Multivariate statistical analyses were used to determine if there were differences in social community, learning community, and perceived learning between male and female students in 12 online graduate education courses in which female students (n = 162) outnumbered males (n = 31). Study results provided evidence that females felt more connected to other students in their courses, felt that their online learning experiences were more aligned to their educational values and goals, and perceived they learned more than their male peers.
Introduction
A foundational premise of online learning programs is that they are educationally effective. In numerous studies, online distance education has been found to be as effective as equivalent face-to-face instruction, as chronicled in the body of “no significant difference” studies (Russell, 1999) and institutional survey research (Allen & Seaman, 2003). Despite this, there remain lingering concerns among educators and researchers whether the level of learning attainment in distance programs is as good as face-to-face courses (Noble, 2002; Phipps & Merisotis, 1999). Concurrent with this debate, there is a growing interest in the dynamics of the online classroom and research into educational improvement, particularly with regards to media selection, instructional design, interactive social dynamics, and pedagogical techniques (e.g., Anderson, 2003; McDonald, 2002; Olson & Wisher, 2002; Peters, 2003) and how these factors can influence learning.
The bulk of online higher education is currently delivered using asynchronous learning network technologies (Walts & Lewis, 2003). Best practices literature (e.g., Palloff & Pratt, 1999) has emphasized the use of learnerinstructor and learner-learner interaction and social networking in the online classroom for the purpose of fostering a strong online learning community. Such literature is often filled with community-building exercises offered in the hope of significantly improving learning effectiveness. While the research into social presence and classroom community in the online environment has supported this assertion (e.g., Richardson & Swan, 2003; Rovai, 2001), it is unwise to presume that one particular online instructional strategy would affect all students equally. Field dependent learners, for example, learn better in socially rich learning environments, while field independent learners prefer working alone (Jonassen & Grabowski, 1993). One might conclude that the geographic separation inherent in online courses would favor field independent learners, but the literature doesn’t bear this out (e.g., Brenner, 1997) and the practitioner literature in online education argues for an instructional style that fits the biases of field dependent learners. Other student characteristics including gender, age, educational background, computer literacy, learning style preferences, and experience with online learning, would likely influence receptivity to particular instructional strategies and influence online learning effectiveness.
Gender in particular is an area worthy of additional consideration, especially since distance education has been extensively marketed to women ever since correspondence course days (Kramarae, 2003). The academic and popular literature is rife with evidence of gender-based communication differences (e.g., Belenkly, Clinchy, Goldberger, & Tarule, 1986; Gray, 1993; Griffin, 1997; Tannen, 1989, 1990, 1991, 1994). In particular, the professional literature suggests that most women seek to establish intimacy in a relationship, whereas most men seek to establish status in a hierarchy, measured in terms of independence (Tannen, 1989, 1991).
Intimacy is key in a world of connection where individuals negotiate complex networks of friendship, minimize differences, try to reach consensus, and avoid the appearance of superiority, which would highlight differences. In a world of status, independence is key, because a primary means of establishing status is to tell others what to do, and taking orders is a marker of low status. Though all humans need both intimacy and independence, women tend to focus on the first and men on the second. It is as if their lifeblood ran in different directions. (Tannen, 1990, p. 26)
Accordingly, “[d]istance education is … yet another institution where gender and power differences are constructed, and to ignore the ways that gender is under construction online is to ignore many difficult experiences of real people” (Kramarae, 2003, p. 269). von Prümmer and Rossié (2001) went further and declared that, “If gender is not seen as relevant, the system will not be equally accessible to women and men and will offer men more chances to succeed” (p. 137). In her counsel to online instructors when considering gender in distance education, Burge (1998) offers guidelines in the areas of constructivism, womenfriendliness, and technology. One of her women-friendly guidelines reads, “Watch for male domination of discussion and other forms of typically male styles of interaction” (p. 38).
Therefore, the purpose of the present research was to examine the possibility that online asynchronous learning is differentially effective based on patterns of socialization, which tend to be gender based. The professional literature (e.g., Belenky et al., 1986; Tannen, 1990, 1994) suggests that students adopting the connected pattern of socialization in society (usually women) are more likely to seek membership in learning communities than their status-seeking (often male) counterparts. Since online programs emphasize interpersonal interactions for deeper learning, in accordance with constructivist epistemology, and females represented the majority of online students in the present study, the authors hypothesized that females would set the tone for online discussions and they would have both a stronger sense of community and a greater perception of learning at the end of their online courses. This hypothesis is consistent with the research reported by Hall (1996) and Herring (1996) that suggested predominantly female online groups are more likely to generate a polite and civil environment of online discussions, an environment friendly to the female pattern of communication.
Gender and Communication
Although early computer-mediated communication (CMC) literature postulated that the limited social cues found online would minimize gender differences (e.g., Kiesler, Siegel, & McGuire, 1984), Herring (1993) claimed that online communication is less democratic than proponents touted because gender-based communication styles carry over into electronic environments. Postmes, Spears, and Lea (1998) conducted research examining the dynamics of such online communication using their social identity model of deindividuation effects (SIDE) framework for analysis. They hypothesized that the effect of deindividuation, “a psychological state of decreased self-evaluation, causing altinormative and disinhibited behavior” (p. 695), is one of greater susceptibility to “the influence of norms, social attraction to group members, stereotypes, and ingroup favoritism” (p. 699). They found the assumption that CMC encourages the liberation of individuals from social influence and general social norms wasn’t supported. Moreover, when communicators shared a common social identity, they were more susceptible to these influences when interacting online.
Scott (1999) elaborated the SIDE model in the following manner: “when personal identity is salient, anonymity may further reduce one’s sense of groupness and adherence to group norms; counterintuitively, when a more social identity (e.g., work group) is salient, anonymity actually increases one’s sense of groupness and adherence to group norms” (p. 459). In other words, individuals tend to engage in heightened stereotypical behavior, conform to group norms, commit to the group, and engage in “us versus them” behavior, given the relative anonymity of text-based CMC. Moreover, Inzlicht and Ben-Zeev (2000) reported that group composition (i.e., the gender mix of a group) can also trigger stereotype-relevant thoughts and behaviors because group composition can make salient one’s social identity and the stereotypes associated with that identity. Because women are stereotypically the primary caregivers of children and the keepers of family history, they are focused on emotional aspects of experience that serve to regulate relationships (e.g., Gilligan, 1982).
Hiltz and Turoff (1978) theorized that people perceived a high degree of impersonality when they were online. This depersonalization might also have a number of effects, including a willingness to express opinions or ideas that would not be expressed in a personalized setting. Even within the online environment, however, Postmes, Spears, Sakhel, and de Groot (2001) observed different behavior in anonymous groups versus identifiable groups in which members shared personal biographies or photos with group members. They found that identifiable groups were less susceptible to conforming to a primed norm than their anonymous counterparts. They wrote, “our understanding of anonymity’s effects in groups … might benefit from distinguishing local and societal norms and making more apparent which each will exert its influence” (p. 1253). Postmes et al. (1998) noted that one flaw in the oft-cited egalitarian view of online communication is that it fails to consider that social norms are not merely externally applied, but are also internalized boundaries that influence personal behavior. “Although CMC gives us the opportunity to traverse social boundaries, paradoxically, it can also afford these boundaries greater power, especially when they define self- and group identity” (p. 689).
This assertion is consistent with the findings of Blum (1999) and Rovai (2001) who concluded that the voice of students engaged in online courses is related to gender. Based on their research findings, the majority of men (and some women) exhibited an independent voice and the majority of women (and some men) used a connected voice. Rovai (2001) noted these differences and commented, “The independent voice tends to be impersonal and assertive, that is, it possesses an authoritative tone. The connected voice, on the other hand, is generally supportive and helpful without being assertive” (p. 45). These reported findings are in line with psychological research (e.g., Cross & Madson, 1997) that provided evidence of gender differences in connected and separate self-concepts as well as epistemological research reported by Belenky et al. (1986) in which women more likely than men use connected knowing while men are more likely to use separate knowing. According to Belenky et al. (1986), individuals who utilize a separate way of knowing distance themselves from the object of knowledge and place an emphasis on objectivity, reason, doubting, analysis, and evaluation. On the other hand, those who utilize a connected way of knowing emphasize understanding, empathy, acceptance, and collaboration, suggesting stronger sense of community within a learning environment.
Shea, Fredericksen, Pickett, Pelz, and Swan (2001) examined students in the State Univeristy of New York (SUNY) Learning Network and found that female students had higher levels of learner-instructor and learner-learner interaction than their male counterparts. Flanagin, Tiyaamornwong, O’Connor, and Seibold (2002) found that men and women communicated in mixed-gender groups differently. Men were more likely to communicate with both men and women just as they would in a face-to-face environment while women preferred to reduce their social cues and thus provide an opportunity for greater power and influence in online conversations in a mixed-gender environment.
Although online communication patterns by gender have been evident in the literature, gender differences have not been as consistent with regards to student achievement or satisfaction. Ory, Bullock, and Burnaska (1997) found no significant gender differences in student attitudes toward asynchronous learning networks after a year of implementation in a university setting. Lim (2001) considered the influence of gender along with a number of other factors (e.g., computer self-efficacy, academic self-concept, age, academic status, etc.) and concluded that computer self-efficacy was the singular predictor of student satisfaction. Clay-Warner and Marsh (2000) found that gender was not a significant influence on students’ acceptance of the use of online communication in instruction. In contrast, Shea et al. (2001) discovered that females in the SUNY Learning Network were more satisfied with online learning than their male peers.
Online Community
A recurring theme in online education literature is the importance of developing a learning community to foster effective instruction. Berg (1999) encouraged online instructors to use virtual teams to foster community development in distance courses. “The notion of community associated by pursuits is one that is useful in relationship to education—it leads to rich connections with both constructivist learning theory and lessons learned from teams in business environments” (p. 33). Palloff and Pratt (1999) called for the development of academic communities in online distance classes and declared, “without the support and participation of a learning community, there is no online course” (p. 29).
Tinto (1993) found that students in traditional learning environments who possess strong feelings of community are more likely to persist than those students who feel alienated and alone. Studies into community in the online classroom have led researchers to similar conclusions. Vandergrift (2002), for example, performed a case study of an online children’s literature class and found the community dynamic to be so pedagogically significant that she advocated online instructors practice “restrained presence” to avoid interfering with the learner-learner dynamics. Such recommendations are consistent with Moore’s (1989) three-fold interaction construction which identified learner-learner interaction as one of the dynamics within a distance course.
McLellan (1999) postulated that the development of online communities would strengthen the online learning experience by fostering a sense of social presence among course participants (p. 40). Moller (1998) similarly encouraged the development of learning communities in asynchronous online courses. She stated that “the potential of asynchronous learning can only be realized by designing experiences and environments which facilitate learning beyond the content-learner interaction. To that end, it becomes necessary to create learner support communities” (p. 115). Such online learning communities provide a framework for social reinforcement and information exchange while girding the learning experience with academic, intellectual, and interpersonal support (p. 116). This is consistent with Wegerif’s (1998) study that found the social dimension of asynchronous learning networks to be an important predictor of the success of the distance learner. He concluded that forming a sense of community is a necessary first step for collaborative learning without which students are likely to be unwilling to take the risks involved in learning.
Measures of Learning
Educators largely view learning in terms of cognitive change. Dumont (1996) and Hiltz and Wellman (1997) report that student grades continue to be the most prevalent measure of student learning outcomes. However, using grades to operationalize learning may not always provide the best results. Classroom grades, particularly for graduate courses, tend to have very restricted ranges, that is, they tend to reflect uniformly superior achievement, thus limiting their use. Additionally, grades can have little relationship to what students have learned. For example, students may already know the material when they enroll or their grade may be more related to class participation, work turned in late, or attendance than to cognitive learning. Furthermore, grades may not be a reliable measure of learning, particularly for performance tests, as different teachers and even the same teacher over time will likely not assign grades in a consistent manner. Therefore, using grades as a measure of cognitive learning can be problematic. Since instruction is designed to foster learning, a self-reported perception of learning should reflect a student’s view of the educational effectiveness of the course. Accordingly, the authors used self-reported perceived learning to operationalize cognitive learning in the present study.
Bloom (1956) conceptualized the educational process as consisting of three domains: affective, psychomotor (i.e., behavioral), and cognitive. The affective domain consists of student attitudes and feelings toward the subject, the psychomotor domain reflects the student’s propensity to engage in subject-oriented behaviors as a result of the learning experience, and the cognitive domain relates to the knowledge acquired about a subject as a result of the learning experience. Research evidence suggests self-reports can be a valid measure of cognitive learning. Pace (1990) supported the validity of student self-reports of cognitive learning based on research evidence that suggested the consistency of results over time and across different populations. He also found that patterns of outcomes vary for self-reports of learning across majors and length of study in the same manner as was established through direct achievement testing.
It is conceivable that a student could, for example, overstate his or her perceived learning in a course and thus undermine the reliability of this as a measure of cognitive learning. The literature seems to indicate, though, that adult students have sufficient educational experience to provide a more accurate assessment of their cognitive learning experiences (e.g., Chesebro & McCroskey, 2000; Corrallo, 1994). In a summary of this literature, Corrallo (1994) noted that “there is a considerable literature concerned with establishing the validity of student self-reports about cognitive outcomes” (p. 23). He concluded that self-reports of cognitive gain are indicative of results obtained through more direct forms of assessment.
Methodology
Participants
A total of 281 students enrolled in 12 online graduate education courses delivered at a distance by an accredited university in the state of Virginia were invited to participate in this study. Of this number, 193 volunteered, resulting in a volunteer rate of 68.7%. Males represented 16.1% of participants and the remaining 83.9% were females. The large percentage of female students is consistent with the relatively large number of female K-12 educational practitioners. The ethnic breakdown was: Caucasian (62.7%), African American (30.1%), Asian (2.1%), and other (5.2%). The mean age of the 187 participants who divulged this information was 39.01 (SD = 9.87).
Setting
The online courses used for this study were asynchronous courses delivered using the Blackboard.comSM e-learning system. This system consists of an integrated set of application tools that are accessible to students via the Internet. All courses were one semester (i.e., 16-weeks) in duration and were taught entirely via the Internet by faculty experienced in online teaching. No course possessed an enrollment of fewer than 10 students. Course titles included “Foundations of Teaching and Learning” and “Technology Integration in Curriculum and Instruction.” A mix of male and female professors taught the courses.
Instrumentation
Data for the study were gathered using an online survey that included: (a) the Classroom Community Scale (CCS) (Rovai, 2002), (b) the self-report measure of perceived cognitive learning (McCroskey, Sallinen, Fayer, Richmond, & Barraclough, 1996), and (c) various demographic questions regarding gender, ethnicity, and age.
The CCS was used to measure classroom community. This instrument consists of 20 self-report items that examine community within the classroom context, such as I feel isolated in this course and I feel that this course is like a family. Following each item is a 5-point Likert scale of potential responses: strongly agree, agree, neutral, disagree, and strongly disagree. Study participants checked the place on the scale that best reflected their feelings about the item. Scores were computed by adding points assigned to each of the 20 five-point items. Items were reverse-scored where appropriate to ensure the least favorable choice was always assigned a value of 0 and the most favorable choice was assigned a value of 4. The CCS produced social community and learning community subscales. Scores on each subscale can range from 0 to 40, with higher scores reflecting a stronger sense of community. Social community represents the feelings of the community of students regarding their cohesion, spirit, trust, safety, interdependence, and social presence. Learning community represents the feelings of community members regarding the construction of understanding through participation in discussions and the degree to which they shared values and beliefs concerning the extent to which their educational goals and expectations were being satisfied by community membership.
The results of a factor analysis reported by Rovai (2002) confirmed that the two subscales were latent dimensions of the classroom community construct. Cronbach’s coefficient alpha for the full classroom community scale was .93. Additionally, the internal consistency estimates for the social community and learning community subscales were .92 and .87, respectively. In the present study, alpha estimates for the full classroom community scale and the social community and learning community subscales were .88, .90, and .72, respectively, suggesting acceptable reliability.
Perceived cognitive learning was measured by student self-reports of their learning. The instrument employed was first used by Richmond, Gorham, & McCroskey (1987) and has been used in many studies related to cognitive learning since then. Study participants were asked to respond to one question: On a scale of 0 to 9, how much did you learn in this class, with 0 meaning you learned nothing and 9 meaning you learned more than in any other class you’ve had? Since the instrument was a single-item scale, no internal consistency reliability estimates were possible; however, McCroskey et al. (1996) reported that testretest reliability over a 5-day period was .85 in a study of 162 adult learners.
Procedures
Data were collected during the final 3 weeks of the semester and for 1 week following the semester for each of the 12 online courses sampled in this study so that students would have substantial exposure to the course about which they were responding. The perceived learning question, along with demographic questions regarding gender, ethnicity, and age were made available to students via an online survey. The researcher e-mailed students on a weekly basis during the 4-week data collection effort, providing directions and encouragement for completing the survey. Participants completed the surveys prior to learning their course grades.
Design and Data Analysis
The present study responds to the following research question: Are there differences in social community, learning community, and perceived learning between male and female students in a predominately female online learning environment? Multivariate analysis of variance (MANOVA) was used to analyze the data followed by separate ANOVAs for each dependent variable and discriminant analysis to determine how accurately student gender could be determined based on a linear combination of the three measures. Effect size was calculated using the eta squared (η2) statistic and interpretation was based on Cohen’s (1977) thresholds of .01 for a small effect, .06 for a moderate effect, and .14 for a large effect. The procedures used for each analysis are described in the results section below.
Results
Table 1 presents the descriptive statistics disaggregated by gender for the 193 study participants, displaying means and standard deviations for all dependent variables. Female students responded with higher scores than males across all variables, reflecting stronger sense community and greater levels of perceived learning for the courses under consideration. Moreover, female participants posted significantly more messages (M = 79.20, SD = 68.82) to course discussion boards during the semester than males (M = 37.57, SD = 32.62), t (90.64) = 5.22, p < .001, η2 = .05. Table 2 displays the Pearson r intercorrelations between all pairs of variables.
Descriptive Statistics Disaggregated by Gender
| n | Social Community | Learning Community | Perceived Learning | |
|---|---|---|---|---|
| Males | 31 | 25.13 (6.00) | 27.58 (4.53) | 6.61 (1.87) |
| Females | 162 | 28.43 (6.58) | 29.77 (4.70) | 7.32 (1.49) |
| Total | 193 | 27.90 (6.59) | 29.41 (4.73) | 7.21 (1.65) |
Note: Displayed are sample size (n) and means with standard deviations enclosed in parentheses. Total possible community scores can range from 0 to 40, with higher scores reflecting a stronger sense of community. Perceived learning scores can range from 0 to 9, with higher scores reflecting greater perceived learning.
Intercorrelations
| Scale | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| 1. Social community | – | .56 | .48 | ns |
| 2. Learning community | – | .56 | .17 | |
| 3.Perceived learning | – | .20 | ||
| 4.Total messages posted | – |
p < .05, ns = not significant.
Figure 1 is a dendrogram based on Pearson correlation coefficients that provides a visual representation of variables being combined into a single cluster using average linkage (between groups) and showing values of the distance coefficients at each step in the clustering process. Connected vertical lines designate joined variables. The dendrogram rescales the actual distances to numbers between 0 and 25, preserving the ratio of the distances between steps. The analysis works upwards to place every variable into a single cluster. Therefore, one ends up with a single cluster that subdivides at lower levels of similarity. For data in the present study, the analysis first combined social community and learning into a single classroom community cluster and then added perceived learning. Finally, at the greatest distance, the final variable, total messages posted, was added to form a single cluster accounting for all four variables.
Table 3 displays the descriptive statistics disaggregated by the 12 courses used in the present study for the 193 study participants, displaying means and standard deviations for all dependent variables. Significant differences were found among courses on the combined dependent measures, Wilks’ Λ = .70, F(33, 528.07) = 2.05, p = .001. The effect size was large, partial η2 = .25. Post hoc ANOVAs on each dependent variable were also conducted. Significant differences were found to exist on each dependent variable: social community, F(11, 181) = 3.86, p < .001, partial η2 = .19; learning community, F(11, 181) = 1.96, p = .04, partial η2 = .11; and perceived learning, F(11, 181) = 2.12, p = .02, partial η2 = .11. No significant differences were noted when the data were disaggregated by the gender of the professor who taught the course.
A one-way MANOVA was conducted to determine the effect of gender on the three dependent variables. Significant difference were found between males and females on all dependent measures, Wilks’ Λ = .96, F(3, 189) = 2.94, p = .04. The effect size was small, partial η2 = .05. Post hoc ANOVAs on each dependent variable were also conducted. Females scored higher than males on each dependent variable: social community, F(1, 191) = 6.73, p = .01, partial η2 = .03; learning community, F(1, 191) = 5.70, p = .02, partial η2 = .03; and perceived learning, F(1, 191) = 5.46, p = .02, partial η2 = .03. All effect sizes were small. Females, therefore, on average felt more connected to other students in their courses, felt that their online learning experience was more aligned to their educational values and goals, and perceived they learned more than their male peers.
Descriptive Statistics Disaggregated by Course
| Course Number | n | Social Ccommunity | Learning Community | Perceived Learning |
|---|---|---|---|---|
| 1 | 15 | 30.20 (5.76) | 29.93 (4.08) | 7.80 (0.86) |
| 2 | 18 | 23.67 (5.63) | 27.94 (4.28) | 6.72 (1.84) |
| 3 | 23 | 25.30 (6.91) | 28.83 (4.59) | 6.87 (1.46) |
| 4 | 15 | 28.20 (3.93) | 31.27 (4.59) | 7.53 (1.36) |
| 5 | 16 | 23.75 (5.54) | 26.38 (4.69) | 5.94 (2.24) |
| 6 | 18 | 30.11 (5.37) | 31.06 (4.87) | 7.28 (1.67) |
| 7 | 15 | 30.80 (5.78) | 31.40 (3.00) | 8.00 (1.31) |
| 8 | 17 | 29.00 (7.66) | 29.82 (5.03) | 7.29 (1.31) |
| 9 | 8 | 24.63 (4.00) | 26.50 (3.93) | 6.88 (0.64) |
| 10 | 13 | 32.46 (4.98) | 29.31 (4.72) | 7.46 (1.61) |
| 11 | 15 | 31.00 (5.31) | 28.80 (5.78) | 7.67 (1.18) |
| 12 | 20 | 27.00 (8.36) | 30.45 (4.73) | 7.28 (1.65) |
Note: Displayed are sample size (n) and means with standard deviations enclosed in parentheses.
A discriminant analysis was also conducted to determine whether the three measures could discriminate between female and male students. The test was significant, Wilks’ Λ = .96, χ2(3, N = 193) = 8.64, p = .004, suggesting that overall the predictors differentiated between females and males. Standardized canonical discriminant function coefficients in order of importance in discriminating between the two groups were social community, .53; perceived learning, .37; and learning community, .31. Predicting gender based on the results of this analysis yielded a correct overall prediction of 65.3% of the cases. By gender, 64.5% of the males and 65.4% of the females were correctly classified. To evaluate how well the classification procedure would predict in a new sample, we estimated the percent of students accurately classified using the leave-one-out technique and correctly classified 63.2% of the cases.
Discussion
The primary intent of the present study was to determine whether men and women differed in their sense of classroom community and levels of perceived learning. The results of this study suggest that indeed these factors differed by gender, with female students consistently scoring higher on each of these measures. Secondarily, the results of this study revealed significant positive correlations between every variable pair under consideration. In particular, the dendrogram at Figure 1 reveals the close relationship of classroom community and perceived learning. A number of implications for research and practice can be derived from these results.
Uniformly higher results demonstrated by females along the self-report variables of sense of classroom community and perceived learning suggest that the asynchronous online environment is one that can be hospitable to women. Additionally, in the present study females posted significantly more messages in course discussion boards than did male students. This result is not consistent with research that suggests males tend to dominate conversations with females due to their desire to achieve a one-up position in the social environment (Tannen, 1991) and to achieve status (Maltz & Borker, 1982). However, this finding is consistent with the research reported by Herring (1999, 2000) that argued female students often participate more than male students in online discussion boards in which the professor, even when the professor is a male, acts as moderator entrusted with maintaining civility and focus in the group.
While this result may seem initially puzzling—how can women be “freer” to participate when they are “controlled” by a group leader?—it makes sense if the leader’s role is seen as one of ensuring a civil environment, free from threats of disruption and harassment. The need for such insurance, rather than reflecting a weakness on the part of women, points to the fundamental failure of “self-regulating” democracy on the Internet to produce anything like equitable participation: when left to its own devices, libertarianism favors the most aggressive individuals, who tend to be male. (Herring, 2000, p. 5)
The promotion of online and distance education as a benefit to women—according to Kramarae (2003) women have been the majority of students in correspondence courses— appears to have been based on the idea that women (particularly married women with children) would be unable to attend traditional university courses but could fit distance education into their busy lives. This study, however, demonstrates that the benefit of online education isn’t merely access but includes educational effectiveness. Women participated in the courses at higher rates than male students and identified their experience as socially richer (as evidenced from the sense of community) and educationally more effective (as evidenced by perceived learning) than men.
The female students who participated in this study appeared to thrive in the asynchronous online environment. Perhaps the discussion-rich approach to online learning evidenced in these courses is particularly wellsuited to females who appeared to be more responsive to social reinforcement and extrinsic motivation than did male students. In a study of online MBA students, Arbaugh (2000) found that women participated more than men in class discussions in an online graduate business course and concluded that men were engaged in the competitive aspects of discussion while the women were more collaborative in their approach. Since collaborative learning and supportive discussion (i.e., constructivism rather than instructionism) are often emphasized in faculty training at the university that provided the participants for the present study, it may be that the instructional design and pedagogical style found in these courses were more harmonious with the communication preferences of female students. Moreover, since female students were in the majority in all courses, the female voice was the predominant one in course-wide interpersonal communications.
A relationally supportive environment is evident in the courses under consideration as reflected not only in the overall sense of community scores but also in the moderate positive correlations between classroom community and perceived learning. This result is consistent with previous research that emphasizes the pedagogical value of online learning communities (e.g., Moller, 1998; Wegerif, 1998). Instructors who are interested in improving the perceived cognitive learning (and presumably their actual cognitive learning) of their students would be wise to consider activities that would promote a sense of classroom community among the students. Palloff and Pratt (1999) proposed seven basic steps for developing academic communities in online distance classes: “clearly define the group’s purpose, create a distinctive online gathering place, promote effective leadership from within, define norms and a code of conduct, allow for a range of roles, permit and facilitate subgroups, and permit students to resolve their own disputes” (p. 24).
Since this population was mostly female, such a discourse style might have simply been constructed by the dominant group (i.e., women) rather than being deliberately fostered by the online professors. Another alternate explanation might be that the men felt significantly outnumbered in the courses and thus did not engage in dialogue or feel a sense of classroom community at the same level that they would have had the class had a 50-50 gender split, although Rovai (2001) reported similar results in a case study of a single course that consisted of an equal number of female and male students. Perhaps a conscious feeling of being outside the demographic majority had a negative effect on the minority group in the present study, in this case, the men.
These hypotheses raise the question of expectations among online learners. It is possible that some online students, particularly those with a more independent and autonomous learning style preference, may not be interested in developing a sense of classroom community and might consider doing so counterproductive to their own learning or not worth the time required to engage in meaningful interactions. Perhaps a significant percentage of men choosing online instruction are less interested in the collaborative and relational opportunities afforded by the medium and thus report lower levels of learning in environments where such is emphasized. Although it’s naïve to ascribe motives and expectations strictly based on gender, these results identify the need for additional gender-based study related to learner expectations in asynchronous online courses.
The discriminant analysis revealed a significant function that differentiated between male and female students in these online courses. Specifically, the standardized discriminant coefficients revealed that feelings of social community were particularly important in determining the gender classification. Female students not only scored higher in social community, but also in the number of messages they posted in their course discussion boards. These two variables could conceivably be grouped together into a construct which the researchers labeled “socially engaged interaction.” Socially engaged interaction reflects students who actively participate in a course through online class discussion and seek both increased knowledge and interpersonal community. Arguably such learners view the course itself as a contentand relationally-rich learning experience and seek benefits in both areas. By considering such socially engaged participation, one can not only differentiate between the male and female students in this study, but one can conclude that the female students were more socially engaged interactive learners than their male counterparts.
However, the discriminant analysis also revealed that each gender is not homogeneous with regard to the dependent measures examined in the present study. Other factors and individual characteristics come into play. The significant differences in outcomes among the 12 separate courses also suggest the presence of additional factors. Moreover, research suggests that gender-linked behaviors and cognition are variable (e.g., Eagly, 1987). Accordingly, further research is needed that examines the role of online course design and pedagogy in promoting classroom community and cognitive learning as well as any changes in group dynamics and gender-linked behaviors in online courses in which males are in the majority and in which the separate, independent voice is established as the group norm. This latter point is particularly important in light of the research of Cross and Madson (1997) that suggested females are more context sensitive than males. Finally, the role of occupation also needs to be examined as all participants in the present study were educators, a profession that includes more women than men. Perhaps the specific occupational roles occupied by men and women helped guide their social behavior in the e-learning environment.
Conclusions
This study revealed the presence of genderrelated differences in participation and perceptions in the online classroom. It provided empirical support for the idea that men and women communicate at different levels, perceive community differently, and have differing views of perceived learning in an online educational environment. The ability to generalize findings beyond the present study is limited because only one university was represented, graduate education courses in which females were in the majority were sampled, and the learner characteristics, course design and pedagogy may not be representative of other university settings. Future research might examine the gender differences in other online instructional settings as well attempt to determine whether such differences are primarily driven by learner expectations and preferences or have more complex psychosocial roots.
