بهبود بهره وری و خلاقیت در گروه های آنلاین از طریق فرایند مقایسه اجتماعی: شواهد جدید برای طوفان مغزی الکترونیکی آسنکرون

Abstract This article extends the findings in synchronous room-based electronic brainstorming about the impact of social comparison process on productivity and creativity in a web-based context of asynchronous electronic brainstorming. Social comparison was manipulated with a feedback informing group members of their respective contributions on the electronic brainstorming task through a shared table regularly updated by a facilitator. In another group, although participants had the possibility to identify each contribution within the newsgroup, they did not receive any feedback in a shared table. Results showed that both group productivity and group creativity are better in the social comparison feedback condition than in the other condition. It appears that social comparison process has a positive impact on productivity and creativity in a web-based context of asynchronous electronic brainstorming, but only when participants have access to a shared table facilitating the comparison among group members. This finding provides some useful recommendations for learning facilitators to improve productivity and creativity in the context of computer-supported collaborative learning over the Internet. It also invites to future innovative technological developments to improve participation in online groups.

1. Introduction The use of computers and communication systems as support for human cooperative work dates back to the early 1980s (e.g., Huber, 1980). From that date, new forms of computer technology – called Group Support Systems (GSS) – have increasingly developed. Such systems, often called groupware, are defined as computer-based network systems which support group work on a common task and provide a shared interface for participating groups (see Ellis, Gibbs, & Rein, 1991). DeSanctis and Gallupe (1987) defined GSS as combining “communication, computer, and decision technologies to support problem formulation and solution in group meetings” (p. 589). These systems were initially developed for commercial use, usually to support decision making processes in organization. They have been widely used for research on electronic brainstorming in groups in which participants generate ideas on computers (see Fjermestad & Hiltz, 1998, for a review). A large majority of studies examined electronic brainstorming in a synchronous room-based context, although the development of Internet has provided over the last years a web-based context without space nor time constraints for testing new forms of electronic brainstorming. The present study does not only extend electronic brainstorming research, shifting from synchronous room-based context to an asynchronous web-based one, but also provides a new test for the impact of social comparison process on productivity and creativity in on-line groups during an asynchronous electronic brainstorming. 1.1. Room-based studies With electronic GSS, group members communicate by exchanging typed messages, instead of verbally speaking in a meeting room (e.g., Nunamaker, Dennis, Valacich, Vogel, & Georges, 1991). Numerous laboratory and field studies have used GSS to facilitate different types of work and learning methods in different tasks such as communication, planning, voting, negotiation, decision-making, problem-solving, idea generation, and so on. These tasks can be accomplished by using specific instruments such as electronic brainstorming, idea organization, vote, and group writing. GSS's have been essentially used for electronic brainstorming in groups1 and have proven useful in improving group performance, particularly for creativity tasks such as idea generation (e.g., Gallupe, Bastianutti, & Cooper, 1991; Nunamaker, Applegate, & Konsyski, 1987; Pinsonneault, Barki, Gallupe, & Hoppen, 1999). This form of brainstorming consists in groups of various sizes simultaneously and anonymously generating ideas on a specific issue on individual computers located in a same electronic meeting room. Individual computers are connected to a central computer which collects the generated ideas and controls their display on a large screen at the front of the room (or on part of the screen). Participants share ideas through typing them on a network-linked computer (a computer console) which is running group brainstorming software (see GroupSystems, Nunamaker et al., 1991). Because participants have to gather at the same time in a same place, the room-based studies have been limited to small groups. Indeed, among 200 experiments listed by Fjermestad and Hiltz (1998) in their review about GSS, only 4% used groups with 10 members or more. Since then, researchers have tested new forms of electronic brainstorming for larger groups in a room-based context. For example, De Vreede, Briggs, van Duin, and Enserink (2000) separated very large groups into sub-groups which had to generate ideas in a serial mode. Once a sub-group had finished to produce ideas, the next began, and so forth until all the sub-groups had performed the task. Although this technique appeared to be more efficient than a technique in which sub-groups worked in a parallel mode, the sub-group members were involved in a synchronous electronic brainstorming task in a traditional room-based context. Because this technique did not eliminate the time and space constraints, and because there may be more group members than there are GSS workstations in the electronic meeting room, other forms of electronic brainstorming may be developed today over the Internet. 1.2. Web-based studies Although the original GSS's were room-based, with strong time and space constraints, the development of the Internet has enabled new forms of electronic brainstorming providing possibilities for individuals to work with others who are geographically remote. The main advantage of this form of electronic communication is that both temporal and spatial constraints are abolished (e.g., Arrow et al., 1996; Cummings, Schlosser, & Arrow, 1996; Gay & Lentini, 1995; Kiesler & Sproull, 1992). There is a new form of GSS derived from Internet such as web-based systems which incorporate various communication tools, synchronous and asynchronous, and can be used by individuals working together. However, when people work together, there is often a need not only to communicate with others but also to share information. That is the reason why shared workspaces are used in the today GSS for the storage of different kinds of information data such as documents, tables, pictures, URL links to web pages, member contact information, and so on. Although many GSS have evolved to the increasingly popular and ubiquitous Internet applications that enable participants to communicate with other group members over the web, the electronic brainstorming remains dramatically synchronous. However, the synchronous mode of communication of web-based GSS confronts people to serious agenda problem reducing the interest for these systems. Consequently, it is often very difficult to support very large groups in a single web-based GSS meeting although the use of room-based electronic brainstorming plays a key role in enhancing productivity in groups, particularly for larger groups (e.g., Valacich, Dennis, & Connolly, 1994). On the contrary to traditional GSS, asynchronous computer conferencing systems such as newsgroups provide many participants with an opportunity to work together without time nor space constraints. Although Computer-Mediated Communication (CMC) has been used in GSS studies, researchers and practitioners have rarely used a traditional asynchronous computer conferencing, i.e. a newsgroup, to conduct an electronic brainstorming session in a web-based environment. One of the objectives of the present study is to use this mode of communication for electronic brainstorming. Apart from the technological environment in which electronic brainstorming can be used, some studies on electronic brainstorming have been guided by group processes which improve (or impair) performance, respectively for individuals – who firstly separately work (without communicating) and later on pool ideas (i.e. nominal groups) – and persons involved in verbal brainstorming (e.g., Gallupe et al., 1991; Valacich et al., 1994; Ziegler, Diehl, & Zijlstra, 2000). 1.3. Group processes in electronic brainstorming Several dozen group processes have been studied in controlled experiments on group computer-supported brainstorming such as production blocking2 (Diehl & Stroebe, 1991), evaluation apprehension (Brown & Paulus, 1996), social loafing (Karau & Williams, 1995), social comparison (Shepherd, Briggs, Reinig, Yen, & Nunamaker, 1996), and so on. It appeared that one of the most important advantages of electronic brainstorming is that it reduces or eliminates the harmful blocking effects which appear in verbal brainstorming (e.g., Diehl & Stroebe, 1987; Gallupe, Cooper, Grise, & Bastianutti, 1994; Valacich et al., 1994). Indeed, because nobody in the group has to wait for a turn to speak, production blocking is eliminated during a synchronous electronic brainstorming session (e.g., Gallupe et al., 1991; Paulus, Legett-Dugosh, Dzindolet, Coskun, & Putman, 2002). However, production blocking is not eliminated by the synchronous mode of communication per se, but rather by the possibility to generate ideas at any moment into a shared workspace, either synchronously or asynchronously. Another advantage of electronic brainstorming is to reduce evaluation apprehension (e.g., Connolly, Jessup, & Valacich, 1990). The anonymity of electronic brainstorming reduces the threat to be negatively evaluated by other participants. However, this advantage can also be detrimental to productivity in groups because anonymity also encourages participants to loaf. Social loafing is the process by which individuals tend to make less effort in group tasks than they do in individual tasks, unless their individual contributions can be identified (e.g., Diehl & Stroebe, 1987; Latané, Williams, & Harkins, 1979; Shepperd, 1993). Alternative to original electronic brainstorming which promotes anonymity of group members, an electronic brainstorming through a traditional newsgroup may provide an identification of the contributions of group members and, consequently, reduce social loafing. In addition, identification is likely to provide group members with an opportunity to compare to each other. Social comparison processes are known to be an important determinant for group behaviours and performance, even in electronic brainstorming. 1.4. Social comparison processes: Upward comparison and downward matching Since their origin, the social comparison processes have been known to be useful for self-evaluation and self-improvement. Indeed, Festinger, 1954a and Festinger, 1954b theory of social comparison processes postulates that there is a “motivation to know that one's opinions are correct and to know precisely what one is and is not capable of doing” (p. 217). Although people prefer objective standards to evaluate themselves, these are hardly available. Under such circumstances, people compare to each other to assess their standing and they prefer “similar others” to “dissimilar ones”. Consequently, there is a trend to stop comparing oneself to others when they are very different. Regarding ability, Festinger, 1954a and Festinger, 1954b postulates that there is a “unidirectional drive upward” in which people make ability comparisons, in particular with “similar others” who are marginally better than themselves. This upward comparison process is generally a good incentive for self-improvement (e.g., Helgeson & Mickelson, 1995; Wood, 1989). Indeed, viewing others performing slightly better than themselves may lead people to set higher personal standards which can motivate efforts to improve themselves, and by doing so, they improve themselves (e.g., Blanton, Buunk, Gibbons, & Kuyper, 1999; Huguet, Dumas, Monteil, & Genestoux, 2001; Seta, 1982; Vrugt & Koenis, 2002). For example, in an academic context, Blanton et al. (1999) found that the academic performance feedback of Dutch school children tended to improve if they compared their examination grades with high-performing students (see also Huguet et al., 2001 for studies among French school children). Similarly, Vrugt and Koenis (2002) recently showed that upward comparison produced higher personal goals which predicted the future scientific productivity of academic staff members. However, the presence of very productive members in a group could motivate increased performance in the other group members, but only if the downward matching process can be counteracted or minimized ( Paulus et al., 2002). Downward matching is the process by which social comparison leads to lower standards for performance when generating ideas within groups. Indeed, over contributors tend to reduce their contributions to match the group's poor standards, whereas under contributors are challenged to reach the level established by others ( Forsyth, 2000). This process tends to lower overall performance levels, also in groups working via computers (e.g., Paulus, Larey, Putman, Leggett, & Roland, 1996), but it can be reduced by increasing a comparison among group members (e.g., Brown & Paulus, 1996; Paulus & Dzindolet, 1993). Indeed, electronic brainstorming studies in room-based context have demonstrated that group members who could monitor others' ideas production in real time eliminated downward matching during an idea generation task (e.g., Paulus et al., 1996; Roy, Gauvin, & Limayen, 1996; Shepherd et al., 1996). Consequently, brainstorming groups are more productive when they are provided with a continuous public display of the ideas generated by anonymous group members projected at the front of the electronic meeting room. Similarly, in a study in which participants announced to the group how many ideas they generated every 5 min, Paulus et al. (1996) demonstrated that the shared performance feedback increased productivity, compared to the “no feedback” condition. Taken together, these results suggest that techniques which provide a real-time – or a delayed performance feedback – seem particularly useful to improve productivity because they create many opportunities for social comparison within the group. Thus, social comparison process is not only useful to evaluate oneself accurately by viewing the performance of group members, but also to improve one's productivity in group through comparison with (slightly) more productive participants (e.g., Monteil & Huguet, 1999). 1.5. Overview and hypothesis The present research aimed to test the influence of the social comparison process in a web-based electronic brainstorming context in which geographically remote participants asynchronously generate ideas in a newsgroup. Participants were randomly assigned in one of the two gender balanced groups. In one group, social comparison was possible through a feedback which informed participants of their own contributions and those of each member within the group by using a shared table regularly updated. In the second group, treated as a control condition, participants did not receive any feedback in a shared table although they had the possibility to identify contributions of group members in the newsgroup. Consistent with findings on room-based studies (Roy et al., 1996; Shepherd et al., 1996), it was assumed that participants who received a feedback of social comparison through a shared table should be more productive than those who did not receive any feedback. Although there is a lack of empirical evidence, it was also assumed that the same pattern should be observed on group creativity measured through the quality of ideas generated in electronic brainstorming.

3. Results 3.1. Pre-task questionnaire: Familiarity and attitudes towards technologies and group learning Analyses of the pre-task questionnaire revealed that all participants were used to computers, the Internet and to asynchronous technologies for learning (range=1–5; browser: M=4.67, SD=0.7; e-mail: M=4.91, SD=0.4; newsgroup: M=3.27, SD=0.9. They were relatively less familiar with the use of synchronous technologies (range =1–5; chat: M=2.41, SD=1.3; web cam: M=1.67, SD=0.7). Results did not reveal any difference between experimental and control condition on each of these measures, nor on attitudes towards information and communication technologies for distance learning and group learning (see Table 1). It may be noticed that although the attitudes towards information and communication technologies for distance and group learning between the two groups were not significant, they all slightly favored the control group for unclear reasons. Table 1. Mean and standard deviation (in brackets) of attitudes and familiarity with technologies in the two conditions Condition of social comparison Feedback (N=14) No-feedback (N=13) Attitudes ICT for distance learning 3.78 (0.5) 4.01 (0.5) Group learning 3.65 (0.6) 3.81 (0.7) Familiarity with communication technologies Browser 4.69 (0.5) 4.64 (0.9) E-mail 5.00 (0.0) 4.82 (0.6) Newsgroup 3.46 (1.4) 3.09 (1.2) Chat 2.38 (1.3) 2.45 (1.4) Web cam 1.62 (0.7) 1.73 (0.8) Table options 3.2. Task-related measures Three measures examining performance and contributions within groups were explored for statistical analyses: (a) number of ideas generated, (b) number of newsgroup's logs, and (c) time spent reading and writing ideas within the newsgroup. In addition to these measures, group creativity was also assessed in the present study. The distributions of the scores for these measures deviated from normal distributions. The Kolmogorov–Smirnov goodness-of-fit test yielded significant results for all distributions. For the number of participations within the newsgroup: K–S Z=1.06, p<0.04, for the number of ideas generated within the newsgroup: K–S Z=1.4, p<0.03, for the time spent reading and writing ideas within the newsgroup: K–S Z=1.5, p<0.03, for the creativity scores: K–S Z=1.6, p<0.02. These deviations from a normal distribution were also apparent when the distributions were separately tested for the two feedback conditions. Moreover, Levene's test for equality of variances between conditions revealed a lack of homogeneity for each dependent measure, except for creativity. Therefore, employing an analysis of variance (ANOVA) or a t test for these dependent measures would not have been appropriate. Consequently, to examine whether there were differences in the contributions in the two conditions, a Wilcoxon rank-test was used; the Wilcoxon rank-test is like a t test for data which deviate from normal distribution ( Bradley, 1968). The obtained mean rank of each dependent measure in the two conditions for the present study is presented in Table 2 (means and standards deviations are also provided in this table for information). The Wilcoxon rank-test showed that the number of ideas generated within the newsgroup was higher in the feedback of social comparison condition than in the non-feedback condition (Wilcoxon Z=1.9, p<0.05). The same pattern of results was found for creativity: the quality of ideas was higher in the feedback condition than in the non-feedback condition (Wilcoxon Z=2.1, p<0.03). The feedback condition and the non-feedback condition did not differ for other measures (number of participations: Wilcoxon Z=0.9, n.s., and time spent reading and writing ideas: Wilcoxon Z=0.1, n.s.). Correlational analyses revealed that creativity scores were not related with productivity scores (r=0.08, ns) suggesting independence between these measures. Table 2. Mean rank, mean and standard deviation (SD) of each dependent measure in the two conditions of the study Condition of social comparison Feedback (N=14) No feedback (N=13) Significance Mean rank Mean SD Mean rank Mean SD Productivity (number of non-redundant ideas) 16.86 30.57 43.6 10.92 8.52 7.83 p<0.05 Number of logs 15.39 2.21 0.89 12.5 1.77 1.4 ns Time spent reading and writing ideas (min) 14.21 37.00 48.36 13.77 30.38 27.18 ns Creativity (originality of ideas) 17.14 2.58 0.62 10.62 1.97 0.74 p<0.03 Table options 3.3. Additional descriptive data In addition to quantitative data, we also collected some qualitative information in the form of written comments during an on-line debriefing session. Participants were given the opportunity to analyse the performance of the two groups and to discuss the utility of the shared table given in the experimental condition. The content of the discussion between participants suggested that social comparison process was at work during the asynchronous electronic brainstorming when participants had access to a shared table presenting the contribution of group members. However, deeper analyses of the comments revealed positive and negative aspect of the social comparison process (see Table 3). Table 3. Qualitative data (comments about human activities during the asynchronous electronic brainstorming in the feedback condition) Negative comments Positive comments I have had a look at the shared table every time I logged in the newsgroup. I have observed the number of contributions, hence the feeling of assessment by the group. This has not enhanced my contribution at all (CS, female) I remember consulting the table on a very regular basis once I had posted my contribution to assess the group evolution and my position within the group … to see if I was in a good position (GP, male) I have had a look at the shared table three or four times, but viewing the great number of ideas and the time spent by others has been a key to relief (DB, female). The shared table is to me a good incentive just as self-correction or feedback (DA, male) The table seems to be a tool for the facilitator but is by no way a federative element for the group (EE, male) It seems that the shared table has played an emulative role for me (HV, male) I have checked the shared table very closely to assess any contribution in comparison to that of others (BN, male)