14

Click here to load reader

Sex and power gender di erences in computer

Embed Size (px)

Citation preview

Page 1: Sex and power gender di erences in computer

Sex and power: gender di�erences in computer-mediated interactions

N.M. Sussman*, D.H. Tyson

Department of Psychology, Sociology, and Anthropology, The College of Staten Island, The City University of

New York, Staten Island, NY 10314, USA

Abstract

A preponderance of psychological literature indicates gender di�erences in written and

oral communication. This study explores a new channel of communication, that of cybertalk.As this method of discourse is not gender-salient, one might argue that sex di�erences incommunication style would be eliminated or reduced. However, we suggest that gendered

power di�erentials in communication style transcend the medium. Archived electronic dis-cussions (n=701) on sex-typed topics were selected and analyzed for length, frequency ofcommunication and discourse content (fact vs. opinion). It was hypothesized that male com-municators would display power behaviors by writing longer postings, by posting more fre-

quently, and by writing more opinionated discourse as compared to female communicators.Congruent with the ®rst prediction, men's discourse entries consisted of a greater number ofwords. However, women communicated more frequently than did men, a ®nding opposite to

the hypothesized direction. The third gendered comparison, while not reaching statistical sig-ni®cance, indicated a modest trend with men writing more opinionated communications intwo out of the three sex-typed categories (masculine and gender-neutral). Findings suggest

that cyberspace, a context where gender of communicators is not salient, remains a male-dominated atmosphere, where gender di�erentiation and power displays in communicationpersist, similar to other communication modes. # 2000 Elsevier Science Ltd. All rights

reserved.

Keywords: Computer interaction; Gender di�erences; Power

1. Introduction

Modern technology and the advent of computers have introduced not only a high-tech means of communication but another milieu in which linguistic and written

Computers in Human Behavior 16 (2000) 381±394

www.elsevier.com/locate/comphumbeh

0747-5632/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.

PI I : S0747-5632(00 )00020 -0

CHB 281p Disk used DTD=4.1.0

*Corresponding author. Tel.: +1-718-982-3763; fax: +1-718-982-3794.

E-mail address: [email protected] (N.M. Sussman).

Page 2: Sex and power gender di erences in computer

di�erences between the sexes can be analyzed. Increasingly, computers are a com-mon and versatile household item. It has been estimated that by the year 2000 nearly50% of the population will own a personal computer and have home access to theInternet. Additionally, as of 1997, 72% of all library systems in the US providedaccess to a computer. With this technological advance more conventional modes ofcommunication (face-to-face interactions, letters, telephone discourse) become aug-mented. Individuals can now communicate both nationally and internationally usingcomputer applications, such as e-mail, chat rooms, and computer-mediated discus-sion groups. Literally at one's ®ngertips, social interaction takes on a new dimensionas time and space condense creating a reachable world. Yet, it is unclear whether allmembers of society communicate similarly in this cyberworld. In particular, how domen and women participate and interact within this communication mode?In one of the earliest studies on gender di�erences in communication, Lako�

(1975) proposed that men and women use language di�erently and attempted toprove that ``women's language'' existed. Lako� contended that the manifestation ofwomen's weaker linguistic behaviors (empty adjectives, e.g. `divine' and `charming',tag questions, and hedges) is a product of the female socialization process, a processwhich abets in maintaining women in social subservience. As a result, women aresaid to speak the language of the powerless, whereas men speak the language of thepowerful.It has been clearly demonstrated that sex-linked stereotypes do exist (Kramer,

1977). Women's speech is perceived as gentle, fast, trivial, emotional, detailed andgreat in quantity. In contrast, men's speech is perceived as boastful, demanding,coarse, dominating, forceful and loud. In a later study, Briton and Hall (1995)expanded these ®ndings in an investigation of the perception of male/female non-verbal communicative behaviors. Results indicated that men were believed to inter-rupt more, speak louder, and display more nervous mannerisms when interacting.Women, on the other hand, were believed to be more expressive, more skilled atsending and decoding nonverbal messages and to participate in more nonverbalcommunicative behaviors.To test these robust stereotypical perceptions, a considerable number of empirical

investigations have been conducted. Researchers have investigated verbal and non-verbal communication by exploring a vast array of variables. Gender di�erences havebeen found in simultaneous talk/interruptions (Breshnahan & Cai, 1996; Drass, 1986;James & Clarke, 1993; Mott & Petrie, 1995; Roger, 1989; Zimmerman &West, 1975),amount of talk (Edelsky, 1981; Hirshman, 1994; James & Drakich, 1993; Mulac,1989), non-verbal communication (Mehrabian, 1969), interpersonal conversationdistance (Sussman & Rosenfeld, 1982), power dynamics (Balkwell & Berger, 1996;Dovidio, Brown, Heltman, Ellyson & Keating, 1988; Fishman, 1978; Johnson, 1994;Meeker & Weitzeil-O'Neil, 1977; Tannen, 1984; Walker, Ilardi & Fennel, 1996) aswell as the e�ect of gender identity on conversation (Drass, 1986).Systematic observations of variations in written discourse have also been investi-

gated, though less extensively. Studies within this domain have focused on theliterature produced by grade-school students (Mulac, Studley & Blau, 1990),undergraduates (Deming & Gowen, 1990), adults (Boser, 1991) and literary authors

382 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 3: Sex and power gender di erences in computer

(Roen, Peguesse, & Abordonade, 1995; Roulis, 1995; Rubin & Greene, 1992). Thelatter investigations focused on structural and grammatical di�erences in the literaryworks of male and female writers. Fewer studies have examined male/female com-munication across other media, such as the telephone (Mott & Petrie, 1995) and inthe less gender-conspicuous atmosphere of cyberspace (Matheson, 1991; McCor-mick & Leonard, 1996; Topper, 1997).Consideration has also been made for contextual in¯uences on communication

dynamics. As a result, communicative gender di�erences have been examined ineducational settings within teacher/student exchanges, as well as in the work place(Johnson, 1994; Mott & Petrie, 1995) and at home (Fishman, 1978; O'Donohue &Crouch, 1996). To further add to this wealth of research, gender-based interactivestyles have been explored in situations where the gender of communicators wasapparent as they occurred in same- and mixed-sex groups (Balkwell & Berger, 1996),and among parties of same and unequal status (Aries, 1982; Johnson, 1994; Meeker& Weitzel-O'Neil, 1977; Mott & Petrie, 1995).Via these and other naturalistic and laboratory studies, researchers have isolated

gender-typical verbal, non-verbal and written patterns of communication. In sum,and despite some inconsistencies in ®ndings, these studies have established thatacross contexts men tend to be more dominant, both from a verbal and non-verbalperspective (Mehrabian, 1969). Men tend to be more competitive, proactive (opi-nionated), task-oriented (suggestive and informational), loquacious (producingmore and longer statements), more likely to exercise opinion leadership and moreapt to use conversational strategies (interruptions and dominant posturing) to con-struct relationships based on power (Aries, 1982; Deaux, 1977; Gilligan, 1982). Incontrast, women are more socio-emotional, reactive, verbally dominant only whenin the presence of men of equal status, ask more questions, write longer and moreexpressive statements, and do more supportive work to prolong the longevity of aconversation (Aries, 1982; Fishman, 1978; Roen et al., 1995; Tannen, 1990).Cyber research with an emphasis on gender has grown rapidly in the past 3 years

evidenced in part by the number of relevant studies appearing as dissertationabstracts. Several studies focus on the e�ect of gender on computer use in generaland have found no di�erences between men and women in computer anxiety(Chmielewski, 1998; Otomo, 1998). Katz, Maitland, Hannah Burggraf and King(1999) ®nd that while women report less comfort using computers as compared tomen, they viewed the computer as more useful than men. More speci®c researchregarding gender and Internet use, in a study of fourth-grade students, indicated nodi�erences between girls and boys in the ability to complete Internet activities,although girls displayed greater enthusiasm for the project (Martin, 1998).Two studies have focused on the e�ect of gender and power on Internet commu-

nication. Carstarphen and Lambiase (1998) suggest that gender barriers in cyber-space are erected, in part, by language and code issues and that the rhetoric ofcyberspace is modeled on the power structures and hierarchies of the dominant dis-course in the ``outernet''. Similarly, Cushing (1996), in an ethnographic study, ®ndsa lack of female voices and actors on the Internet and suggests that male rituals andlinguistic patterns dominate.

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 383

Page 4: Sex and power gender di erences in computer

Two dominant viewpoints have emerged in the literature which add explanatoryand theoretical support for the ever present gender distinctions. The socializationperspective posits that roles attached to and socially maintained across the life-spancondition individuals to behave in alliance with gendered expectations. Women andmen are said to belong to di�erent sociolinguistic subcultures (Tannen, 1984), adi�erence which is then internalized and re¯ected in male/female language usage. Inaccordance with gender-role socialization, men's linguistic style is tainted with acompetitive edge, while women engage in relational discourse. Tannen (1990) labelsthis linguistic phenomenon as a process of ``report talk'' fueling men's discourse and``rapport talk'' as the dynamo of women's. Some feminist writers have taken aradical stance with the contention that language is itself a male-governed phenom-enon, against which the female voice is often pitted (Gilligan, 1982).Status theory, however, places more importance on status as the pivotal determi-

nant of gendered communication. Men, being the dominant social group, engage insocial behavior which re¯ects on a micro-social level (via language) the macro-socialgendered hierarchical social structure Ð a dominant patriarchal system. Socialpower is therefore re¯ected in men's language as a function of their ruling masculinestatus (Henley, 1977). Expectation States Theory (EST) has been proposed to eluci-date the relationship between gender and the expression of power in communication(Berger, Rosenholtz & Zelditch, 1980).Dovidio et al. (1988), in an investigation of gender-based task familiarity on

communication power displays, found that both men and women display more ver-bal and nonverbal power dynamics when involved in discussions within their genderdomain. Similarly, Leet-Pellegrini (1980) as cited in Woods (1989), found that statusbehaviors occur within mixed-sex groups but only when gender interacts with inter-actant's expertise on the topic of discussion.The purpose of this study is to determine whether gender-based power di�eren-

ces in communication would generalize to an anonymous communication situa-tion, cyberspace, where discourse content could take precedence over gender ofcommunicators.Shimano� (1980) suggests that stability in behavior across contexts is a clear

indication that a behavior is rule governed. Male power-display in communicationhas been consistent across those contexts that have been investigated in the past.Thus, it is expected that, despite the opportunity which cyberspace o�ers for gender-neutral communication, gender (and the degree to which topics lie within one'sgender domain) would still serve as a status characteristic and thereby foster powerdynamics during cyber interactions. Speci®cally, we hypothesized the following:

1. Men would produce longer postings than would women. In accordance withStatus Theory, it is hypothesized that dominant status behavior is revealed viathe length of newsgroup postings by males. This hypothesis parallels similar®ndings for verbal and written communication.

2. Men would communicate more frequently than would women. Status theoryand prior gender-based communication research ®ndings indicate male verbaland written behaviors to be characterized by competition, task orientation and

384 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 5: Sex and power gender di erences in computer

production of more communicative utterances. We predict similar behavior innewsgroup postings.

3. Men would generate more opinionated statements, regardless of sex-typedtopic. Exercise of powered communication is re¯ected in males' frequent use ofopinion leadership in communication contexts.

Thus, three indices of power-behavior, length of communication, frequency inwhich communication is initiated and proportion of opinions expressed, were uti-lized to test these three hypotheses.

2. Method

2.1. Overview

Newsgroups are a means of `public discussion' allowing people to communicateon a global level. Anyone who has access to a computer equipped with usenet soft-ware can gain access to newsgroups. In a manner similar to sending an e-mail,individuals post messages, questions or comments to `local' news servers who at alater point send the data to be publicly displayed on `neighboring' servers (Bell,1998). Depending on the rules of the server, messages are displayed for days, weeksor months. The public is at liberty to read and respond to the posted messages.There are over 25,000 di�erent newsgroups and the number is continually grow-

ing. Newsgroups are named and organized hierarchically by subject (Hauben, 1993).The range of topics is diverse. There are newsgroups dealing with social issues (e.g.soc.culture.French), political issues (e.g. talk.politics.theory), sports, religion toname a few. This global connectivity o�ers many rewards. Netizens (individuals whoparticipate in the Internet society) can increase their social connections, ®nd othersholding similar interests, participate in collective work, improve quality of life, ®ndemployment, improve communication with friends, and be exposed to, in a non-threatening manner, opposing point of views (Hauben, 1995). Hauben does notethat there is a large male to female ratio among the Net population and that femalescan be the target of net harassment.

2.2. Selection of discussion topics

Out of thousands of discussion topics located on a newsgroup server, 30 wereselected (10 masculine, 10 feminine, and 10 gender-neutral) and in a pilot test,were judged to be of interest either primarily to males, primarily to females or toboth. A one-page questionnaire was constructed as a premeasure to determine thesex typing of discourse topics. Topics were randomly ordered into three rows of 10topics each. A three-point categorical scale (1=feminine, 2=masculine and3=gender-neutral) was used to make the ratings. For example, if the topic `®shing'was deemed as masculine, the rater would insert the number `2' as the appropriateselection in the space provided. The questionnaire was anonymous. Questionnaires

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 385

Page 6: Sex and power gender di erences in computer

were distributed to 50 (21 males, 29 females) undergraduate students at The Collegeof Staten Island, The City University of New York, and sex-typed ratings weremade for each of the 30 discussion topics. Ratings were aggregated for each topicand mean scores were calculated to further sub-divide topics into masculine, femi-nine and gender-neutral categories. Topics with the two highest means within eachgendered category were selected and used in this study. As a result, ®shing andbaseball were selected as representative of the masculine category, ballet and ®gureskating represented the feminine category and news media and theatre were selectedas more androgynous or gender-neutral topics.An example of ®gure skating postings is the following:

Query: ``Just wondered who the fastest skaters are? From TV it seems that its(sic) Paul Wylie and Liz Manley. . . Although they it (sic) hard to tell on TVbut I can still see the speed Paul has.''Response: ``I think Elena Liashenko, Yulia Lawrenchuk, Anna Rechnio, IrinaSlutskaya, and Latitia Hubert are very fast and powerful ladies (sic) skatersbut maybe not necessarily the most elegant skaters.''

An excerpt for the theatre newsgroup read as follows:

I am performing the role of Celia from Shakespeare's As You Like It and amhaving great di�culty ®nding viewpoints, essays and ideas on the play, andspeci®cally, the character. Has anyone got any information or viewpoints onthe play and Ð especially the character: her role in the play, why she marries soquickly, her personality, etc? I would be most grateful for any assistance.

2.3. Measures

2.3.1. Predictor variablesThis study incorporated two quasi-independent variables: sex of communicator

(male/female) and gendered topic (masculine, feminine and gender-neutral). Bothvariables were used as markers for power-behavior in communication. Gender ofcommunicator was determined by the signature attached at the end of discourseentries. In order to reduce the subjectivity involved in this method, postings made byanonymous communicators, those with initialized signatures and ambiguous names(e.g. Pat and Leigh) were excluded from the study. As mentioned above, sex-typingof topic was determined a priori by the pre-test. Since gender di�erences do exist intopic choice (Bischoping, 1993), and familiarity on the issue of discussion clearlyin¯uences power displays (Dovidio et al., 1988), both were selected as markers forcommunication power dynamics.

2.3.2. Criterion variablesThree variables were measured: length of communication, frequency of commu-

nication, and discourse content (fact vs. opinion). Length of discourse entries werecalculated using a word-count method, where the actual number of words within

386 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 7: Sex and power gender di erences in computer

written communications were counted by encoders. Quoted information and statis-tical reproductions were given a one-word count as to better assess the actual verb-osity of communicants.To determine the frequency in which communication was initiated, the ®rst 25

communicants of determinable sex who posted on each of the six selected topics onthe randomly selected days, were assigned a subject number. All of their newsgrouppostings were tallied for the day being investigated. Bales (1950) found a distinctcorrelation between the frequency in which interaction is initiated and expressions ofpower. This corroborates ®ndings by Rosa and Mazur (1979), who found thatcommunicants with higher power initiate speech more often.Content was determined through the use of an author-developed scale. To make

determinations of fact and opinion, four criteria were used in rating discourseentries, 1=factual, 2=opinion supported by fact, 3=opinion without factual sup-port, and 4=miscellaneous (inquiries, advertisements, questions). Postings werecoded as factual if it was considered general knowledge that its composing state-ments were indeed factual. Postings incorporating con®dent statements of opinion,such as ``I believe'', `I suggest'', ``It is my opinion'' or entries indicative of the writersa�ect were classi®ed as opinionated. When the composition of postings was suchthat factual information was used in support of the writer's opinion a classi®cationof ``opinion supported by fact'' was assigned. All other entries (questions, adver-tisements, inquiries) were categorized as miscellaneous.An example of a posting which was coded as ``opinionated'' follows:

I would just like to let everyone know that (name deleted) is complete and totalpsycho ignorant idiot!! Don't ever waste your time auditioning for (name of citydeleted) ballet you will hate life if you dance for this man. Life's too short towaste time on someone who will only cause you grief and frustration during thefew years you may have as a professional dancer.

The following excerpt was coded as ``opinion with fact'':

I can't help but think of the irony involved, if Sammy Sosa wins the home runcrown this year, for the National League. Last year McGuire had more thananyone in the majors, but didn't win the crown because he changed leaguesmid-season. This year he broke Maris' record. . . Just seems a little ironic to me.

2.4. Procedure

Public electronic discourses on the six previously sex-typed topics were archivedover a 2-month period. Eight randomly selected days were then isolated. The codingand content analysis was performed by two encoders (one male and one female).They were each randomly assigned discourse entries on three of the sex-linked topicsfor each of the 8 days. Inter-rater reliability was calculated by means of the Pearsonproduct-moment correlation and prior to coding, coders were trained to encode thedata until a signi®cant level of inter-rater reliability was achieved for each measure

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 387

Page 8: Sex and power gender di erences in computer

(r=0.997, for word count, r=0.987 for gender of communicator, and r=0.78 forcontent). Inter-rater reliability was also con®rmed twice during each month of cod-ing in order to ensure that reliability was maintained among encoders. The ®rst 25postings in which gender of communicant could be determined were then coded forlength, frequency and content. A total of 701 discussions written by 464 males and237 females were analyzed.

3. Results

The length of postings to the newsgroups were initially analyzed in an overall 2(gender)�3 (discussion topics) analysis of variance but revealed no signi®cant maine�ects or interaction. This may be attributed to violations of the assumptions ofequality of variances of the treatment groups. This violation is particularly criticalwith unequal sample sizes such as those found in this study. Due to the predicteddirectionality of the hypotheses, t-tests of independent samples were performed totest each of the hypotheses.The ®rst analysis tested the hypothesis that men would write longer postings than

would women. See Table 1 for the mean length of discourse as a function of genderof communicator and sex-typed topic. Male-generated discourse ranged from 1 to2286 words per posting with a mean length of 105.1 words. Discourse produced byfemales ranged from 2 to 1108 words with an average length of 81.2 words. Overall,and congruent with the prediction, postings made by men consisted of a greaternumber of words than postings generated by women. A Levene's test for equality ofvariance was signi®cant (F=5.929, P=0.01) and indicated that the variances wereheterogeneous. A t-test of means with unequal variances indicated that the di�er-ence between the means was signi®cant, t(691.45)=ÿ2.07, P<0.05, two-tailed. Adetailed analysis of sex of communicator by discussion topic revealed that there wasalso a signi®cant di�erence in the length of male (M=101.8) and female (M=61.1)discourse on masculine topics, t(183)=ÿ2.65, P<0.01, two-tailed. No signi®cantdi�erences were found in the length of discourse made by men and women when

Table 1

Length of communication as a function of gender of communicator and gender-typing of topic (n=701)

Gender-typed topic Gender

Male Female

Masculine 101.8a* 61.1*

Feminine 118.6 84.3

Gender-neutral 100.5 95.4

All topics 105.1** 81.2**

a Represents the average number of words in each posting.

*P<0.01.

**P<0.05.

388 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 9: Sex and power gender di erences in computer

writing on feminine and gender-neutral topics. However, the means were in theexpected direction so that even on feminine topics (ballet and ®gure-skating) males'postings were longer than were females' postings.The second hypothesis predicted that men would post more frequently than would

women. See Table 2 for the mean number of times in which communication wasinitiated each day as a function of gender- and sex-typed topic. A two-tailed inde-pendent t-test for frequency reached signi®cance, but opposite to the hypothesizeddirection, t(300.23)=2.86, P<0.01 in that individual women communicated moreoften on a single day (M=1.86) than did men (M=1.42). The gender di�erence,however, was mediated by gender-typed topic such that on feminine-typed issues,females communicated an average of 2.02 times per day as compared with men whocommunicated an average of 1.36 times, t(230.36)=3.06, P<0.01. Communicationfrequency was not signi®cantly di�erent for men and women on either masculine orgender-neutral typed topics but continued the trend for females to post more fre-quently than males.The third hypothesis predicted that men would produce more opinionated dis-

course. Due to unequal sample sizes and categorical data, percentages were calcu-lated to assess the proportion of entries as a function of gender-and sex-typed topicthat were coded as opinion without factual support, opinion with factual support,factual, and miscellaneous (Table 3). A chi-square test comparing the content ofmale and female postings indicated non-signi®cance. Not surprisingly for discoursein an Internet newsgroup, nearly 65% of the content was opinionated, either with orwithout fact for both males and females. However, in an examination of the per-centages by discussion topic found among masculine-typed topics and gender-neutral topics, male postings tended to be more opinionated without fact than werewomen's postings.While not speci®cally hypothesized, it is interesting to note that the total number

of male postings (n=464) on these six newsgroup sites was nearly double the num-ber of female postings (n=267). Even on feminine-typed topics, the frequency ofmale postings (n=101) was similar to the number of female postings (n=168). In

Table 2

Frequency in which communication was initiated as a function of gender of communicator and gender-

typing of topics (n=701)

Gender-typed topic Gender

Male Female

Masculine 1.52a 1.57

Feminine 1.36* 2.02*

Gender-neutral 1.28 1.32

All topics 1.42* 1.86*

a Represents the average number of postings by each individual on a particular day.

*P<0.01.

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 389

Page 10: Sex and power gender di erences in computer

contrast, on masculine-typed topics, male postings (n=249) far outnumbered femalepostings (n=44).

4. Discussion

In this study, we examined the communicative styles of men and women within acontext where sex of interactants was not salient. Past research has consistentlyrevealed power-di�erentials in communication and further, Matheson (1991) indi-cated that the mere expectation of the gender of one's communication partnerin¯uenced computer-mediated social behavior. Thus, it was expected that gender-pertinent power expressions would manifest during cybertalk. The length of post-ings, frequency at which communication was initiated, and discourse content (opi-nion vs. fact) were used as markers of power-related behaviors.We anticipated that male power dynamics would appear most clearly within the

masculine and gender-neutral sex-typed categories. Hypothesis (1) predicted thatmen would generate longer postings and this ®nding was con®rmed. Men had muchmore to say than women irrespective of sex-typed topic. The two greatest di�er-entials in the length of men's and women's posted discourse occurred, as expected,in the relation of masculine and (more so) gender-neutral issues. Since quantity oftalk or length of holding the ¯oor is an index of leadership, our ®ndings parallelthose of Walker et al. (1996) who found that when in mixed-sexed groups men are®ve times more apt than women to assume a leadership role. Our ®ndings, therefore,support the contention that gender-based di�erences in behavior persist as theyrelate to length of cyber-conversation.Hypothesis (2) predicted that men would initiate discussions more frequently. Our

®ndings, however, were in the opposite direction. The ®ndings indicate that womeninitiated discourse more than men but particularly on female-typed topics. We sug-gest three possible explanations. First, women may have experienced a greaterdegree of comfort when interacting on female-linked issues anticipating that theircyber correspondents would be all female. Thus, power communication rules wouldnot be in e�ect. Second, women may have demonstrated greater expertise in their

Table 3

Content of posting as a function of gender and sex-typed topic (n=701)

Opinion without fact Opinion with fact Fact Miscellaneous

Male Female Male Female Male Female Male Female

Gendered topic

Feminine 38.6a 39.9 27.7 24.1 15.8 18.5 17.8 17.3

Masculine 34.5 31.8 32.9 31.8 14.5 9.1 18.0 27.3

Gender-neutral 44.7 28.0 17.5 36.0 13.2 16.0 24.6 20.0

All topics (%) 37.9 37.1 28.0 27.0 14.4 16.4 19.6 19.4

a Percentage of postings coded in each of four content categories.

390 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 11: Sex and power gender di erences in computer

female-typed discourses. This latter alternative is supported in that women's post-ings were slightly more factual than were men's on female topics, indicative of feel-ings of expertise. Finally, the literature indicates that women's communication issocio-emotionally oriented and that women engage in supportive work to prolongthe longevity of a conversation. Perhaps in the realm of Internet discussions, fre-quency of postings is the tactic employed to maintain a conversation.It is worth noting, however, that twice as many men as women participated in the

six newsgroups. So while individual men did not multiply post during a single day,overall males dominated these discussion groups. Power behavior then is re¯ected intotal male presence on these site and supports Cushing's (1996) contention that malerituals and linguistic patterns dominate the Internet.Third, we predicted that men would produce more opinionated discussions. We

expected that women would rely on factual as opposed to opinionated statementsand that the masculine and gender-neutral categories would be the arena for powerplay. While the chi-square test did not reveal signi®cant di�erences in behavior, themeans were in the predicted direction. Men as compared to women were more opi-nionated when discussing male-linked issues and even more so, when the categorywas androgynous. The overall high percentage of opinionated postings for bothmale and females may have skewed this measure and we suggest a more precisemeasure of content of postings be developed.In summary, cyberspace has the potential to allow communicants to become dis-

inhibited from sexualized bounds and explore true freedom of expression (Siegel,Dubrovsky, Kiesler & McGuire, 1986). One might hope that via cyberspace womenand men could transcend the socialized constraints on their communicative expres-siveness and adopt a more androgynous style of interaction. However, our ®ndingssuggest that power-behaviors in communication, especially regarding length ofpostings and the ration of males to females in each newsgroup, have becomeintransiently socialized into behavioral dynamics, as Socialization Theory posits,such that discourse medium becomes irrelevant. Power di�erentials in communica-tion still persist and it appears that cyberspace is a male-dominated atmosphere(McCormick & Leonard, 1996).On a methodological note, some limitations constrained our interpretations. Due

to both the `®eld' nature of the data collection and participant anonymity, noinformation was known about and no control was made for race, ethnicity, socio-economic status of communicators, use of aliases, and the in¯uence of personalityon behavior. Drass (1986) examined the relationship between gender identity andverbal behavior in same-sex interactions and found support for the notion thatbehavior is indeed a�ected by one's sense of self. As a result, more androgynousindividuals, communicating with both masculine and feminine communicativecharacteristics could in¯uence the validity of our ®ndings. We attempted to mini-mize such confounds through the large sample (701) of postings.Overall, this study adds weight to the theory that gender di�erences in commu-

nication are mediated by power and status. One might say that power behaviorswhich are socialized early in life, later become rule governed such that even innon-gender-salient contexts, we rely on gender power dynamics. Our results revealed

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 391

Page 12: Sex and power gender di erences in computer

that overall men post more often than women, men engaging in power discoursecommunicate longer irrespective of the topic being discussed and deliver more opi-nionated speech on masculine and gender-neutral topics. Women, however, in amore narrow domain of feminine-typed topics, initiate factual interactions morefrequently, perhaps indicating expertise rather than power.Findings of continued gender power di�erences have implications for the use of

the Internet in both work and academic settings. Faculty who introduce Internetdiscussion groups as a pedagogical improvement, for example, may ®nd that malestudents predominate and dominate the discussions adding yet another context forgender inequalities in communication.

Acknowledgements

The authors thank Maury Silverman and Scott Sugarman for their assistance inthis project. A portion of this paper was presented as a poster at the 70th AnnualMeeting of the Eastern Psychological Association, 16±18 April 1999, Providence,Rhode Island.

References

Aries, E. J. (1982). Verbal and nonverbal behavior in single-sex and mixed-sex groups: are traditional sex

roles changing? Psychological Reports, 51, 127±134.

Bales, R. (1950). Interaction process analysis. Reading, MA: Addison-Wesley.

Balkwell, J. W., & Berger, J. (1996). Gender, status, and behavior in task situations. Social Psychology

Quarterly, 59, 273±283.

Bell, J. (1998). News.newusers.questions: what newsgroups are and how they work. Availabe at: http://

people.ne.mediaone.net/babay/how-it-works.html (accessed on 25 January 2000).

Berger, J., Rosenholtz, S. J., & Zelditch Jr., M. (1980). Status organizing processes. Annual Review of

Sociology, 6, 479±508.

Bischoping, K. (1993). Gender di�erences in conversation topics, 1922±1990. Sex Roles, 28, 1±28.

Boser, J. A. (1991). Gender di�erences: let's see them in writing. Paper presented at the annual meeting of

the Mid-South Educational Research Association, Lexington, KY (ERIC Document Reproduction

Service No. ED 341 980).

Breshnahan, M. I., & Cai, D. H. (1996). Gender and aggression in the recognition of interruption. Dis-

course Processes, 21, 171±189.

Briton, N. J., & Hall, J. A. (1995). Beliefs about female and male nonverbal communication. Sex Roles,

23, 79±90.

Carstarphen, M., & Lambiase, J. J. (1998). Domination and democracy in cyberspace: reports form the

majority media and ethnic/gender margins. In B. Ebo, Cyberghetto or cybertopia?: race, class and gender

on the internet (pp. 121±135). Westport, CT: Praeger.

Chmielewski, M. (1998). Computer anxiety and learner characteristics: their role in the participation and

transfer of internet training. Dissertation Abstracts International, 59(3A), 791A.

Cushing, P. J. (1996). Gendered conversational rituals on the internet: an e�ective voice is based on more

than simply what one is saying. Anthropologica, 38, 47±80.

Deaux, K. (1977). Sex di�erences. In T. Blass, Personality variables in social behavior (pp. 357±377).

Hillsdale, NJ: Earlbaum.

Deming, M. P., & Gowen, S. G. (1990). Gender in¯uences on the language processes of College Basic

writers. Community/Junior College Quarterly of Research and Practice, 14(3), 177±187 (ERIC Docu-

ment Reproduction Service No. EJ 413745).

392 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394

Page 13: Sex and power gender di erences in computer

Dovidio, J. F., Brown, C. E., Heltman, K., Ellyson, S. L., & Keating, C. F. (1988). Power displays

between women and men in discussions of gender-linked topics: a multichannel study. Journal of Per-

sonality and Social Psychology, 55, 580±587.

Drass, K. A. (1986). The e�ect of gender identity on conversation.Social PsychologyQuarterly, 49, 294±301.

Edelsky, C. (1981). Who's got the ¯oor? Language in Society, 10, 383±421.

Fishman, P. M. (1978). Interaction: the work women do. Social Problems, 25, 397±406.

Gilligan, C. (1982). In a di�erent voice: psychological theory and women's development. Cambridge, MA:

Harvard University Press.

Hauben, R. (1993). The evolution of usenet news: the poor man's arpanet. Available at: http://stu-

dentweb.tulane.edu/�rwoods/netbook/ch.1_poorman_arpa.html (accessed on 26 January 2000).

Hauben, M. (1995). The Net and Netizens: the impact the Net has on people's lives. Available at: http://

studentweb.tulane.edu/�rwoods/netbook/ch.7_netizen.html (accessed on 19 January 2000).

Henley, N. M. (1977). Body politics: power, sex and nonverbal communication. Englewood Cli�s, NJ:

Prentice Hall.

Hirschman, L. (1994). Female±male di�erences in conversational interaction. Language in Society, 23,

427±442.

James, D., & Clarke, S. (1993). Women, men and interruptions: a critical review. In D. Tannen, Gender

and conversational interaction (pp. 231±267). New York: Oxford University Press.

James, D., & Drakich, J. (1993). Understanding gender di�erences in amount of talk: a critical review of

research. In D. Tannen, Gender and conversational interaction (pp. 281±312). New York: Oxford Uni-

versity Press.

Johnson, C. (1994). Gender, legitimate authority, and leadership subordinate conversations. American

Sociological Review, 59, 122±135.

Katz, L., Maitland, M., Hannah, R., Burggraf, K., & King, S. (1999). The e�ects of gender and academic

program on learning styles and attitudes of undergraduate students using multimedia, web-based

anatomy labs. In J. Chambers, Selected papers from the 10th International Conference on College

Teaching and Learning (pp. 81±87). Jacksonville, FL: Florida Community College at Jacksonville.

Kramer, C. (1977). Perceptions of female and male speech. Language in Speech, 20, 151±161.

Lako�, R. (1975). Language and women's place. New York: Harper & Row.

Martin, S. (1998). Internet use in the classroom: the impact of gender. Social Science Computer Review,

16, 411±418.

Matheson, K. (1991). Social cues in computer-mediated negotiations: gender makes a di�erence. Compu-

ter in Human Behavior, 7, 137±145.

McCormick, N., & Leonard, J. (1996). Gender and sexuality in the cyberspace frontier. Women & Ther-

apy, 19, 109±119.

Meeker, B. F., & Weitzel-O'Neil, P. A. (1977). Sex roles and interpersonal behavior in task-oriented

groups. American Sociological Review, 42, 91±105.

Mehrabian, A. (1969). Signi®cance of posture and position in the communication of attitude and status

relationships. Psychological Bulletin, 71, 359±372.

Mott, H., & Petrie, H. (1995). Workplace interactions: women's linguistic behavior. Journal of Language

and Social Psychology, 14, 324±336.

Mulac, A. (1989). Men's and women's talk in same-gender and mixed-gender dyads: power or polemic?

Journal of Language and Social Psychology, 8, 249±270.

Mulac, A., Studley, L. B., & Blau, S. (1990). The gender-linked language e�ect in primary and secondary

students' impromptu essays. Sex Roles, 26, 439±469.

O'Donohue, W., & Crouch, J. (1996). Marital therapy and gender-linked factors in communication.

Journal of Marital and Family Therapy, 22, 87±101.

Otomo, Y. (1998). The relationship of computer anxiety, mathematics anxiety, test anxiety, gender, and

demographic characteristics among community college students. Dissertation Abstracts International,

59(6A), 1957A.

Roen, D. H., Peguesse, C., & Abordonade, V. (1995). Gender and language variation in written commu-

nication. In D. L. Rubin, Composing social identity in writing language (pp. 113±131). Hillsdale, NJ:

Earlbaum.

N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394 393

Page 14: Sex and power gender di erences in computer

Roger, D. (1989). Experimental studies in dyadic turn-taking behavior. In D. Roger, & P. Bull, Con-

versation (pp. 75±95). Clevedon, PA: Multilingual Matters Ltd.

Rosa, E., & Mazur, A. (1979). Incipient status in small groups. Social Forces, 58, 18±37.

Roulis, E. (1995). Gendered voice in composing, gendered voice in evaluating: gender and the assessment

of writing quality. In D. L. Rubin, Composing social identity in writing language (pp. 151±185). Hills-

dale, NJ: Earlbaum.

Rubin, D. L., & Greene, K. (1992). Gender-typical style in written language. Research in the Teaching of

English, 26, 7±40.

Siegel, J., Dubrovsky, V., Kiesler, S., & McGuire, T. W. (1986). Group processes in computer-mediated

communication. Organizational Behavior and Human Decision Processes, 37, 157±187.

Shimano�, S. (1980). Communication rules: theory and research. Beverly Hills, CA: Sage.

Sussman, N. M., & Rosenfeld, H. M. (1982). Culture, sex and language: e�ects on interpersonal distan-

cing. Journal of Personality and Social Psychology, 42, 66±74.

Tannen, D. (1984). Gender and conversational interaction. New York: Oxford University Press.

Tannen, X. (1990). You just don't understand: women and men in conversation. New York: William

Morrow.

Topper, A. (1997). Comparing face-to-face and electronic discourse: issues and questions raised in research

study. Paper presented at the meeting of the American Educational Research Association, Chicago, IL

(ERIC Document Reproduction Service No. ED 408 578).

Walker, H. A., Ilardi, B. C., & Fennell, M. L. (1996). Gender, interaction and leadership. Social

Psychology Quarterly, 59, 255±272.

Woods, N. (1989). Talking shop: sex and status as determinants of ¯oor apportionment in a work setting.

In J. Coates, & D. Cameron, Women in their speech communities (pp. 141±157). New York: Longman.

Zimmerman, D. H., & West, C. (1975). Sex roles, interruptions and silences in conversation. In B. Thorne,

& W. Henley, Language and sex: di�erence and dominance (pp. 105±129). Rowley, MA: Newbury

House.

394 N.M. Sussman, D.H. Tyson /Computers in Human Behavior 16 (2000) 381±394