they formed of these interactants. Subjects were then thanked for their participation and debriefed regarding the nature of the study.

Calculations

The self and observer judgments allowed for the calculation of three separate types of "difference scores." Actual-ideal difference scores were calculated to assess actual and ideal self-ratings relative to one another. Two other difference scores were calculated in a similar fashion which would be used to compare self and observer judgments. Judged-Actual difference scores were calculated by measuring the discrepancy between the average judgment made by group members for a given subject and that subject's actual self-rating. The Judged-Ideal difference score was based on the discrepancy between the average judgment made by the subjects' group members and the subjects' ideal score. All difference scores were calculated along all five dimensions for each subject.

Coding A coding scheme was devised which measured the occurrence of behaviors relating to the Big Five dimensions and to disinhibition in the logs (IRC) and videotapes (FTF). Each session was coded individually by two experimenters, in order to assess the degree of interrater reliability obtained. Coded behaviors included the following:

1) Total number of smiles per person. In the IRC conditions, smiles were considered to be any emoticon or variation on the smiley face, including winks and frowns. In the FTF condition, a smile was counted whenever a subject smiled or conveyed emotion through the use of a facial expression.

2) Total number of laughs per person. In the IRC conditions, this consisted of any written indication of laughter, such as "lol", "*laugh*", "heehee", or any action conveying laughter (i.e., "John laughs"). In the FTF condition, any distinct occasion when a subject laughed or giggled was counted as one laugh.

3) Total number of utterances per person and per group. In the IRC conditions, an utterance was defined as any line of "speech" that was separated from another by a carriage return. In the FTF condition, an utterance was defined as any uninterrupted period of speech by one subject.

4) Number of times subjects' names were mentioned during the session. The mention of names consisted of occasions in which a subject was directly addressed by name or was talked about by another individual.

5) Total number of self-directed actions per person. These were defined as actions directed toward the self. For example, in the IRC conditions, the typed action "John yawns" would be considered a self-directed action. In the FTF condition, a subject physically yawning or stretching would be considered a self-directed action.

6) Total number of other-directed actions per person. These were defined as actions involving a subject doing something onto another subject. In the IRC conditions, these consisted of typed actions that were directed at another group member, such as, "John hands a rose to Mary". In the FTF condition, an other-directed action was defined as any deliberate action characterized by physical movement toward or regarding another group member, such as nodding ones head in agreement, or moving out of the way of that person.

7) Total number of self-directed statements per person. These were defined as utterances that revealed information about the self, or that conveyed one's own knowledge. For example, the statements "I am a college student" and "I don't like that song" would be coded as self-directed statements, in that both disclose information about the person who made them.

8) Total number of other-directed statements per person. These were defined as utterances that elicited information from another individual, related to something someone else said, or otherwise related to a group member. Thus, the question "What classes are you taking?" and the statements "You're funny" and "Be nice to him" would all be coded as other-directed statements.

Utterances were coded as both other-directed and self-directed if they conveyed both types of information. For example, the statement, "I can see how that might be difficult for you" contains elements that are characteristic of both self-directed and other-directed speech and would thus be coded as both.

In making decisions about specific utterances, the context in which the utterance was made was taken into account. For example, the response "yeah" could be coded as self-directed, other-directed, or both, depending on the context in which it was uttered.

From these coded variables, certain other variables were computed. These included: the percentage of a given subject's total utterances that were self-directed; the percentage of a given subject's total utterances that were other-directed; the percentage of a group's total utterances that were made by a given group member; the expected percentage of a group's total utterances that should have been made by each group member (ie, 25% if the group consisted of four subjects, 33% if the group consisted of three subjects); and the difference between the actual percentage of utterances per person and the expected percentage of utterances per person (an indicator of the equalization of participation in discussion among group members).

Results

Subject Age. The overall mean age of subjects across conditions was 22.1 years. A t test revealed a significant difference in age between the IRC experts and the subjects recruited from Haverford College (t(54)=4.84, p<.001; M_expert=28.9, M_other=18.8).

Hypothesis 1(a&b): Subjects in the FTF group should present themselves and thus be perceived as more in line with their actual self ratings than subjects in the IRC groups. Conversely, subjects in the IRC groups should present themselves and thus be perceived as more in line with their ideal self ratings than subjects in the FTF group.

In order to assess whether subjects employed self-presentational strategies in line with their actual vs. ideal NEO-FFI scores, it was necessary to establish which of the coded self-presentational behaviors (footnote #1) were indicative of which of the Five Factors. Thus, correlations were computed between the coded self-presentational measures and the mean actual, ideal, and judged NEO-FFI scores for each condition. Table 1 presents the correlations between the dimensions and the coded behaviors which were significant at the .05 level. Since few meaningful relationships were uncovered, it was concluded that the self-presentational behaviors for which we coded did not adequately correspond to the Five Factors; therefore, Hypothesis 1 could not be tested.

Hypotheses 2 through 4 required a comparison of the judged-actual and judged-ideal difference scores (in order to assess whether perceptions were more in line with actual or ideal self-ratings). Thus, an ANOVA was conducted treating these two difference scores as two levels of a within subjects variable. Interactions were uncovered between group and the bi-level variable for the dimensions of Neuroticism order one (footnote #3) [F(2,26)=4.50, p<.05], Neuroticism - order two [F(2,24)=2.9, p<.10], Extraversion [F(2,53)=2.76, p<.10], Openness [F(2,53)=7.86, p<.01] and Conscientiousness [F(2,53)=3.65, p<.05] dimensions.

Hypothesis 2(a&b): Subjects in the FTF group should be perceived more in line with their actual self ratings than subjects in the IRC groups. Conversely, subjects in the IRC groups should be perceived as more in line with their ideal self ratings than subjects in the FTF group.

Tukey HSD post-hoc comparisons revealed between group differences for the dimensions of Extraversion (p<.10) and Openness (p<.05). IRC Experts were judged more like their ideal Extraversion scores than were either IRC novices or FtF subjects (p<.01). IRC Experts were also judged somewhat more like their actual Extraversion scores than were FTF subjects (p<.10). In addition, IRC experts were judged significantly more like their ideal Openness scores than were IRC novices (p<.01). For the same dimension, FTF subjects were judged more like their actual scores than were IRC experts (p<.05). No significant between group differences were found for the Neuroticism (either order) or Conscientiousness dimensions.

Hypothesis 3(a&b): Within the IRC groups, subjects should be judged as more like their ideal than their actual self ratings. Conversely, within the FTF group, subjects should be judged as more like their actual than their ideal self ratings.

Tukey's HSD post-hoc comparisons revealed that subjects in both the IRC Novice and FTF conditions were judged as closer to their actual self ratings than to their ideal self ratings on the Neuroticism - order one (p<.001), Neuroticism - order two (p<.05), Extraversion (p<.001) and Openness (p<.001) dimensions. IRC experts were also seen as closer to their actual than ideal selves on the Neuroticism - order one (p<.10) and Extraversion (p<.01) dimensions. No significant within group differences were found for the Conscientiousness dimension.

Hypothesis 4: Subjects in the IRC groups should display more disinhibited behavior patterns than subjects in the FTF group. This disinhibition should be manifested in greater use of behaviors such as "smiles" and "hugs", as well as a greater degree of self-disclosure/ intimacy as compared to the FTF group, more equalized participation in conversation amongst group members, and perhaps a greater display of aggression.

A Mann Whitney U test was conducted comparing the difference scores between expected and actual percentages of participation in the FtF condition with those from the IRC conditions (combined). This test revealed that there was a trend towards greater equalization of verbal participation across subjects in the two IRC conditions as compared to the FtF condition [U=194.5, p<.10].

An overall ANOVA was conducted on the percentage of total utterances per subject which were other-directed. A significant main effect of condition was found on the mean percentages of other-directed speech [F(2,49)=5.34, p<.01]. A higher proportion of speech in the IRC novice than in the IRC expert condition was coded to be other-directed (p<.01). The mean percentages of other-directed speech in the IRC conditions are presented in Table 2.

Several coded self-presentational behaviors were analyzed through t-tests between the two IRC conditions (footnote #4). Differences in the use of text-based actions were revealed in these comparisons. Table 3 displays all coded self-presentational data for the expert and novice IRC conditions.

{Note to anyone who's actually reading this: these are the 2 paragraphs that fit in!}:

In order to assess differences in the relationship between actual and ideal self ratings across the three conditions, a difference score was computed by subtracting each subject's ideal NEO-FFI score from his actual NEO-FFI score on each of the five dimensions. These difference scores revealed that, across conditions, subjects wished to be less neurotic, more extraverted, more open, more agreeable, and more conscientious than they reported themselves as actually being (p<.001).

ANOVAs were then conducted to compare the actual-ideal difference score across the three conditions for each of the five factors. A significant main effect of condition on the actual-ideal difference score was found for the Neuroticism dimension with counterbalanced questionnaire order one (F(2, 26)=4.50, p<.05) and for the Openness dimension (F(2, 53)=7.86, p<.01). Tukey's HSD post-hoc comparison revealed that on both dimensions, the IRC experts had a smaller discrepancy between their actual and ideal scores than did the subjects in the IRC novice (p<.05) and FTF (p<.10) conditions. A significant main effect of condition on the actual-ideal difference score was also found for the Conscientiousness dimension (F(2, 53)=5.13, p<.01). A Tukey's HSD post-hoc comparison showed that the IRC novices had a larger discrepancy between their actual and ideal Conscientiousness scores than did the IRC expert (p<.05) and FTF (p<.10) subjects. A nearly significant main effect of condition on the actual-ideal difference score was found for the Neuroticism dimension with questionnaire order two (F(2, 24)=2.89, p<.10) and for the Extraversion dimension (F(2, 53)=2.76, p<.10). Tukey's HSD post-hoc comparisons revealed that for both dimensions, IRC experts had a smaller discrepancy between their actual and ideal scores than did the IRC novices (p<.10). No main effect of condition on the actual-ideal difference score was found for the Agreeableness dimension (p>.10).

Footnote #1: Correlations were conducted in order to assess the degree of interrater reliability for the codings of the IRC logs and FTF videotapes for self presentational behavior. The correlation coefficient was .90 or above for all of the coded variables; because such a high degree of interrater reliability was obtained, the two experimenter codings were averaged for each of the self presentational measures.

Footnote #2: Correlations were conducted in order to assess the degree of agreement between subjects in their judgments of a particular group member. Thus, for each subject, three peer correlations were conducted (between peers one and two, two and three, and one and three) for each of the five dimensions, yielding a total of 15 correlations. These correlations are depicted in Table 1. Of the 15 correlations, nine were significant at the p<.05 level. All inter-peer correlations on the Extraversion and Neuroticism dimensions were significant at the p<.05 level, while none of the inter-peer correlations were significant for the Conscientiousness dimension. The judgments of a particular subject by his three group members were averaged to obtain an overall "judged" score on each of the five dimensions, for use in all subsequent analyses.

Footnote #3: An ANOVA revealed a significant main effect of counterbalanced questionnaire order on subjects' ratings of their ideal selves on the Neuroticism dimension (F(1, 50)=8.97, p<.01). Subjects who completed the NEO-FFI for their actual self followed by their ideal self (order one) had a lower ideal Neuroticism score than did subjects who received the questionnaire in the opposite order (order two). Thus, all subsequent analyses involving ideal self ratings on the Neuroticism dimension were examined separately for each counterbalanced order.

Footnote #4: Because the FTF group was not comparable to the two IRC groups on many of the variables (for example, total number of utterances was higher in the FTF condition because of the lack of the constraint of typing speed), the majority of the self-presentational measures could only be compared between the two IRC conditions.

Discussion

Self-Presentation. It was predicted that the subjects in the reduced-cues IRC conditions would present themselves closer to their "Ideal" personality profiles than would those in the FtF condition. Similarly, it was hypothesized that subjects in the FtF condition would present themselves closer to their "Actual" personality profiles than would subjects in the IRC conditions. It became apparent, however, when creating the coding scheme, that clear-cut relationships between coded behaviors and the NEO-FFI personality dimensions could not be posited. Therefore, the hypotheses concerning differences in self-presentational strategies across conditions could not be reliably tested.

Impression Formation. The hypotheses surrounding impression formation were only partially supported. Specifically, it was predicted that the subjects in the FtF condition would be perceived more like their "Actual" personality profiles than would subjects in the IRC conditions. This was shown to be the case only for the dimension of Openness. Within Extraversion, the opposite was found to be true, in that IRC Experts were judged closer to their "Actual" selves than were subjects in the FtF condition.

It was also hypothesized that subjects in the IRC conditions would be perceived more like their "Ideal" selves than those in the FtF condition. This was found to be true for Extraversion only - IRC experts were judged closer to their "Ideal" selves than the subjects in both remaining conditions. IRC novices, however, were not perceived more like their "Ideal" selves than subjects in the FtF condition on any of the five personality dimensions. Another unexpected finding was that IRC Experts were perceived more like their "Ideal" selves than the IRC novices within the dimensions of Extraversion and Openness.

Contrary to the predictions made, across all conditions, no subjects were perceived as closer to their "Ideal" selves than their "Actual" selves; and, in fact, for the Extraversion dimension, the opposite was true. Within the IRC Novice and FtF conditions, similar results were found for Neuroticism and Openness. A trend in the same direction was discovered within Neuroticism among IRC.

Overall, the results found concerning impression formation were not as powerful as anticipated. It would seem that while some results suggest that they were perceived more like their "Ideal" selves than subjects in the FtF condition, IRC Experts are not capable of completely obscuring their true selves when interacting with other experienced IRC users. This was evidenced by their having been judged closer to their "Actual" profile scores along the dimension of Extraversion.

Perhaps it is because they were interacting with other experienced IRC users that this result was found. Other IRC Experts, well versed in the self-presentational strategies common to CMC, may be able to look beyond the text-based surface and see the true selves of other experienced participants. Perhaps it is this knowledge of the medium that results in the finding that the IRC Experts rated each other closer to their "Actual" profile scores than did the subjects in the FtF condition. Because specific self-presentational strategies are not as salient during everyday FtF interactions, subjects in that condition might be less aware of the likelihood that their fellow participants would selectively self-present.

Disinhibition. As detailed earlier, it has been suggested that the anonymity of a reduced-cues medium results in increased disinhibition online. One way in which this effect was evidenced in our subjects was by greater equalization of participation among the subjects in both IRC conditions. This would suggest that the intimidation commonly felt in a face-to-face encounter, perhaps due to inferior status or to physical stature, is not as present in CMC. Another manifestation of disinhibition was thought to be the more frequent use of "smiles" and "hugs" by IRC users than by their face-to-face counterparts. It was found that these gestures were used much more in the IRC Expert condition than in either of the remaining conditions. The difference in use between the IRC Experts and the IRC Novices can be explained by greater familiarity with the medium among the experts.

Reid (1995) proposes that online communication allows for more intimate disclosure, another function of disinhibition. The expert subjects in the present study exhibited this in their discussions, as many individuals disclosed incredibly personal information about their lives. While one discussed her somewhat rocky relationship with her mother, another shared her experience of the death of a best friend. These are only some examples of the many personal conversations held online between the expert IRC users. While the novice online group did not get nearly as intimate with their fellow participants, this may be due to their infamiliarity and resulting awkwardness with the online experience.

Although it was also suggested by Reid (1995) that disinhibition online would result in more aggressive behavior, such was not found in the present study. This may be due to the fact that this was an examination of the first encounter between individuals. Perhaps aggression among online IRC users develops over time, as relationships build and change. It is also possible that because the IRC users knew that they were being observed and were part of an investigation into online communication, they were on their best behavior.

Limitations of the Present Study. A significant difference in age between the subjects in the IRC Expert condition and the subjects in the other two conditions may have played a part in the results found. It had been the assumption of the investigators that the expert IRC users would be relatively matched in age to the college students used for the other conditions. This was not the case however, as it was found that most of the college-aged IRC experts contacted to be in the study were reluctant to participate, perhaps due to other time commitments. Further study into this area should focus more attention on recruiting subjects matched in age across all conditions.

As has been mentioned earlier, the subjects in the IRC Novice condition were, by definition, unfamiliar with the medium. This may have hindered their ability to utilize all of the alternative self-presentational tools at their fingertips, such as "emoticons" and actions. It may also have interfered with their ability to form full impressions of other individuals online, as they were not accustomed to the form of communication at hand. Finally, a factor which may have greatly affected the level of disinhibition among the IRC novices is that they all attended one of two small colleges, and the probability of encountering one another at a later time was therefore quite high. This threat to anonymity may have resulted in less disinhibition during their conversations.

Implications of the Present Study. Overall, these findings suggest that that there are some substantial differences in self-presentational strategies between IRC and face-to-face interactions. Specifically, it is possible that the reduced-cues medium allows for selective self-presentation, and the creation of a more ideal profile of the participant, particularly within the dimension of Extraversion. This, in turn, affects the impression formed of the individual by other members of the online community.

Because of the unreliability of the correlations between behaviors and personality dimensions, however, we were unable to determine completely whether or not people present themselves in a manner closer to their ideal personality profiles than to their actual selves. It is possible that the investigation focused on the wrong behaviors, and that there are indeed many significant correlations to be found between the complex strategies used in communication and the personality traits of the communicator. Perhaps these rich links can be found in the choice of words used, or the topics of conversation breached.

The level of intimacy among the IRC Experts, while also a function of disinhibition due to the anonymity of the medium, may also be directly due to the absence of nonverbal cues. In a typical FtF interaction, even superficial conversations are not devoid of personal meaning, as each participant can feel sure that they are speaking with an actual person. With online communication, however, one is not afforded that luxury. The nature of the medium enables the user to forget that there is a human being sitting in front of another monitor somewhere. The inclusion of personal information enables the individual to feel as though they are connecting to someone on a personal level, not merely reading words on a screen. A result of this compensation is that the conversations online become even more intimate than the average FtF interaction, such that the same dialogue, in person, would probably be considered too intense and overwhelming, or at least reserved for close friends.

Implications for Further Research. Further investigation into this arena should focus attention on creating settings more true to life. Many of the subjects in the present study, both in the IRC Novice and the FtF condition, remarked that it was strange to be asked to converse with complete strangers for an hour, either in an unfamiliar computer medium or in front of a video camera. As the examination of self-presentation and impression formation relies heavily on the everyday, perhaps unconscious processes of human interaction, an unforced, comfortable setting is ideal.

The present study revealed a strong difference in impression formation across conditions for the dimension of Extraversion only, which would suggest a more specific investigation into this personality trait would be fruitful. Furthermore, though the Big Five taxonomy has been shown to be a robust model, it is not all-inclusive. Therefore, there may be other aspects of personality that would be more relevant to this area of study. Future research should extend the present study to include other personality dimensions in the investigation of self-presentation and impression formation online. Further examination into this area may also prove to be helpful to members of minorities, or individuals with physical disabilities. Perhaps even the average individual who feels stifled by face-to-face interactions due to their own insecurities could benefit from this medium. Research in this area would shed light on the impressions formed online, and may enable such individuals to voice their opinions in a safer environment.

Many people fear that computers will soon take over every aspect of our lives, while others view this prospect as incredibly exciting and valuable. This trepidation concerning the growth of the computer world is evidenced by the confusion felt by the IRC Novice subjects. The Expert condition, however, proved that one can indeed grow to understand and fully utilize the various resources available, and become quite skilled. This new technology may change the tone of the business lunch, the college class, or conversations with family and friends around the world. Any knowledge we can gain about the culture and personality of the Internet is a valuable tool in understanding the course of its rapid and expansive growth.

References

Baym, N. K. (1995). The emergence of community in computer-mediated communication. In Jones, S. G. (Ed.), CyberSociety: Computer-Mediated Communication and Community (pp. 138-163). Thousand Oaks, CA: Sage Publications.

Computational Mechanics, Inc. (1996). IRC Usage Report v. 2.0b. Available: ftp://ftp.comco.com/pub/irc/stats

DePaulo, B. M. (1992). Nonverbal behavior and self-presentation. Psychological Bulletin, 111, 203-243.

Forgas, J. P. (1985). Interpersonal Behavior: The Psychology of Social Interaction. Sydney, Australia: Pergamon Press.

Fuller, R. (1994). Human-computer-human interaction: How computers affect interpersonal communication. Arachnet Electronic Journal on Virtual Culture [On-line]. Available: ftp://ftp.lib.ncsu.edu/pub/stacks/aejvc/aejvc-v2n02-fuller-human computer-human

Goffman, E. (1959). The Presentation of Self in Everyday Life. New York: Anchor Books.

Reid, E. (1991). Electropolis: Communication and community on Internet Relay Chat [On-line]. Available: http://www.ee.mu.oz.au/papers/emr/electropolis.txt

Reid, E. (1995). Virtual worlds: Culture and imagination. In Jones, S. G (Ed.), CyberSociety: Computer-Mediated Communication and Community (pp. 164-183). Thousand Oaks, CA: Sage Publications.

Serpentelli, J. (1992). Conversational structure and personality correlates of electronic communication [On-line]. Available: ftp://ftp.parc.xerox.com/pub/MOO/papers/conv-structure.txt

Turkle, S. (1995). Life on the Screen: Identity in the Age of the Internet. New York: Simon and Schuster.

Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational perspective. Communication Research, 19, 52-90.

Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal and hyperpersonal interaction. Communication Research, 23, 3-43.