Computers in Human Behavior 24 (2008) 2757–2770
Contents lists available at ScienceDirect
Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh
Relationship between the level of intimacy and lurking in online social network services Pei-Luen Patrick Rau, Qin Gao *, Yinan Ding Department of Industrial Engineering, Tsinghua University, Beijing 100084, China
a r t i c l e
i n f o
Article history: Available online 15 May 2008
Keywords: Computer mediated communication (CMC) Social networking Interpersonal relationship Intimacy
a b s t r a c t The rapid growth of online social network services (SNSs) leads to new research questions. Unlike in other online communities, people in SNSs expect to gratify social-emotional needs rather than informational needs, and they are connected in a person-to-person manner which is more direct and interpersonal. The author argued that the factors influencing members’ public posting in SNSs differ from those in traditional online communities. Interpersonal intimacy was postulated to influence lurking behaviors in SNSs. To investigate the relationship between intimacy level and posting frequency in SNSs, an online survey was conducted in Wallop, a SNS provided by Microsoft. Responses (102) were collected, in which the first 40 posters and the first 40 lurkers were selected for statistical analysis. The result shows significant differences exist in both verbal and affective intimacy level between lurkers and posters. The level of verbal intimacy and the level of affective intimacy are positively correlated with posting frequency. The result of discriminant analysis shows that verbal intimacy and affective intimacy are useful for discriminating posting/lurking groups of users. In addition, significant gender differences in perceived intimacy and posting behaviors were found. The result implies that people lurk in SNSs because they believe that their social-emotional needs may not be satisfied even if they post. Ó 2008 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +86 1062788750; fax: +86 1062794399. E-mail addresses:
[email protected] (P.-L.P. Rau),
[email protected],
[email protected] (Q. Gao). 0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chb.2008.04.001
2758
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
1. Introduction Online communities exist not only for task-orientated communications, but also for personal relevant information sharing, trust and intimacy creation, and social relationships building. During the past five years, explicit social networking sites (SNSs) have mushroomed together with the wave of Web 2.0 technologies. These systems are designed specifically to help people build online presences and building social networks. Users’ active and public participance is critical to the success of such services. Active posting is desired. However, the motivation and behavior of people using such environments are supposed to differ from that in traditional online communities, and reasons why they post or lurk may also differ. Previous studies found that people do not post in online communities since their informational needs can be satisfied without posting. However, people rarely use online social network services solely to seek information. Contrarily, social-emotional supports, friendship and intimate interactions are sought out in SNSs. In addition, people in SNSs are connected in a person-to-person manner, which is more direct and interpersonal than in other online communities. Therefore, the social ties and interpersonal relationships among members of SNSs are supposed to influence posting behavior of user in SNSs. Intimacy is an essential aspect of the social network tie and interpersonal relationship. It motivates people to seek warm, close, and validating experiences. This study is to understand the reasons behind lurking behaviors in SNSs with the aim to help the developer improve the design of such services and ultimately enhance user experience with such services. The author examined the relationship between the intimacy level in SNSs and the posting behavior in SNSs, with a hope to facilitate further exploration of user behavior in SNSs. 2. Literature review 2.1. Computer mediated communication and online community Considerable variability exists in the assumptions about the influences of computer mediated communication (CMC) on social interaction processes and interpersonal relationship developments. On the one hand, some researchers suggested that CMC was improper for complex, emotional, and equivocal communications, due to its lack of social-emotional cues (Barua, Chellappa, & Whinston, 1997), reduced social presence (Connell, Mendelsohn, Robins, & Canny, 2001; Short, Williams, & Christie, 1976) and lower media richness (Daft & Lengel, 1986; Daft, Lengel, & Trevino, 1987). Only illusions of real communities can be created in such an environment (Beniger, 1987). Other researchers, however, argued that some early findings depreciating CMC were based on inadequate field observation (Walther, 1992). The social information processing theory, proposed by Walther (1992), emphasized that people need to develop relationships even in the absence of visual and oral cues, and this need motivates people to adapt the textual cues for complex tasks such as forming interpersonal impressions and attainting psychological-level knowledge. People may even find experiences in CMC communication more desirable than face-to-face interactions because of selective self-presentation and the resulting idealized-perception of partners within CMC (so-called hyper-interaction). Some other researchers adopted the uses and gratifications approach (Blumer & Katz, 1974; Rosengren, Wenner, & Palmgreen, 1985) to analyze the use of new media and the impact of CMC. The uses and gratifications approach arose out of the functionalist perspective on media (Herzog, 1944). Essentially, it assumes that audience members of a medium actively use the medium to fulfill specific needs, and the gratification opportunities provided by the medium influence the usage of the medium. They suggested that CMC provides disparate gratification opportunities compared with traditional media, and could be a superior to richer media in the case that richer does not fit the needs or the constraints (Dimmick, Kline, & Stafford, 2000). 2.2. Online social network services The notion of social network, coined by Barnes (1954), denotes a description of the underlying patterns of social structures. Social network theory models persons as nodes of a graph and their relation-
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
2759
ships as ‘ties’ connecting the nodes (Wasserman & Faust, 1994; Watts & Strogatz, 1998). Two friends who are directly connected are one link away from one another; a friend of a friend is two links away. In this way, all relationships of individuals can be modeled as paths on the graph. It has been found by experiment that anyone can be connected to another one on the planet through a short chain of acquaintances, typically consisting of no more than five intermediaries (Dodds, Muhamad, & Watts, 2003; Milgram, 1967). Researchers have found that the Internet complements traditional communications and enhances traditional social ties (Birnie & Horvath, 2002; Katz & Aspden, 1997; Pew Internet and American Life Project, 2000; Green & Pearson, 2005). Although all communication applications have social networks embedded in their design, SNSs is the first application in which people explicitly articulate their social networks. Commercialized SNSs sites, like orkut, facebook, friendster, have incorporated numerous concepts in social network theory into their design, such as the rule of 150 and six degrees of separation. SNSs differ from other online communities in three major aspects. First, they are designed specifically to help people establish online presence and build social networks (Ahn, Han, Kwak, Moon, & Jeong, 2007), while the majority of traditional online forums or communities are built to ‘‘improve one’s understanding of the topic” (Preece, Nonnecke, & Andrews, 2004). Though some online communities are designed to exchange emotional supports, such exchange mainly occurs between strangers rather than acquaintance in real life. Consequently, the motivations of users for visiting and postings are different. Secondly, users in SNSs are connected in networks rather than in hierarchical groups as in traditional online communities. Each user is represented by a public profile which contains links to selected members. These links to other users make up the network through which information is distributed, while each user can define his/her community egocentrically. Wellman (1992) argued that human communities are developed in networks, not groups. Therefore SNSs are believed to reflect real-life relationships of people more accurately than traditional online communities and more innate to human cultural experience (Kiehne, 2004). Mayfield (2005) compared the structure of online communities and that of online SNSs, stating that online SNSs are bottom-up developed, people-centric, user-controlled, context-driven, decentralized, and self-organizing, whereas online communities are top-down developed, place-centric, moderator-controlled, topic-driven, centralized and architected. Thirdly, SNSs users are connected in a person-to-person manner and they must explicitly state their relationship with other people. These make the relationships among members in a SNS more visible, direct and interpersonal than the relationships among members in other online communities. In SNSs, connections come before contents, whereas in traditional online communities, content comes before connections (Mayfield, 2005). It is often assumed that people will more authentically define their identity on SNSs sites compared with on other online communities to ensure meaningful connections (Boyd, 2004). Moreover, most SNSs sites facilitate relationships creating and maintaining by providing tools for self-presentation (e.g., avatars and blog), for managing connections (e.g., buddylists), for keeping updated with acquaintances and for initiating new connections (e.g., self profile and introduction). 2.3. Lurking in online communities The value of online SNSs lies in that they help people to capitalize their social networks (Boyd, 2004). The form and the level of user participation is critical to the survival of SNSs. Users can participate in online communities either in a public manner (i.e., posting and commenting) or in a non-public manner (regularly reading but not posting, also called lurking). Public participation is often desired. Preece (2001) proposed that the number of posted messages per member or per active member indicates how engaged people are with the community and should be taken as a determinant to the success of online social software. However, a large portion of members in online communities are lurkers rather than active posters (Nonnecke & Preece, 2001). It was reported over 90% members of some large online groups such as MSN and AOL were lurkers (Katz, 1998). Understanding the reasons why people do not post publicly is important to the survival and the success to SNSs as to other online communities. Researchers often define the concept of lurking operationally, such as ‘‘no messages sent during a three month period” (Nonnecke & Preece, 2001), and ‘‘regular visits to the community but reticence or very seldom posting” (Rafaeli, Ravid, & Soroka, 2004). While Smith and Kollock criticized that lurkers
2760
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
were free-riders, i.e., non-contributing, resource-taking members (Smith & Kollock, 1999), other researchers argued that lurking is a normal behavior in the majority of the population (Nonnecke & Preece, 2000a). Lurking is a ‘‘vicarious learning” process, during which lurkers can benefit cognitively and socially from observing other’s learning (Lee & McKendree, 1999; McKendree, Stenning, Mayes, Lee, & Cox, 1998). A recent study (Preece et al., 2004) suggested that whether lurking is a problem depends on the community context. While in large and active communities, the existence of lurkers is desirable, communities where little or no message posting occurs cannot survive if lurking behavior becomes dominant in the community. To understand why this ‘‘silent majority” does not post, Nonnecke, Preece, and Andrews jointly conducted a series of research. They found that lurking is the result of a complex set of actions, rationales, and contexts. The reasons for lurking fell into various categories ranging from personal to work related reasons (Nonnecke & Preece, 2001). In later studies (Nonnecke, Preece, Andrews, & Voutour, 2004; Preece et al., 2004), it was found that lurking was a post joining developed habit rather than a conscious decision from the outset. The biggest reason for lurking was that the information needs of users could be satisfied without posting (53.9% of 1188 participants). This is consistent with the previously proposed gratification model for lurking (Nonnecke & Preece, 2001). Previous research also postulated various models to explain lurking (Nonnecke, 2000; Nonnecke & Preece, 2000b; Nonnecke & Preece, 2001), and suggested de-lurking strategies and improvements for elevating the lurker’s experience (Nonnecke et al., 2004; Schultz & Beach, 2004). However, these studies only investigated lurking behavior in traditional online communities, such as BBS, online forums, and e-mail/discussion lists. Lurking in SNSs communities remains unstudied. Since SNSs sites are designed to satisfy social and emotional needs, the reason for lurking in SNSs are supposedly related with interpersonal relationships and emotional closeness. 2.4. Interpersonal relationship and computer mediated communication The relationship between people in a social network can be described by social ties (Garton & Wellman, 1997; Wellman, Salaff, Dimitrova, Garton, & Gulia, 1996). Previous research found that tie strength of interpersonal relationships exerts a remarkable impact on media use (Haythornthwaite, 2002, 2005; Haythornthwaite & Wellman, 1998). Intimacy was agreed as both an important measure of tie strength (Granovetter, 1982; Marsden & Campbell, 1984) and an essential part of tie formation (Wellman, 2001). In previous research, intimacy was conceptualized as the quality of interactions and relationships (Laurenceau, Barrett, & Pietromonaco, 1998; Timmerman, 1991), or as enduring motivation to experience closeness, warmth and communication (Sullivan, 1953). Although different studies conceptualized intimacy differently, self/partner-disclosure and emotional support were widely accepted as central to the make-up of intimacy (Harper, Schaalje, & Sandberg, 2000; Hu & Smith, 2004; Moss & Schwebel, 1993; Tolstedt & Stokes, 1983). Intimacy was developed through a dynamic process whereby an individual discloses personal relevant and revealing information to a partner; receives a response from the partner; and interprets that response as understanding, validating, and caring (Reis & Shaver, 1988). In different studies, overt verbal exchange, such as self-disclosure and partner disclosure, also known as verbal intimacy, was identified as an essential condition which must be met for intimacy to occur (Kjeldskov et al., 2004; Timmerman, 1991), a significant component of the intimacy process at an interaction-by-interaction level (Laurenceau et al., 1998), and a key characteristic that differentiates intimate from non-intimate relationships (Moss & Schwebel 1993). It is often used as an operational measure of intimacy. Buhrmester and Furman (1987) and Carpenter and Freese (1979) even equate intimacy with self-disclosure. Emotional support, also called affective intimacy, reflects feelings of closeness and emotional bonding, involving intensity of liking, moral support, and the ability to tolerate flaws in the significant other (Tolstedt & Stokes, 1983). It reflects the depth of awareness intimates have of one another’s emotional world and the emotional exchanges they share, involving the perception the closeness (Sinclair & Dowdy, 2005). The level of affective closeness in friendships and romantic relationships is reported as related to the level of intimacy of that relationship (Levinger & Senn, 1967), and being able to better predict the relation satisfaction (Deenen, Gijs, & van, 1994).
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
2761
An interesting finding in previous research in this area is the significant gender differences in how intimacy is developed and perceived (Ahn et al., 2007; Hook, Gerstein, Detterich, & Gridley, 2003; McNelles & Connolly, 1999). Girls were reported more likely than the boys to establish intimacy through discussion and self-disclosure, and their relationships are more self-disclosing than those of males, whereas males’ relationships focus on shared activities and experiences and they are more likely to establish intimacy through shared activities (Barrett & Lally, 1999; Sussman & Tyson, 2000). There are also studies examining the gender difference in user behaviors and perceptions in online communities. Males are found doing the most of talking and attracting, and the articles are often lengthy, sarcastic, and self-promoting, while females post less, and their postings are often supportive, attenuating, and less opinionated (Berscheid, 1969; Sussman & Tyson, 2000). The dominance of male culture in the cyberspace has been recognized, and females were found less comfortable with the computer use and the online culture than males (Herring, 1994). These findings provide contradictory implications for the possible gender differences in perceived intimacy and posting behaviors in SNSs. Whereas female pattern of communication in the real life tend to be network-oriented (Tannen, 1991), which can be facilitated by the network structure of SNSs, females also suffer from higher computer anxiety and perceive less ease of use with computer-mediated communications (Berscheid, 1969; Frenkel, 1990; Gilroy & Desai, 1986), which may make them less easily enjoy online communication services. 3. Research framework and questions In information-seeking orientated communities, the biggest reason for lurking was reported as ‘‘just browsing/reading is enough” (Preece et al., 2004). This indicates that people lurk since they can satisfy their informational needs without explicitly posting in communities. However, people go to SNSs sites since they see the value of connecting to friends, and most of the writings/postings in SNSs are documenting the writer’s own life. While the informational content of such posts may be low and seemingly trivial to outsiders, the posting/commenting act itself can be laden with emotional significance for those involved (Vetere et al., 2005). The motivation to communicate with others and the desire for warm, close and validating experiences is supposed to affect posting behavior in SNSs. The more intimate a relationship, the more interactions are needed, and the more likely people are to adapt and expand their media use to support the exchanges they need (Haythornthwaite & Wellman, 1998). Therefore, the level of interpersonal intimacy is hypothesized to influence user posting behavior, as shown in Fig. 1.
Fig. 1. Research framework containing components of verbal intimacy and affective intimacy influencing posting in SNSs.
2762
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
3.1. Research question 1: does the level of verbal intimacy in a SNS have an influence on posting frequency in that SNS? Self-disclosure, or verbal intimacy, is an important ingredient of tie strength or intimacy (Granovetter, 1983; Marsden & Campbell, 1984). People sharing an intimate relationship communicate more frequently, discussing topics of a wider breadth, and the depth of self-disclosure is deep (Laurenceau et al., 1998; Marsden & Campbell, 1984). Previous research has shown that where ties are strong, communicators can influence each other to adapt and expand their use of media to support the exchanges important to their tie; those more strongly tied used more media to communicate than weak ties (Haythornthwaite, 2001, 2002, 2005; Haythornthwaite & Wellman, 1998). When users are connected to those with whom they share a high level of verbal intimacy in SNSs, they are more likely to disclose different aspects themselves in different depths, by writing a blog post, posting an article on semipublic/public BBSs, uploading a photo or just updating the message describing their current status on profile pages. On the contrary, when connections in a SNSs are of low verbal intimacy, topics users can discuss with their connections are narrow and shallow, and are often descriptive (telling the facts) rather than evaluative (revealing feelings). Consequently, the need and the possibility for self-disclosure are low, which leads to lurking behaviors. 3.2. Research question 2: does the level of affective intimacy in a SNS have an influence on posting frequency in that SNS? Affective intimacy refers to the presence of emotional closeness and affective feelings, as well as reciprocal support. Affective intimates share each others innermost emotional worlds and are more likely to exchange emotional supports (Tolstedt & Stokes, 1983). The risks associated with self-expression decrease when people are aware that they are loved and liked, and they are more willing to share their ideas and feelings (Berscheid, 1969). It is expected that people that feel affectively intimate with their connections in a SNS will post more because they seek more emotional supports from these connections and the possibility that they can receive satisfactory supports is high. On the other hand, when the level of affective intimacy between a SNSs user and his/her connections in this SNSs, the user will have lower motivation to post texts or photos about their own life and feelings, since people rarely confide in people who appear unconcerned or indifferent to them (Berscheid, 1969). In addition, the risk of being misunderstood by a less intimate acquaintance is high. This increases the apprehension of casually stated opinions being criticized and the meaning of equivocal sentences being distorted. Such fears of negative responses also hold people back from public posting (Nonnecke et al., 2004). 3.3. Research question 3: how lurkers and posters differ in terms of their verbal and affective intimacy levels? Based on the definition of lurking given by Nonnecke and Preece (2000a) and Rafaeli et al. (2004), lurking was defined in this study as less than 3 posts over a three-month period on the premise of visiting a site at least once a month on average. Both verbal and affective intimacy levels were expected to influence posting behaviors, and consequently influence whether a user becomes a lurker or a poster. Accordingly, this study was designed to investigate whether lurkers and posters differed in their intimacy levels and whether verbal and affective intimacy levels can discriminate lurkers from posters.
4. Methodology 4.1. Participants A web-based survey was conducted among users of Wallop (http://www.mywallop.com), a SNSs provided by Microsoft. Wallop was chosen because it is a feature-rich social network service with rich, emphasizing life sharing and conversation building in the context of social networking. The survey
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
2763
was administered over a three-week period, from the end of May to the middle of June in 2005. According to earlier experience, response rates from lurkers can be very low (Mason, 1999). Invitations (768) were sent altogether and 102 responses were collected. Although the response rate was not high, the result was satisfactory because: (1) the sampling methods resulted in 41.2% responses from lurkers and (2) the only incentive for participating was a promise that the study results would be sent to participants if they were interested. 4.2. Questionnaire design The questionnaire consisted of three sections. Section A contained measures of the posting frequency and lurking level in Wallop. Respondents were asked to estimate the amount of time per month they visited Wallop, the number of posts in Wallop during the past three months, and the average posting frequency (times/month). In addition, the average amount of time the user spent surfing the Web per week was asked. In Sections B (verbal intimacy) and C (affective intimacy), the respondents were told to describe the most intimate relationship they experienced in Wallop according to the questions. Questions were encoded into seven-point Likert scale, anchored by strongly disagree (1) and strongly agree (7). Verbal intimacy was measured by the depth and the breadth of interactions (Joinson, 2001; Parks & Floyd, 1996). The depth of intimacy referred to the extent to which importance, risky, and private information was revealed to friends; the breadth of intimacy related to the variety of topics, activities and communication channels Parks. The subscales for these two aspects adopted items from Parks and Floyd’s scale of development levels in on-line relationships (Parks & Floyd, 1996) and modified them to fit the current context and topic. Items B1–B4 addressed the breadth of interaction. Items B5–B11 addressed the depth of interaction. B1, B6, and B9 were reversely coded. Sample items for each subscale are, ‘‘Our communication ranges over a wide variety of topics.” (breadth) and ‘‘I feel I could confide in this person about almost anything.” (depth) Section C addressed affective intimacy, which refers to the feeling of closeness and emotional bonding (Hu & Smith, 2004). Items 1–16 were adopted from Parks and Floyd’s scale (Parks & Floyd, 1996), including subscales of interdependence (C1–C6, sample item: ‘‘We have a great deal of effect on each other.”), interpersonal predictability and understanding (C7–C11, sample item: ‘‘I can usually tell what this person is feeling inside.”), and commitment (C12–C16, sample item: ‘‘This relationship is a big part of who I am.”). Items 17–22 were six items selected from the Miller Social Intimacy Scale (MSIS) (Miller & Lefcourt, 1982), addressing the frequency of affective intimacy. The overall intimacy was measured by the sum of the verbal intimacy and the affective intimacy in this study. At the end of the questionnaire were five items of demographic information. Finally the respondents were told that they can receive a copy of the research result if they were interested and provided their e-mail address. To measure the internal consistency of the scales, Cronbach’s alpha were used. The resulting alphas were 0.73 for verbal intimacy and 0.89 for affective intimacy, respectively, indicating an acceptable level of internal consistency. The Cronbach’s alpha for overall intimacy was 0.9, showing a good level of internal consistency. 4.3. Data collection A pilot test was conducted among six graduate students to ensure that the questions were unambiguous and that there were no technical errors that would impede data collection. A small number of questions were amended after the pilot test to improve the clarity. Because SNSs are characterized by an aggregation of ego-centered networks and there are no public boards or administrative mediators, the snowball sampling method was used. The snowball sampling method is considered more efficient in crawling the web and it can capture the network typologies better than node sampling and link sampling (Ahn et al., 2007; Kwak, Han, Ahn, Moon, & Jeong, 2006). Starting from the node of ‘Wallop Team’ which had a large number of connections, we sent invitation letters to people who were directly connected to the Wallop Team. We called these people as users on the first layer. Invitation letters were sent to participants through the messaging function of
2764
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770
the system. More invitation letters were then sent to people who were directly connected to users on the first layer (these people are called users on the second layer). The process continued in this way and each step was recorded carefully to make sure that the same person would not be invited twice. To balance the lurker and poster response rate, more invitation letters were sent to those whose name bars were semi-transparent (indicating a low level of posting frequency in Wallop). 4.4. Data analysis Descriptive statistics were calculated for Internet use, posting frequency, and each item in the verbal intimacy scale (section B) and in the affective intimacy scale (section C). t-Tests were conducted to evaluate the differences between two groups: lurkers and posters. The Pearson correlation coefficients were computed and the significance of the coefficients was tested to determine whether the level of verbal and affective intimacy had a positive influence on posting frequency in Wallop. Then a twogroup discriminant analysis was used to determine how lurkers and posters differ based upon their intimacy levels. The independent variables were the level of verbal intimacy and affective intimacy, and the dependent variable was group membership (lurkers or posters), a nominal scale variable. Equal priorities were assigned to the two predicators. In addition, the influence of gender was examined. 5. Results 5.1. Descriptive statistics Among the 102 respondents, 100% visited Wallop at least once a month. Thus, all of them satisfied the premise of being regarded as lurkers. The first 40 lurkers and the first 40 posters were identified as two groups for statistical analysis. 48 were males, and 34 were females, aging from 18 to 49 (mean = 24.1, SD = 5.0). Sixty four out of 80 were students, among which 46 were undergraduate students. Almost all of the respondents had a college or above level education. The average amount of computer use was 34.3 hours/week (SD = 23.1). The sample represents a highly educated population with abundant Web experiences. As shown in Table 1, the lurker group included 19 males (47.5%) and 21 females (52.5%), aging from 20 to 49 (mean = 23.5, SD = 4.4). The poster group included 28 males (70%) and 12 females (30%), aging from 18 to 44 (mean = 24.8, SD = 5.4). there was no significant differences in age (Wilcoxon two sample test: Z = 1.40, p = 0.16) and time spent on Web (Wilcoxon two sample test: Z = 1.44, p = 0.15) between lurkers and posters, and but lurkers posted significantly less frequently than posters (Wilcoxon two sample test: Z = 6.89, p < 0.001). Table 2 shows the descriptive statistics for verbal intimacy; Tables 3 and 4 show that of affective intimacy. Posters rated higher for each intimacy item, both verbal and affective, than did lurkers. To test the significance of such differences, t-tests were used. Where variances were not homogeneous (B3, B8, C8, C12, and C18), Satterthwaite t-tests were used. The result showed that all the differences were significant. Posters in Wallop have a significantly higher level of verbal intimacy in their social networks in Wallop (mean = 54.30, SD = 8.58) than lurkers do (mean = 34.03, SD = 8.10, p < 0.001), and a significant higher level of affective intimacy (mean = 112.1500, SD = 15.23) in their social networks in Wallop than lurkers do (mean = 69.23, SD = 16.56, p < 0.001). In addition, the difference in overall intimacy was examined, showing that post-
Table 1 Descriptive statistics of age, gender, time spent on Web and posting frequency for posters and lurkers Age (year)
Lurkers Posters
Time on Web (hours/week)
Posting frequency (time/month)
Mean
SD
Gender Female
Male
Mean
SD
Mean
SD
23.5 24.8
4.4 5.4
21 12
19 28
28.8 39.8
18.7 28.7
1.3 12.8
1.23 16.3
2765
P.-L.P. Rau et al. / Computers in Human Behavior 24 (2008) 2757–2770 Table 2 Verbal intimacy statistics for posters and lurkers Items
Group
Mean
SD
tValue
p
Difference
Verbal intimacy (total)
Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers Posters Lurkers
54.30 34.03 4.2502 2.317 5 2.55 5.475 4.075 5.3 3.325 5.4 3.7 3.925 3.05 4.85 2.375 2.475 4.4 5.15 3.375 5 3.45 5.075 3.25
8.58 8.10 1.4848 1.354 1.2195 1.3195 1.0374 1.5914 1.6204 1.859 1.3359 1.5225 1.7005 1.3765 1.545 1.3528 1.4521 1.7802 1.406 1.3716 1.6641 1.6478 1.4916 1.4806
10.87