The Impact of Group Membership on Collaborative Learning with Wikis

Introduction

The development of the internet into a social web changes the way people handle information.^1,2 Wikis, ³ blogs,^4–7 social network sites (SNSs),^8,9 and other social software tools increasingly play a role in the context of learning, education, and academic achievement^10,11 and help stimulate learning through collaboration. ¹² The availability of social software tools enables large and heterogeneous groups to collaborate. ¹³ Especially wikis have great potential for encouraging learners to exchange their knowledge with others, acquire new knowledge, and be an active part of a knowledge-building community.^14,15

The social web is as heterogeneous as the real world and brings together people of different groups, affiliations, interests, or races. In many cases, the information in the social web is associated with information about its author: A blog is conceptualized as a personal journal of an author, ⁶ SNSs provide information about authors, ⁸ and in a wiki, which might seem to be an unidentifiable mixture of different authors' information, authors and their contributions are identifiable by the version history and authors' names. ¹⁶ Even if users do not display names or pictures that inform others about sex or race, and even if users choose to disguise their identity by nicknames and avatars, their choices often reveal personal characteristics that leave impressions. ¹⁷ In addition, the artifacts produced by authors (e.g., posts, links, and comments) reveal attitudes and might foster the deduction of underlying opinions, group memberships, or personal characteristics. Thus, most information in the social web is presented together with information about the author directly or indirectly, so that deductions about author characteristics are likely.

In wikis, it is not the authors, but the information that takes center stage. However, as we have explained that the deduction of personal characteristics of authors can hardly be avoided by readers, the current research investigates whether learners who use wikis are affected by indirect information about previous authors. There is evidence that certain information, namely, information about authors' group memberships, has an impact on information processing. Thus, we assume that group-membership information may also influence collaboration and learning processes when people use wikis.

Literature Review and Hypotheses

According to a social comparison theory, ¹⁸ relevant others help individuals to define and validate reality. Information from individuals that share a social category (i.e., ingroup members) are particularly relevant, because they are perceived as more trustworthy^19,20 and more independent in their judgment ²¹ than outgroup members. Similarly, the self-categorization theory ²² proposes that members of a salient group perceive themselves as interchangeable with other ingroup members. Consequently, divergences between one's own view and the ingroup view violates the expectation of a shared view^22,23 and, thereby, elicits active processing of information. ²⁴ Taken together, this theorizing proposes that ingroup information should induce learning more readily than outgroup information.

Persuasion research has indeed demonstrated that messages from ingroups were more persuasive than outgroup messages. More specifically, if the message concerns a topic that is relevant to one's group, ingroup information profits from more intensive processing (i.e., longer processing times and different levels of persuasion due to weak or strong messages), whereas outgroup messages are not processed deeply and are more likely to be rejected. ²⁵ However, outgroup messages have this disadvantage in processing depth, ²⁶ agreement, and recall ²⁷ only when the groups are salient. It is noteworthy that a message does not have to be prototypical or even relevant to the topic of the intergroup situation, the ingroup will still prefer a message coming from an ingroup member. If the topic itself is irrelevant to the ingroup, ingroup messages are not elaborated upon, but still readily accepted, whereas outgroup messages remain rejected. ²⁵

In sum, research has demonstrated that information profits from coming from the ingroup: it is more deeply processed and is more readily accepted than outgroup information. In other words, as soon as relevant group memberships are involved in situations in which information is shared, assessed, and integrated, information from an ingroup has advantages over outgroup information. So far, this effect has been demonstrated for messages and arguments on recall, elaboration, and persuasion of information. It has not been tested, however, whether learning—in terms of integrating new information into prior knowledge and in terms of an increase of factual knowledge—is also affected by group membership.* The effects of group memberships on knowledge integration and the acquisition of factual knowledge have not been investigated for collaborative learning in the social web. When using the social web, it is almost impossible to avoid personal information about authors, which often includes references to social categories (e.g., gender or religious affiliations). In collaborative wikis, this information about the authors might affect the way the wiki information triggers learning. On the basis of previous group membership research, we state the following hypotheses:

H1: Ingroup information in a wiki leads to stronger knowledge integration (i.e., stronger integration of new wiki information into one's own prior knowledge) than outgroup information.

H2: Ingroup information in a wiki leads to a stronger increase in factual knowledge than outgroup information.

Method

Results

We expected main effects of group on knowledge integration (H1) and factual knowledge (H2). To test these predictions, we conducted separate t-tests for the dependent measures. As expected, for knowledge integration, ingroup information was more strongly related to prior knowledge (M=1.71, SD=1.84) than outgroup information (M=0.78, SD=1.17), t(55.61)=−2.51, p=0.015, d=0.62. Likewise, for factual knowledge, wiki information from the ingroup led to more factual knowledge (M=8.85, SD=1.54) than information from the outgroup (M=8.00, SD=1.85), t(68)=− 2.09, p=0.040, d=0.51 (see Table 1).

It is likely that the increase of factual knowledge is more effective if learners integrate new knowledge into existing knowledge structures. We, therefore, explore whether knowledge integration mediated the effect of group on the development of factual knowledge by mediation analyses. ³² When regressing group only, factual knowledge was stronger when individuals encountered an ingroup wiki (β=0.25, p=0.040), but this effect ceased upon including knowledge integration in the regression (β=0.18, p=0.139); whereas knowledge integration increased factual knowledge to marginal significance (β=0.22, p=0.074, see Fig. 1).

OPEN IN VIEWER

FIG. 1.

Mediation analysis of the indirect effect of group on factual knowledge via knowledge integration (N=70).

The bootstrap confidence intervals based on 1,000 bootstrap samples confirmed the indirect effect of group on factual knowledge via knowledge integration, b=0.11, SE=0.07, CI α=0.05 [0.087; 0.306].

Discussion

The current study tested whether group memberships of previous authors (ingroup vs. outgroup) affected collaborative learning with wikis. We expected and found that ingroup information in a wiki induces stronger knowledge integration and the acquisition of more factual knowledge than outgroup information. Moreover, the findings demonstrate that this advantage is due to the fact that ingroups cause superior integration of new information into prior knowledge compared to outgroups in a shared wiki.

The current findings are the first application of group membership effects on learning through collaboration in the social web. They demonstrate that even in a wiki—where information is assumed to be neutral—learners are affected by the group membership of other authors. This finding underlines the fact that learning with wikis does not take place unconditionally or without bias, but that learners have to consider the knowledge presented in a wiki. If the information comes from a doubtful source (e.g., an outgroup member), the information is integrated to a lesser degree into one's own mind and induces less learning. In line with the social comparison theory ¹⁸ and self-categorization theory, ²² information by relevant others has a greater chance of being considered and inducing integration and learning, whereas new information by irrelevant others is disregarded.

It is particularly noteworthy that the group effect occurred in wikis that had a stronger focus on the information itself (and not its authors) than other social software tools. It is also striking that this effect was found even though the topic (i.e., a medical diagnosis) was totally unrelated to group membership (i.e., soccer or the national team), and even though the group memberships had been manipulated very indirectly. Thus, if group memberships are salient, even unrelated information provided by an outgroup member is disregarded. Considering that we found this effect in a controlled laboratory setting, where participants are usually highly motivated to comply with the instructions, the effect is probably even more pronounced in the overcrowded social web that is full of unreliable information from various sources. If practitioners want to induce collaborative learning with wikis in formal or informal environments, the current findings provide evidence that the context can become a serious boundary condition for learning. Three implications might be derived for practitioners: (1) Consider letting individuals collaborate who already share group memberships (a rather conservative suggestion). (2) Improve understanding of common group membership among collaborative learners. Subgroup memberships should be made nonsalient and the common learning group should be emphasized. (3) Create an unprejudiced atmosphere, where diversity is expected and appreciated. ³³

The current results demonstrate that it is worthwhile investigating the impact of group effects on collaborative learning with wikis. Future studies should include a control group, where no group membership is given. Moreover, findings should be replicated in the field with (unpaid) participants who interact naturally. This would allow testing whether personally relevant topics are also affected by topic-unrelated group memberships, even though existing research ³⁴ would suggest that this is not the case. Our results have implications for learning through collaboration in the social web, as group memberships are omnipresent in names, nicknames, pictures, or avatars and will inevitably play a role. ¹⁷ These implications are even more significant for real wikis or SNSs, where group memberships are often related to the expression of certain attitudes. Expression of attitudes may render the ingroup source even more persuasive ³⁵ and, thus, harden the fronts between the groups. This would reduce the chance that diverse information would stimulate cognitive conflicts which, in turn, induce learning. ¹² Future research should find means to overcome limiting group boundaries and develop mechanisms to stimulate unbiased collaboration with wikis.