JCMC 9 (3) April 2004
Collab-U  CMC Play  E-Commerce Symposium Net Law  InfoSpaces Usenet
  NetStudy VEs VOs O- Journ HigherEd Conversation Cyberspace Web Commerce
Vol. 6 No. 1  Vol. 6 No. 2  Vol. 6 No. 3  Vol. 6 No. 4  Vol. 7 No. 1  Vol. 7 No. 2  Vol. 7 No. 3
 Vol. 7 No. 4  Vol. 8 No. 1  Vol. 8 No. 2   Vol. 8 No. 3  Vol. 8 No. 4   Vol. 9 No. 1   Vol. 9 No. 2

Getting to Know You: Exploring the Development of Relational Intimacy in Computer-mediated Communication

Lee Bee Hian
Asia Insight Research Consultants Pte Ltd.

Sim Li Chuan
Ministry of Community Development and Sports, Singapore

Tan Mon Kiat Trevor
TODAY, MediaCorp Press Ltd.

Benjamin H. Detenber
Nanyang Technological University, Singapore


Abstract

This study investigates the development of relational intimacy in computer-mediated communication (CMC) by comparing it to face-to-face (FTF) interaction in a laboratory experiment. It also examines the influence of task type, gender, and group composition. Using a mixed-model factorial design, zero-history participants (n = 48) were grouped into dyads and asked to cooperate on low- and high-equivocality tasks over a series of three meetings conducted on three consecutive days. Results indicate that relational intimacy increased over time, but a significant interaction with medium and subsequent analyses reveal that this was only true for the CMC condition. Task type also interacted with time, but it did not have a significant effect on relational intimacy, nor did it interact with medium. Similarly, there was no main effect for either participant gender or group composition, nor did these factors interact with medium. The findings provide some support for the hyperpersonal communication model (Walther, 1996), which predicts that, under certain circumstances, interpersonal relationships can develop in CMC to a greater extent than they can in FTF interactions.


Introduction

Computer-mediated communication (CMC) is a form of interaction that has become extremely prevalent in many people's social lives. Despite technological advancements in networking and computing systems, CMC is still primarily text- based. Therefore, it is considered a limited medium in terms of information exchange, and unsuitable for carrying out tasks or social functions that require rich, detailed and nuanced communication. In this technologically deterministic view, CMC's usefulness as a communication medium is circumscribed by its technology (see Soukop, 2000; Spears & Lea, 1994; Walther, 1996).

However, challenges have emerged to counter this technologically deterministic perspective. Walther and Burgoon (1992) questioned whether media shaped communication or were used by communicators to suit their own needs. Evidence now exists indicating CMC-related modes of interaction such as instant messaging and computer-based group support systems do cater to interpersonal gratification needs (Leung, 2001) and reduce relational distance among users (Chidambaram, 1996). In fact, Walther (1996) proposed the concept of hyperpersonal communication, hypothesizing that online relationships sometimes create more social and intimate experiences than offline ones. However, as Walther (1996) himself and other researchers have noted, the hyperpersonal communication model fails to take into account context variables such as the nature of tasks communicators are engaged in or the communicators' group composition (e.g., Hancock & Dunham, 2001). In this study, we aimed to achieve a better understanding of CMC by considering theoretically relevant context variables that have not been explored in prior research. In this way, we hope to test and extend the hyperpersonal communication model (Walther, 1996).


Literature Review

CMC has usually been defined through the use of examples such as email, Internet relay chat, instant messaging and bulletin boards (c.f., Soukup, 2000; Walther, 1992, 1996). For the purposes of this study, CMC is defined as communication carried out through the use of networked personal computers. There are two perspectives on the relationship between CMC and face-to-face (FTF) communication. One is the technologically deterministic view which holds that, unlike FTF interaction, the limitations that characterize CMC as a technology restrict its utility as a channel of rich and varied social information. The other perspective has emerged from the notion that verbal and textual content can convey rich social information just as FTF interaction does (Walther & Burgoon, 1992).

CMC: From a Lean Medium to Hyperpersonal Communication

The technologically deterministic view of CMC arose from early research, most of which involved laboratory studies or experiments where small groups worked on structured problems for limited periods of time. Results from these early studies emphasized the social disadvantages of CMC, with the probability of highly developed personal relationships found to be relatively low online (Parks & Floyd, 1996). These findings were generally explained by the "cues filtered-out" (CFO) perspective, an umbrella term for several related theories, which posit that the lack of non-verbal cues in CMC causes it to be more impersonal than FTF interaction. For example, in social presence theory (Short, Williams & Christie, 1976), CMC is thought to convey little social immediacy or relevance compared to FTF communication due to the former's inability to transmit non-verbal cues. CMC is thus more impersonal. Media richness theory (Daft & Lengel, 1986) also focuses on CMC's predominantly lexical mode of interaction, deeming it a lean medium compared to FTF interaction, which has multiple cues and a high degree of personalization.

More recently, due to the burgeoning use of the Internet for social purposes, anecdotes of online encounters have shown that people can have intimate relationships in the CMC environment. Theories have emerged that attempt to address these mediated relationships, and they typically exchange the technological determinism inherent in the CFO perspective for a more nuanced social outlook. One of these, the social identification/deindividuation (SIDE) model (Spears & Lea, 1994), acknowledges the lack of cues in CMC but shifts the focus to social identity variables, such as in-group/out-group characterization, that are transmitted through and frame CMC interactions (Hancock & Dunham, 2001). The SIDE model also focuses on the cognitive processes by which communicators make inferences about others on the basis of minimal information (Lea & Spears, 1991, 1995; Spears & Lea, 1994). These generalizations and over-attributions may be drawn from stereotypes, especially over the Internet, where the slightest cue about the other's group identity may be seized upon and used as information for categorization and in-group/out-group definition (Spears & Lea, 1994). Thus, despite the lack of non-verbal social context cues afforded by CMC, there is paradoxically a more intense personal impression of one's communicative partners, based on the few cues available.

Another theory that challenges the CFO perspective is the social information processing (SIP) theory (Walther, 1992), which asserts that all communicators experience similar needs for uncertainty reduction and affinity, regardless of medium. To meet these needs, CMC users, like communicators using other media, will solicit and present socially revealing relational behavior (Walther, 1994). The difference, then, between CMC and FTF is not the ability to engender social interactions, but the rate at which social information exchange takes place. That is, given time, as interaction accumulates, relationships develop the same way in CMC as in FTF interactions.

Recently, another theory has been articulated that is an extension of both the SIDE and SIP perspectives. Walther's (1996) hyperpersonal communication model introduces factors that explain how the CMC environment can allow the individual to experience a level of closeness above the norm in FTF condition. Walther describes three necessary conditions for hyperpersonal communication to occur: 1) the receiver's idealization of the other due to over-attributions, whereby the receiver assigns magnified positive values to his or her partners; 2) sender's selective self-presentation, in which the sender has the advantage of being able to optimally edit his message before transmitting; 3) feedback loop or reciprocity of interactions, whereby the interplay of idealization and self- presentation becomes a dynamic process and creates a self-reinforcing cycle. If these conditions are met, people can develop a sense of closeness and rapport in their CMC interactions.

The hyperpersonal communication model has found support in a number of studies. Walther and Burgoon (1992) found that although CMC is slower in terms of rate of interaction than FTF, relationships developed in the same way in the two media as interaction accumulated. This finding was echoed in a study using original measures of impression development (Walther, 1993, cited in Walther, 1996) where it was shown that although CMC communicators' interpersonal impressions are less developed initially in the first meeting, the impressions approximated to the level of FTF by the third meeting. A subsequent experimental study by Walther (1995) even found that CMC interactions already exceeded FTF interactions with respect to relational intimacy after the first meeting.

In most studies, hyperpersonal communication has been examined by measuring relational intimacy (Walther, 1995, 1997; Walther & Burgoon, 1992; Walther, Slovacek & Tidwell, 2001). The concept suggests more than mere affiliation. Rather, it reflects a kind of closeness and reciprocity that is highly valued in personal relationships. Changes in relational intimacy can be personally significant and can be considered to be a reasonable index of relational development in both CMC and FTF interactions. Prager (1995) offers a three- tiered definition of intimacy as a kind of interaction in which partners (1) self- disclose; (2) feel good about themselves, the other and the interaction they share; and (3) perceive the interaction as having improved and progressed, or reflecting the partners' understanding of each other. For Burgoon and Hale (1987) the concept of relational intimacy is comprised of three non-independent factors: immediacy, similarity and receptivity, and it is from their operationalization that most current measures are derived, including those used in this study.

An important point to note is that the hyperpersonal communication model describes changes over time and the development of relationships. Studies that examine single interactions are likely to find a preference for FTF over CMC. Similarly, studies that place time limits on interactions (Cappel & Windsor, 2000; Kahai & Cooper, 1999; Warkentin, 1997) yield distinctly different results from studies without time limits (Walther & Burgoon, 1992; Walther, 1993; Walther, 1994; Walther et al., 1994; Walther, 1995; Walther, 1997). This point is underscored by a meta-analysis of CMC studies that found the degree of socio-emotional communication is greater in CMC when interaction time is not restricted (Walther, Anderson, & Park, 1994). Looking at the development of CMC-based relationships across geographical boundaries, Walther (1997) found that long- term group members experienced the greatest affection/social attractiveness from their partners, while short-term group members experienced the least affection/social attractiveness. Clearly, time is a critical variable in the development of relationships, regardless of the mode of interaction. In fact, Walther (1995) proposed that time may actually be a confound in early CMC experiments because time has been found to interact with medium. Due to CMC-users' reliance on typing to communicate, they typically reduce the rate at which they interact as compared to FTF communicators. Thus, experiments that assume comparability of CMC and FTF interaction in a single, time-limited session are inherently shortchanging the CMC communicators.

That is not to say that CMC communicators' relationships invariably gain intimacy as they interact more, or that different CMC relationships develop at the same rate. In a series of studies featuring distributed groups of students working on tasks exclusively over CMC, Walther, Boos and Jonas (2002) found that prior experience with the CMC technology affected the way students rated their working partners in terms of impression development, relational immediacy/affection, and social attraction. Students who appreciated the challenges of working over CMC rated their partners better and had better working relationships with them, compared to students who did not have prior experience working collaboratively via CMC.

Other medium-related factors might also influence the different rates at which relationships develop in CMC and FTF interaction. For example, Tidwell and Walther (2002) found that CMC-users adapted to the medium by modifying their uncertainty reduction behaviors. Specifically, "CMC partners forgo the peripheral questions and answers that mark the normal, superficial exchanges among new acquaintances in FTF encounters" (Tidwell & Walther, 2002, p.338). A study by Horman, Bordia, Irmer and Chang (2002) found that CMC groups exhibited more process and relationship conflict than FTF groups in their first interaction, but this difference disappeared in subsequent interactions. These behavioral changes in CMC are also likely to contribute to differences in relational development in CMC versus FTF interactions.

Based on prior research we believe that the FTF condition will lead to higher relational intimacy than CMC after the initial meeting. As the number of meetings accumulate, relational intimacy between communicators in both channels increases. According to the hyperpersonal communication model (Walther, 1996), CMC lets communicators optimally self- edit and present themselves, encouraging reciprocation of an idealized other, so intimacy between communicators in CMC should increase at a higher rate than, and eventually overtake, communication in FTF conditions. Hence we posit the following hypotheses:

H1a: Relational intimacy is lower in the CMC condition than in the FTF condition after the initial meeting (i.e., meeting 1).

H1b: Relational intimacy increases at a higher rate in the CMC condition than in the FTF condition; that is, the increase in level of relational intimacy between meeting 1 and meeting 3 in the CMC condition is greater than that in the FTF condition.

Task Equivocality

Although Walther criticizes the technological determinism inherent in the CFO approach, his model of hyperpersonal communication does not appear to take into account context variables, such as situation and user motivation. Such contextual variables might have played an important role in undermining the technological determinism of the CFO approach. Thus, in addition to testing Walther's hyperpersonal communication model, this study also aimed to explore its limitations by introducing a situational variable: task equivocality.

In their media richness theory, Daft and Lengel proposed that a rich medium, such as FTF communication, "facilitates equivocality reduction by enabling [individuals] to overcome different frames of reference" (1986, p. 560), and conversely, a medium that is less rich would be more suitable for less equivocal tasks. Task equivocality is distinguished from task uncertainty, in that equivocality refers to " ambiguity, the existence of multiple, conflicting interpretations" (p. 556), whereas uncertainty refers to "lack of information." Thus, FTF communication is the medium of choice for high-equivocality (HE) tasks, while leaner media such as CMC are more suitable for low-equivocality (LE) tasks because "rich media provided communicators with too much information" (Dennis & Kinney, 1998, p. 259). When the medium-task fit is right, that is, when HE tasks are resolved through rich media and LE tasks through lean media, task performance is better ( Daft & Lengel, 1986) and task satisfaction should also be higher (Dennis & Kinney, 1998).

Despite its apparent importance, the phenomenon of improved performance when medium is matched to task equivocality has received little empirical support (Dennis & Kinney, 1998; Rice, 1992). In fact, studies have found that email, traditionally defined as a lean medium in media richness theory, can be a relatively rich channel of information (c.f., Contractor & Eisenberg, 1990; Fulk, Schmitz & Steinfeld, 1990). These findings seem to complement results from Walther's studies (1992, 1994, 1996, 1997), but it should be noted that the studies were conducted among employees already embedded within stable organizations. In other words, familiarity among participants was likely in previous media richness theory studies and may have acted as a confound. One could argue that the lack of familiarity between zero-history CMC communicators would hinder their use of the medium to tackle HE tasks; that is, lack of familiarity would hinder the use of CMC as a rich, hyperpersonal medium. So, with the assumption that media richness theory would hold for zero-history communicators, we came up with the following hypotheses:

H2a: There is an interaction of task equivocality and medium on relational intimacy of zero-history communicators in a dyadic interaction such that, for the FTF condition, level of relational intimacy is higher for the HE task condition than for the LE task condition; and for the CMC condition, level of relational intimacy is higher for the LE task condition than for the HE task condition.

In addition, taking into account the deleterious effect HE tasks may have on the hyperpersonal nature of CMC, we hypothesize that:

H2b: For the CMC condition, level of relational intimacy develops more slowly for the HE task condition than for the LE task condition; that is, the increase in level of relational intimacy between meeting 1 and meeting 3 is smaller in the HE task condition than that in the LE task condition.

When conditions are right in CMC, hyperpersonal communication - by definition an intimate, emotion-rich experience - can occur in a supposedly "cues filtered-out" situation. By incorporating task equivocality into our study, we can assess whether it has a differential impact on the rapid relational development that characterizes hyperpersonal communication.

Gender and Group Composition

Gender may also be a factor in CMC relational development, for it is likely to play a role in the idealization of the other and selective self-presentation, which contributes to the heightened sense of closeness between partners. Gender differences in communication behaviors and goals are well documented. For example, Tannen (1990) noted that women tend to focus on reaching consensus while men focus on establishing status. Furthermore, men "tend to stress competition, maintaining some distance from others, operating by rules, interacting in groups, and doing," whereas women are inclined towards "cooperation, making and sustaining different relationships, [and] participating in close, dyadic friendships" (Wood, 1993, p. 34). Such gender differences, which Walther (1996) did not touch upon, may be a factor in the hyperpersonal communication model.

In fact, gender differences featured prominently in past CMC studies, most of which have mapped medium differences by studying how different genders perform or communicate within each medium. For example, communication style has been studied because it qualifies as an information cue used in interpersonal evaluation, which is an ongoing process in relationships (Adkins, 1995). Jaffe, Lee, Huang and Oshagan (1999) theorized that males and females have separate and distinct styles of communication that are attributable to learned social norms. Generally, males tend to be more aggressive, assertive of power, active and challenging when asking questions whereas females were found to be more accommodating, affectionate, passive and more dependent on non-verbal cues (Aune, Buller & Aune, 1996; Kaplan & Farrell, 1994; Rosenthal & Christensen, 1982; Soukup, 1999).

Research on communication between members of both genders in the CMC and FTF environment has yielded discrepant views with respect to the difference in the interactional processes that occur in the two environments. Some believe the presence of an equalization effect online, due to the absence of gender-specific cues, figures largely (Herring, 1993; Weisband, 1992, cited in Spears & Lea, 1994; Kiesler, Seigel & McGuire, 1984, cited in Dennis & Kinney, 1999). As Jaffe et al. (1999) put it, CMC has been described as "democratizing" because it neutralizes one's social status cues. However, there are others who regard CMC to be very much like FTF interactions because the lack of unique gender norms in CMC causes people to fall back on the stereotypes and gender norms that they rely on in the FTF environment (Matheson, 1992; We, 1994). Meanwhile, research in the FTF environment has shown that males are more able to get their communication partners to behave in a way that confirm the males' expectations (Snyder, Tanke & Bersheid, 1977, cited in Walther, 1997) whereas females were better at fulfilling expectations by reacting with confirming behaviors (Rosenthal & Christensen, 1982).

Taking these gender differences across media into account, and considering hyperpersonal communication, questions emerge about how members of each gender would behave when in the process of shaping a positive impression of the other and also selectively presenting themselves in a positive light. In addition, we can inquire about the differences that show up in the comparison between CMC and FTF interaction, in terms of the development of relational intimacy. Thus, we seek to answer the following research question:

RQ1. Does subject gender interact with medium in the development of relational intimacy, and if so, how do these variables interact?

Research on communication between members of both genders in the online environment has largely been focused on communication patterns that are observed in groups, rather than in dyads. The contexts of such research have typically been multi-user dungeons (MUDs), Internet Relay Chat (IRC) chat-rooms and group email correspondence (e.g., Cappel & Windsor, 2000; Flanagin, Tiyaamornwong, O'Connor, & Seibold, 2002; We, 1994). These studies typically find that CMC is influenced strongly by the gender of the person/persons with whom the individual is communicating. That is, while same-gender groups display the communication style of the common gender, mixed-gender groups vary across the continuum in terms of level of satisfaction and group development (Savicki & Kelley, 2000). Given these findings, we believe it is worthwhile to study the effects of group composition in a dyad, especially given the preponderance of group-based CMC studies, and relative lack of dyadic CMC research. The value of dyadic CMC research becomes even greater when one considers the growing popularity of online instant messaging software (e.g. MSN messenger or Mirabilis ICQ), which are typically one-to-one communication media.

Based on the above, we have reasons to believe that group composition in a dyad is important in determining the kind of interaction that takes place. Likewise, it would be interesting to find out if medium interacts with group composition to affect the level of relational intimacy between members in a dyad. Thus we pose the following research question:

RQ2. Does group composition interact with medium in the development of relational intimacy? If so, how do these variables interact?


Method

Design

The study had a 2 (CMC vs. FTF) X 3 (meetings) X 2 (high vs. low task equivocality) mixed-model design, with meeting being a within-subject factor and medium and task type manipulated between subjects. Counterbalancing gender and group composition across conditions, the participants were divided into 16 same-gender pairs (8 male-male; 8 female-female) and 8 mixed-gender pairs (see Table 1).

Computer-mediated Communication
Face-to-face Communication
High-equivocality task
Low-equivocality task
High-equivocality task
Low-equivocality task
MM
FF
MF
MM
FF
MF
MM
FF
MF
MM
FF
MF
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs
2 pairs

Table 1. Experiment group assignment


Participants

A total of 48 undergraduates (24 females) with a mean age of 22.2 years old from a local university were recruited via leaflets delivered to campus dormitories and bus stops. Participants were offered S$50 as an incentive for completing the three sessions of the experiment. Before the first meeting, the participants had no prior interaction with their partners.

Tasks

Six collaborative activities were selected from various sources (Kirby, 1992; Kirby, 1992; Morris, 1998; Savicki & Kelley, 2000) to represent HE and LE tasks (three of each). According to Daft & Lengel (1986), a task high on equivocality is ambiguous in that different individuals may understand instructions differently. Characteristically, HE tasks do not have a single right answer but a multitude of possible answers or solutions. On the other hand, LE tasks are of a less disputable nature and typically have a single answer or solution. An independent sample of 45 undergraduates provided a manipulation check of the task type variable. Their assessment of the activities yielded a significant difference between the task categories (LE and HE) for all three tasks (Task 1: t(43) = -5.01, p < .001; Task 2: t(43) = -3.81, p < .001; Task 3: t(43) = -8.59, p < .001)

Procedure

The experiment was carried out in three sessions over three consecutive days. Before the start of the first session, participants were briefed and instructed not to discuss the sessions with friends. The participants were reminded that there was no time limit to the tasks and to inform the moderator once they completed the task. Participants were randomly assigned to a group (either male-male, female-female, or male-female) and task type condition (HE or LE) and given a different task each day.

For the CMC condition, computer terminals connected via a local area network and equipped with Microsoft Windows 2000 and the instant-messaging software Mirabilis ICQ were used. CMC participants communicated with their partners using the "chat" mode of the instant-messaging software ICQ. Each pair was assigned computer terminals in two separate rooms. Care was taken to ensure that the participants in each pair did not meet each other throughout the experiment. For each session, the participants were given instruction sheets according to their assigned task condition. The participants were only told that they would be working with the same partner for all the three sessions; no additional information about their partners was given. Upon completing the task, the moderator was alerted and self-report questionnaires were given to the participants.

In the FTF condition, each pair was placed in a separate room and given the same stimuli as their CMC counterparts. Again, participants were only told that they would be working with the same partner for all the three sessions; no other information about their partners was given. When the task was completed, the moderator was alerted and self- report questionnaires were administered.

At the end of the third session, the participants were debriefed regarding the specific objective of the study as well as the funding source. They were also reminded not to discuss the experiment with anyone until the study was completed. Incentives were then paid to the participants.

As expected, sessions in the FTF condition took a shorter time to complete, as the median time for tasks to be completed is about 10 minutes. In contrast, sessions in the CMC condition took a little longer with the median time for tasks to be completed in CMC being about 30 minutes.

Measures and Data Analysis

The measurement of the dependent variable - relational intimacy - was based on an instrument initially developed by Burgoon and Hale (1987) and subsequently adapted by Walther and Burgoon (1992). Their 64-item scale was modified to suit the dyadic interactions in this study. In order to guard against response-set bias, the items were randomized so that the self-report questionnaire was different for all three meetings. Only 33 of the 64 items are directly related to relational intimacy, and although they can be further separated into immediacy/affection, similarity/depth and receptivity/trust, these components of relational intimacy are non-independent (Walther & Burgoon, 1992). As such, we decided to combine all of the items into a relational intimacy index for our analysis. The reliability of this 33-item composite measure was very high (Cronbach's alpha = 0.92).

The data were analyzed via repeated measures analysis of variance, with meeting as the within- subject factor and medium and task equivocality as between-subjects factors for tests of Hypotheses 1a, 1b, 2a and 2b. For the research questions on gender and group composition, separate analyses were run with medium, gender and group composition as between-subjects variables.


Results

Medium Effects

There was a marginally significant main effect for medium (F(1, 44) = 2.84, p < .10), such that relational intimacy was greater in the CMC condition than in the FTF condition (see Table 2 for means).

CMC
FTF
Meeting
M
SD
M
SD
Meeting 1
3.54
.49
3.53
.38
Meeting 2
3.69
.37
3.45
.37
Meeting 3
3.79
.40
3.56
.37
Overall
3.67
.42
3.51
.37

Table 2. Relational intimacy means for CMC and FTF groups by meeting


To test whether the relational intimacy in meeting 1 was lower for the CMC condition than for the FTF condition (Hypothesis 1a), an independent- samples t-test was run. The difference in relational intimacy at meeting 1 between CMC and FTF was not significant, t(46) = -.08, p = .94, so Hypothesis 1a was not supported.

The results also show that relational intimacy increased significantly with time (F(2, 88) = 3.56, p < .05). However, a marginally significant time x medium interaction (F(2, 88) = 3.05, p = .05) and examination of the means suggest that this was only true for the CMC condition ( see Figure 1).


Figure 1. Relational intimacy in FTF vs. CMC conditions by meeting


Hypothesis 1b postulated that relational intimacy would increase at a higher rate in the CMC condition that the FTF condition. A planned comparison indicates that there was a differential rate of development of relational intimacy between the CMC and FTF conditions (F(1, 44) = 6.99, p < .05). A comparison of relational intimacy levels between meeting 1 and meeting 3 confirms that there was significant increase in the CMC condition (t(23) = 3 .04, p < .05), but not for the FTF condition (t(23) = .39, p = .70).

Task Equivocality Effects

Overall, relational intimacy did not differ by task type. There was a significant interaction between task equivocality and time, however (F(2, 88) = 4.31, p < .05). Inspection of the means (see Table 3) and a significant linear contrast (F (1, 44) = 5.36, p < .05) indicates that relational intimacy only increased for the high equivocality task. A paired-samples t-test confirms that there was significantly greater relational intimacy after the third meeting than after the first for those in the HE condition (t(23) = 3.60, p < .01), but no difference for those in the LE condition (t < 1).

High-equivocality
Low-equivocality
Meeting
M
SD
M
SD
Meeting 1
3.50
.39
3.57
.47
Meeting 2
3.68
.27
3.46
.45
Meeting 3
3.76
.37
3.58
.41
Overall
3.65
.34
3.54
.44

Table 3. Relational intimacy means for high-equivocality and low-equivocality tasks by meeting



In terms of the task equivocality hypotheses, the analyses found no support for them. Hypothesis 2a postulated that there would be an interaction between task equivocality and medium. However, the ANOVA results show there was no significant interaction between task equivocality and medium (F < 1). As to whether relational intimacy increased at a higher rate for LE tasks than for HE tasks in the CMC condition (H2b), there was no significant interaction between task type and time in that condition (F(2, 44) = 1.67, p = .20). The linear contrast of the interaction also indicates that there was no significant difference in the development of relational intimacy between the two tasks (F(1,22) = 1.50, p = .23).

Gender and Group Composition Effects

Investigating RQ1, whether gender and medium interacted, the analyses indicated that relational intimacy did not vary significantly with gender (F < 1), nor was there a significant interaction between gender and medium (F < 1).

Our other research question (RQ2) addressed whether group composition had a differential impact on relational intimacy based on medium. The analyses indicate that relational intimacy did not differ by group composition ( F < 1), nor was there a significant interaction between gender and medium (F < 1). However, a marginally significant interaction effect was found for group composition and time (F(4, 80) = 2.34, p < .10), such that participants in the male-male groups showed no development in relational intimacy over time while participants in the other types of groups did (see Table 4).

Male-male
Female-female
Male-female
Meeting
M
SD
M
SD
M
SD
Meeting 1
3.71
.45
3.56
.37
3.33
.41
Meeting 2
3.55
.48
3.59
.34
3.57
.34
Meeting 3
3.73
.54
3.69
.25
3.59
.35
Overall
3.66
.49
3.61
.32
3.50
.37

Table 4. Relational intimacy means for group composition by meeting


Furthermore, there was a significant three-way interaction between group composition, medium and time (F(4, 80) = 3.08, p < .05). This significant finding seems to be the result of the male-male and female-male groups reporting higher levels of relational intimacy over time in the CMC condition while that of the female-female groups seems to stagnate (see Table 5).

Male-male
Female-female
 Male-female
Meeting
M
SD
M
SD
M
SD
Meeting 1
3.61
.51
3.62
.44
3.37
.54
Meeting 2
3.75
.39
3.80
.22
3.53
.44
Meeting 3
3.89
.52
3.72
.29
3.75
.38
Overall
3.75
.47
3.71
.32
3.55
.45

Table 5. Relational intimacy means for group composition by meeting in CMC condition


In contrast, the male-male groups in the FTF condition reported a significant drop in relational intimacy from the first meeting to the second, and although there seemed to be a bit of rebound after the second meeting, the overall trend was negative (see Table 6).

Male-male
Female-female
Male-female
Meeting
M
SD
M
SD
M
SD
Meeting 1
3.80
.39
3.50
.29
3.28
.27
Meeting 2
3.34
.50
3.38
.31
3.61
.45
Meeting 3
3.57
.55
3.67
.24
3.43
.25
Overall
3.57
.48
3.52
.28
3.44
0.33
Table 6. Relational intimacy means for group composition by meeting in FTF condition


On the other hand, the female-female and the male-female groups reported overall increases in intimacy over time, but differed from each other in their development patterns. That is, there was a drop in reported intimacy between the first meeting and the second for the women-only groups (similar to the men-only groups) before reaching the highest level after the final meeting. For the male-male groups the highest level of relational intimacy was reported after the second meeting.


Discussion

The findings of this study lend support to the hyperpersonal theory in that relational intimacy does increase at a faster rate in CMC than in FTF interactions. The number of meetings was found to be a strong predictor of the development of relational intimacy in CMC while task equivocality, gender and group composition had no significant main effects. The finding that relational intimacy did not vary by medium initially also supports a rejection of the technological deterministic perspective of CMC (i.e., CMC is inherently inferior in fostering intimate social interaction).

Hyperpersonal Communication

Results of the analysis support the hyperpersonal communication model, which posits that the idealization of one's partner and the ability to edit one's self-presentation make CMC more conducive to development of relational intimacy. After the first meeting, in which users of the two media were not significantly differentiated with regards to their relational intimacy, the rate of relational intimacy development was significantly higher for the CMC condition. On the other hand, it was observed that relational intimacy did not increase significantly in FTF interaction at all. It is possible that given more time relational intimacy in the FTF condition would increase as well.

A lower level of intimacy in the FTF scenario may be caused by the sense of awkwardness that is common even in a non-experimental environment when strangers meet for the first time in person. The participants' inhibitions could have been manifested, consciously or unconsciously, as nervous movements. Such movements could be characterized by fidgeting and lack of eye contact, both of which were frequently observed in the FTF settings in the present study. These manifestations of each other's inhibitions resulted in a tense atmosphere that was not conducive to building relational intimacy. In this case, the ability of the FTF channel to transmit non-verbal cues was actually detrimental to relational development. It seems the richness that might characterize FTF interactions can sometimes work against the development of relational intimacy.

Task Equivocality

Results indicating that task equivocality did not significantly affect relational intimacy and did not interact with medium can be interpreted as running counter to the technologically deterministic view, which posits that medium characteristics dictate how and what type of communication develops. In this case, the richness or leanness of medium failed to affect the medium's suitability to tasks of different equivocalities, and hence the medium's ability to encourage relational development. Thus, this study, like many others before it, points out the deficiencies in media richness theory, and as a whole, the rigidity of the CFO perspective.

By failing to consider the versatility of CMC, researchers may miss out on studying newly evolved practices in the CMC environment and thus lose the opportunity to re-investigate the relationship between CMC and FTF channels. For example, computer-mediated cues such as emoticons and paralanguage (Lea & Spears, 1992) have evolved from the use of CMC where symbols, generated by typing on keyboards, communicate feelings and emotions of users. Though superficially different from nonverbal cues in the FTF channel, both computer-mediated cues and non-verbal cues could be fulfilling the same intrinsic communication roles. In fact, there exists a potential that the use of these computer-mediated cues might expedite or replace, either fully or partially, the functions of traditional FTF cues. Such a possibility should be addressed and explored.

The results also indicate that regardless of medium, HE tasks engendered faster relational development than LE tasks. This is likely to stem from HE tasks' greater requirement for information exchange to disambiguate the activity's goals and processes. Beyond just instrumental information, it was noted that there was social information exchanged as well, and for participants doing the HE tasks this probably facilitated building relational intimacy at a faster rate.

Gender and Group Composition

Results indicate that males and females individually did not experience significant differences in the level of intimacy felt with their partners, nor did the genders differ across CMC and FTF conditions. The absence of observed gender differences may appear to be in contrast with findings from previous research. However, there are a number of possible explanations for the apparent lack of consistency with previous studies. First, this study did not examine motivations or communication behaviors, domains where gender differences are most profound. Second, the investigation of relational intimacy in this context and with these kinds of tasks is somewhat uncommon, so it is possible that the lack of gender differences was due to some idiosyncrasy of the experimental design, setting, or participants, or an interaction among these factors. For example, the laboratory setting could have suppressed gender differences by prompting a mood of cooperation. It is also possible that the lack of significant differences between genders may have been due to the use of zero -history participants. The interactions tended to start and end on a polite tone and this may have been a function of lack of familiarity. The absence of differences also could have been due to characteristics of the participants, English-speaking Asian university students, for gender differences in interpersonal relationships are known to exist across cultures (Watkins, et al., 1998). All of these explanations are speculative, however, and we want to exercise caution in any interpretation of null findings.

Similarly, group composition did not significantly affect relational intimacy, nor did it interact with medium. Post hoc analyses indicate that the marginally significant interaction between group composition and meeting was due solely to the difference between the high level of intimacy reported by members of the male-male groups compared to the mixed-gender groups after the first meeting. There were no other significant differences over time or across groups. It is often very difficult to make sense of higher-order interactions, and the pattern of means for the significant three-way interaction between group composition, time, and medium (see Tables 5 and 6) does not lend itself to easy interpretation. Post hoc analyses indicate that only a few of the pairwise comparisons of relational intimacy are truly different; it is the overall difference in patterns that is significant. Even so, one interpretation might be that the patterns in the CMC condition are more consistent and trending upward while those in the FTF condition are not really coherent or indicative of overall change. This might be a result of better impression management in CMC or selective self -presentation, which is one of the necessary conditions for hyperpersonal communication to occur. Unfortunately, the data seem somewhat equivocal on this point, so it will be left to future research to corroborate both the finding and the interpretation.

Limitations of the Study

The findings of this study may suffer from some limitations in its experimental design. First, individual tasks were not counterbalanced because the small sample size did not allow for a full permutation of tasks with the different conditions. Second, the sample size could not be as large as we would have liked due to time and resource constraints. Third, although the study design did not inhibit participants with respect to the amount of time they could use to interact for each of the three sessions, there was still a limit in terms of the number of sessions they could meet. Hence, we feel that observing a development of relational intimacy over a longer period would be a better way to profile the development of relational intimacy across both the CMC and FTF conditions.

In the study, we had arranged for CMC participants to be grouped into two groups, which were then installed in two different rooms such that partners did not see each other. However, we did not account for the time pressure that participants in individual rooms might have felt when they noticed others in the same room leaving upon task completion. This could have resulted in the remaining participants' hurrying through whatever was left of their task, which would constitute an effect of time limitation. Future research should avoid this confound by isolating each participant.

Researchers should also try to increase the ecological validity of their future studies. It is rather rare for CMC communicators to be interacting on only a one-to-one basis in real life, as they did in our study and other experiments. Multiple chat windows and interacting with a few friends on the contact list of the instant online messaging software are more common. As such, in order to increase ecological validity, researchers might want to conduct their future CMC studies by having participants use multiple chat windows in multiple dyad settings.

Implications

The study of relational intimacy development in CMC is relevant because it is related to other concepts such as productivity, task performance and the development of romantic relationships. According to Burgoon and Hale (1987), the factors of immediacy, similarity and receptivity constitute the concept of relational intimacy, which we have decided to adopt in this study. In the work environment, it would be advantageous for co-workers to feel positive about each other and be receptive to each other's opinions as relational intimacy can serve to enhance working relationships and pave the way for better cooperation and better work performance. Likewise, to begin or maintain a romantic relationship, partners need to feel affection, commonality and trust in one another. To then place the study of relational intimacy within the context of different media helps to differentiate what would be the optimal way of achieving relational intimacy. Such information can aid work organizations and dating services in their quest for building and maintaining relationships.

Previous studies and our results have shown that time is a major factor in the development of relationships in any medium. For example, Hollingshead, McGrath, and O'Connor (1993) commented on the importance of longitudinal research and strongly advocated that future research adopt a longitudinal approach to the study of the relational impact of communication technology. Thus, longitudinal studies comparing CMC and FTF in an unlimited- time setting are necessary to fully gauge the differences in the development of relationships in both media. It will be interesting to see if there is a ceiling effect, whereby the level of intimacy reaches a point of stagnation and cannot be improved anymore, especially in CMC.

While hyperpersonal communication was conceived in the context of CMC, not every relationship exists wholly online. Parks and Floyd (1996) found that about a third of the respondents in their study had used the telephone and FTF communication to contact their online friends. They also found that nearly two-thirds of their respondents with online personal relationships used communication channels other than the computer. In fact, Etzioni & Etzioni (1999) proposed a kind of hybrid system of FTF/CMC in which groups of people meet in person and form some measure of shared understanding, and then maintain bonding through CMC. Thus, in many cases, CMC may just be a starting point for personal relationships to develop before it is translated into the real life or vice versa. Future research might want to look into this area by looking at how participants' relationship starts first in one medium and then develops via a hybrid use of different media.

Furthermore, our study did not look into commonly shared visual codes such as emoticons that exist in CMC. Such codes may take the form of paralinguistic cues or keyboard representations, for example, a smile :-) or a frown :-( signifying happiness or sadness respectively (Lea & Spears, 1992; Walther & D'Addario, 2001). As Walther (1999) reiterated, "demystifying visual cues [in CMC], analyzing their functions, and recombining and re-representing them an element at a time, we might find that, like the value of well-chosen words in CMC, a little information can go a long way." Researchers have noted that certain textual behavior involving such codes in CMC actually accomplishes what nonverbal behaviors and cues typically perform in FTF communication (Walther & Tidwell, 1995). Future research concerning new categories of cues, paralanguage and an increased exploration of their functions vis-à-vis FTF interaction should contribute to better understanding of Walther's hyperpersonal communication model and CMC in general.


Acknowledgments

The first three student authors contributed equally to the study in collaboration with their faculty supervisor, the fourth author. This project was supported in part by a grant from the Social Development Unit of the Singapore Ministry of Community Development and Sports. A previous version of the paper was presented to the Communication & Technology Division of the International Communication Association at the annual conference in San Diego CA, USA May 2003.

References

Adkins, M. (1995). The power of language in computer-mediated groups. Management Communication Quarterly, 8(3), 289-318.

Aune, K.S., Buller, D.B. & Aune, R.K. (1996 ). Display rule development in romantic relationships: Emotion management and perceived appropriateness of emotions across relationship stages. Human Communication Research, 23(1), 115-145.

Burgoon, J. K. & Hale, J. L. (1987). Validation and measurement of the fundamental themes of relational communication. Communication Monographs, 54(1), 19-41.

Cappel, J. & Windsor, J. (2000). Ethical decision-making: A comparison of computer-supported and face-to-face group. Journal of Business Ethics, 28(2), 95-107.

Chidambaram, L. (1996). Relational development in computer-supported groups. MIS Quarterly, 20(2), 143-165.

Christensen, D. & Rosenthal, R. (1982). Gender and nonverbal decoding skills as determinants of interpersonal expectancy effects, Journal of Personality and Social Psychology, 42(1), 75-87.

Contractor, N. S. & Eisenberg, E. M. (1990). Communication networks and new media in organizations. In J. Fulk & C. W. Steinfeld (Eds.), Organisations and communication technology (pp. 117-140). Newbury Park: Sage.

Daft, R. L. & Lengel, R. H. (1986). Organisational information requirements: Media richness and structural design. Management Science, 32(5), 554- 571.

Dennis, A. & Kinney, S. (1998). Testing media richness theory in the new media: The effects of cues, feedback and task equivocality. Information Systems Research, 9(3), 256-295.

Dennis, A. & Kinney, S. (1999). Gender differences in the effects of media richness, Small Group Research, 30(4), 405-437.

Etzioni, A. & Etzioni, O. (1999). Face-to-face and computer-mediated communities: A comparative analysis. The Information Society, 15(4), 241-249.

Flanagin, A. J., Tiyaamornwong, V., O'Connor, J. & Seibold, D.R. (2002). Computer- mediated group work: The interaction of member sex and anonymity. Communication Research, 29(1), 66-93.

Fulk, J., Schmitz, J., & Steinfeld, C. W. (1990). A social influence model of technology use. In J. Fulk & C. W. Steinfeld (Eds.), Organisations and communication technology (pp. 117-140). Newbury Park: Sage.

Hancock, J. T., & Dunham, P. J. (2001). Impression-formation in computer-mediated communication revisited. Communication Research, 28(3), 325-347.

Herring, S.(1993). Gender and democracy in computer-mediated communication. Electronic Journal of Communication, 3(2). Available: http://www.cios.org/www/ejc/v3 n293.htm

Herring, S. (1999). Interactional coherence in CMC. Journal of Computer-Mediated Communication, 4(4). Retrieved April 1, 2004 from http:// www.ascusc.org/jcmc/vol4/issue4/herring.html.

Hollingshead, A. B., McGrath, J. E., & O'Connor, K. M. (1993). Group task performance and communication technology: A longitudinal study of computer-mediated versus face-to-face work groups. Small Group Research, 24(3), 307-333.

Horman, E. V., Bordia, P., Irmer, B. & Chang, A. (2002). The expression of conflict in computer-mediated and face-to-face groups. Small Group Research, 33(4), 439-465.

Jacobson, D. (1999). Impression formation in cyberspace: Online expectations and offline experiences in text- based virtual communities. Journal of Computer-Mediated Communication, 5 (1). Retrieved April 1, 2004 from http://www.ascusc.org/jcmc/vol5/issue1/jacobson.html .

Jaffe, J.M., Lee, Y., Huang, L. & Oshagan, H. (1999). Gender identification, interdependence, and pseudonyms in CMC: Language patterns in an electronic conference. The Information Society, 15(4), 221-249.

Johnson, D. W., & Johnson, F. P. (1994). Joining together: Group theory and group skills. Needham Heights, MA: Allyn and Bacon.

Kahai, S. & Cooper, R. (1999). The effect of computer-mediated communication on agreement and acceptance. Journal of Management Systems, 16(1), 165-188.

Kirby, A. (1992). Games for trainers: Volume 1. Aldershot, England: Gower.

Kirby, A. (1992). Games for trainers: Volume 2. Aldershot, England: Gower.

Lea, M., & Spears, R. (1992). Paralanguage and social perception in computer-mediated communication. Journal of Organizational Computing, 2, 321-341.

Lea, M., & Spears, R. (1995). Love at first byte? Building personal relationships over computer networks. In J. T. Wood & S. Duck (Eds.), Under-studied relationships: Off the beaten track (pp. 197-233). Thousand Oaks, CA: Sage.

Leung, L. (2001). College student motives for chatting on ICQ. New Media & Society, 3(4), 483-500.

Matheson, K. (1991). Social cues in computer-mediated negotiations: Gender makes a difference. Computers in Human Behavior, 7, 137-145.

Morris, W. (undated). Post-nuclear war survival. Retrieved April 1, 2004 from http://www .ofcn.org/cyber.serv/academy/ace/soc/cecsst/cecsst024.html.

NetValue. (2002, March 10). Home Net users increasing in Asia. Retrieved April 1, 2004 from http://www.nua.com/surveys/index.cgi?f= VS&art_id=905357534&rel=true

Parks, M. R., & Floyd, R. (1996). Making friends in cyberspace. Journal of Communication, 46(1), 80-96.

Prager, K. J. (1995). The psychology of intimacy. New York: Guilford.

Rice, R. E. (1992). Task analyzability, use of new media, and effectiveness: A multi-site exploration of media richness. Organization Science, 3(4), 475-500.

Rubin, R.B., Palmgreen, P., & Sypher, H. E. (1994). Communication research measures: A sourcebook. New York: Guilford.

Savicki, V. & Kelley, M. (2000). Computer mediated communication: Gender and group communication. CyberPsychology & Behavior, 3(5), 817-826.

Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunication. London: John Wiley.

Soukup, C. (1999). The gendered interactional patterns of computer-mediated chatrooms: A critical ethnographic study. The Information Society, 15(3), 169-176.

Soukup, C. (2000). Building a theory of multi-media CMC: An analysis, critique and integration of computer-mediated communication theory and research. New Media & Society, 2(4), 407-425.

Spears, R., & Lea, M. (1994). Panacea or panopticon: The hidden power in computer-mediated communication. Communication Research, 21(4), 427-259.

Tannen, D. (1990). You just don't understand: Women and men in conversation. New York: Ballantine.

Tidwell, L. C., & Walther, J. B. (2002). Computer-mediated communication effects on disclosure, impressions and interpersonal evaluations: Getting to know one another a bit at a time. Human Communication Research, 28(3), 317-348.

Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational perspective. Communication Research, 19(1), 52-90.

Walther, J. B. (1994). Anticipated ongoing interaction versus channel effects on relational communication in computer-mediated interaction. Human Communication Research, 20(4), 473-501.

Walther, J. B. (1995). Relational aspects of computer-mediated communication: Experimental observations over time. Organization Science, 6(2), 186-203.

Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal and hyperpersonal interaction. Communication Research, 23(1), 3-43.

Walther, J. B. (1997). Group and interpersonal effects in international computer-mediated collaboration. Human Research Communication, 23( 3), 342-369.

Walther, J. B. (1999, May). Visual cues and computer- mediated communication: Don't look before you leap. Paper presented at the annual meeting of the International Communication Association. San Francisco, CA.

Walther, J. B., Anderson, J. F. & Park, D. W. (1994). Interpersonal effects in computer-mediated interaction: A meta-analysis of social and anti-social communication. Communication Research, 21(4), 460-487 .

Walther, J. B., Boos, M., & Jonas, K. J. (2002). Misattribution and attributional redirection in distributed virtual groups. Proceedings of the 35th Hawaii International Conference on System Sciences.

Walther, J. B. & Burgoon, J. K. (1992). Relational communication in computer-mediated interaction. Human Research Communication, 19(1), 50-88.

Walther, J. B., & D'Addario, K. P. (2001). The impacts of emoticons on message interpretation in computer-mediated communication. Social Science Computer Review, 19, 323-345.

Walther, J. B., Slovacek, C. L., & Tidwell, L. C. (2001). Is a picture worth a thousand words? Photographic images in long-term and short-term computer-mediated communication. Communication Research, 28(1), 105-134.

Warkentin, M. E., Sayeed, L., & Hightower, R. (1997). Virtual teams versus face-to-face teams: An exploratory study of a Web-based conference system. Decision Sciences, 28(4), 975-996.

Watkins, D., Akande, A., Fleming, J., Ismail, M., Lefner, K., Regmi, M., Watson, S., Yu, J., Adair, J., Cheng, C., Gerong, A., Mcinerney, D., Mpofu, E., Singh-Sengupta, S., & Wondimu, H. (1998). Cultural dimensions, gender, and the nature of self-concept: A fourteen-country study. International Journal of Psychology, 33(1), 17-31.

We, G. ( 1994). Cross-gender communication in cyberspace. Electronic Journal on Virtual Culture, 2(3). Retrieved April 1, 2004 from http://www.monash.edu.au/journals/ejvc/we.v2n3.

Wood, J. T. (1993). Engendered relations: Interaction, caring, power, and responsibility in intimacy. In Steve Duck (Ed.), Social context and relationships (pp. 26-54). Newbury Park: Sage.


About the Authors

Lee Bee Hian graduated with a bachelor's degree (Honors) from the School of Communication Studies, Nanyang Technological University. She is currently a Senior Research Executive at Asia Insight Research Consultants, a private commercial market research company located in Singapore.
Address: Address: 53 Jalan Binchang, Singapore 578546. Phone (65) 6258-4996.

Sim Li Chuan graduated from the School of Communication and Information with a bachelor's degree (Honors) in Communication Studies in 2002. He is currently working at the Ministry of Community Development and Sports, Singapore.
Address: 5 Sembong Road, Singapore 758330.

Tan Mon Kiat Trevor graduated with a bachelor's degree (Honors) from the School of Communication Studies, Nanyang Technological University. He is currently working as a photojournalist at TODAY, a local free daily newspaper in Singapore.
Address: MediaCorp Press Ltd, 24 Raffles Place #28-01/06, Clifford Centre, Singapore 048621. Phone (65) 6236-4835.

Benjamin H. Detenber (Ph.D., Stanford University) is an Associate Professor in the School of Communication and Information, Nanyang Technological University, Singapore, and Head of the Communication Research Division. His research interests focus on psychological processing and communication technologies, and cognitive and emotional responses to mediated presentations.
Address: School of Communication and Information, 31 Nanyang Link, Singapore 637718. Phone (65) 6790-5809, Fax ( 65) 6792-4329.

©Copyright 2004 Journal of Computer- Mediated Communication