Do two heads search better than one? Effects of student collaboration on web search behaviour and search outcomes

This study compared Pairs of students with Single students in web search tasks. The underlying hypothesis was that peer-to-peer collaboration encourages students to articulate their thoughts, which in turn has a facilitative effect on the regulation of the search process as well as search outcomes. Both hypotheses were supported by the results. Pairs located the target information more often and in less time than Singles did. Pairs also employed a richer repertoire of search strategies and were more proficient in monitoring and evaluating their search behaviour. Implications of these findings for practice and further research are discussed. Introduction About thirty years ago, information seeking was a specialism and librarians could call themselves information specialists for a reason. People in need of specific information generally consulted a librarian who had access to various information retrieval systems and was able to reveal its contents by developing sophisticated queries. The librarian would sit down with the information seeker to identify his/her information needs. Taylor (1968) classified this presearch interview as a negotiation process in which the information specialist works with the inquirer back to his/her information needs and then translates these needs into a useful search strategy. Nowadays, most public and research library catalogues are available online, and the Internet hosts tons of information that is readily accessible to the general public. Internet technology has also found its way into schools where, among other things, the web is being used as an arena for students to obtain information. Asking students to collect their own learning materials is consistent with contemporary notions of 466 British Journal of Educational Technology Vol 36 No 3 2005 © British Educational Communications and Technology Agency, 2005. learning that advocate knowledge building through constructive, self-directed activities. The web is therefore considered a potentially powerful means to mediate this new form of learning. The advantages of online public access notwithstanding, most students are unlikely to locate information efficiently and effectively. One reason is that the web suffers from a lack of indexing conventions. Controlled vocabularies, thesaurus classifications, and human indexing are uncommon for web search engines. More importantly, the web opens up so much information that designated skills to manage the information flow are called for. Students generally lack these self-regulatory skills, which caused about half of the high school freshmen from various European countries to consider themselves incapable of searching the web efficiently and effectively (Pelgrum, 1999; Ten Brummelhuis & Slotman, 2000). In-depth studies by Branch (2001) and Lazonder (2000) showed that teenagers are largely unable to select appropriate search strategies (planning), check their progress (monitoring) and assess the relevance of search outcomes (evaluating). Improving searchers’ abilities to self-regulate seems called for, especially because such regulatory skills are the key to successful web searching. Yet, there is little research on effective web-searching instruction. Colaric (2003) examined three instructional treatments to aid searchers in developing queries (ie, planning). The results were equivocal. Undergraduate students receiving worked examples yielded superior syntactic knowledge than students in the two conceptual model conditions. That is, students in the worked-example condition were more proficient in structuring a search query using terminology that a search engine can interpret correctly. In contrast, conceptual model instruction was the most effective method for increasing the students’ declarative understanding of how a search engine works. Lazonder (2001) compared the instructional efficacy of two versions of self-regulatory skill instruction to a control group. His instruction was more comprehensive than Colaric’s in that it taught students to plan, monitor, and evaluate a search. Results indicated that students from both self-regulatory skill groups needed about 25% more practice time than students in the control group (who were merely taught procedural skills to operate the search engine). However, self-regulatory skill instruction did not enhance search performance on test tasks. A possible explanation for this finding is that students had developed the self-regulatory skills during practice, but failed to recognise when the acquired skills should be applied on the test tasks. The work reported here explored whether collaboration might help overcome this inert knowledge problem. The basic assumption is that experienced web users have at least some understanding of the self-regulatory skills entailed in efficient web searching. The presence of an equally knowledgeable peer might encourage these users to articulate their thoughts. Such verbalisations might have a positive impact on problem solving by supporting a variety of self-regulating functions (Teasley, 1995). That is, peer interaction is assumed to incite users to negotiate the suitability of search strategies and the Effects of student collaboration on web search behaviour and search outcomes 467 © British Educational Communications and Technology Agency, 2005. relevance of search outcomes. Another alleged advantage of collaborative information seeking is the opportunity for users to critically observe and monitor each other’s actions, which may facilitate early detection and correction of errors. Additionally, higher levels of self-regulation are assumed to be associated with better search outcomes. Research on collaborative web searching has recently begun to emerge. Some studies have focused on peripheral issues such as the roles individuals can adopt in collaborative information seeking (Prekop, 2002), attitude changes in intergenerational collaboration (Caskey, 2003), and gender differences in group-based information seeking (Large, Beheshti & Rahman, 2002). Another line of research is directed towards designing interfaces and support tools to foster collaborative information-seeking activities (Chau, Zeng, Chen, Huang & Hendriawan, 2003; Druin et al , 2003; Twidale & Nichols, 1998). Yet, none of these studies report how groups of users (as opposed to individuals) organise and perform their search process, nor do they provide empirical evidence on the alleged advantages of collaborative web searching. Evidence on the effects of collaboration on task outcomes can be gleaned from research in other domains. Collaboration has been used successfully to promote learning across a range of curriculum areas including mathematics, science, reading, and language arts. Its benefits appear to be relatively independent of group size, student age, and ability. The magnitude of these effects was shown in meta-analytical studies, comparing collaborative learning with whole-class instruction or individual learning (eg, Cohen, 1994; Lou, Abrami, Spence & ’d Apollonia, 2001). These studies report substantial facilitative effects of collaborative learning on both group task performance and achievement on individual posttests. Zimbardo, Butler, and Wolfe (2003) further demonstrated that these findings extend to collaborative test taking. Students who prepared for a test individually performed significantly better when they took that test with a fellow student than on their own. This result is relevant to web searching as teachers generally require students to utilise existing search expertise to find information on the web. Practical constraints such as the availability of computers further necessitate teachers to let students perform these searches collaboratively. Very few studies have investigated whether collaboration promotes self-regulation. To illustrate, in Lou et al ’s (2001) meta-analysis on the effects of small-group learning with technology, only 3 of the 122 studies addressed the use of self-regulating strategies. Although the results indicated that collaboration yielded higher frequency of using appropriate strategies, the nature of these strategies remains unknown. Fleming and Alexander (2001) give a more detailed account of the benefits of collaboration. They studied fourth-graders’ strategy use during memory tasks and found that triads used more sophisticated strategies and were more aware of the usefulness of these strategies than did children who worked individually. Despite these favourable results, several properties of this study constrain its generalisation. These include the nature of the target audience, the experimental task, and the use of question prompts to promote the children’s understanding and application of appropriate strategies. Consequently, there 468 British Journal of Educational Technology Vol 36 No 3 2005 © British Educational Communications and Technology Agency, 2005. is no guarantee from Fleming and Alexander’s study that collaboration will foster adolescents’ self-regulation during web searching. Research supports the assumption that self-regulation improves academic achievement. Schauble, Klopfer, and Raghavan (1991) found that higher levels of planning lead to higher achievement. Schraw (1994) reported similar effects for monitoring and evaluating. Other studies have found that self-regulation is a relatively general process that operates in the same fashion across different domains (eg, Veenman, Elshout & Meijer, 1997). Together, these findings substantiate that planning, monitoring, and evaluating can increase achievement on web search tasks. However, as these studies were conducted on individual students, further research is needed to show whether their findings generalise to collaborative web searching. In summary, the cited studies strongly suggest that both collaboration and selfregulation improve search outcomes. Whilst it seems plausible that the benefits of collaboration are at least in part attributable to higher levels of self-regulation, there is insufficient evidence to substantiate this claim. The present study therefore explored whether collaboration stimulates self-regulatory activities in students searching for information on the web. It sought to answer this question by comparing pairs of students with individual students in web search tasks. Pairs were expected to perform these tasks faster and better than single students. Pairs were also expected to show relatively higher proportions of self-regulatory activities such as planning, monitoring, and evaluating.