Chapter 3 presents the results of the first study for the quantitative questionnaire data and the content analysis of the e-mail communication. Since an individuals’ perception of the collaborative group process depends on the activities of all group members, the evaluations are interdependent. This has important implications for the statistical method that can be used for analysis. Multilevel modelling (MLM) appears to be the best suited technique to analyse the questionnaire data and since it has not been applied to a small number of observations, it will be discussed in detail. In addition, results of the analysis of the e-mail communication regarding the amount of coordination and content-focused statements are reported and the content analysis method is described to some extent.

The usefulness of roles to support small group performance can often be read; however, their effect is rarely empirically assessed. This article reports the effects of functional roles on group performance, efficiency, and collaboration during computer-supported collaborative learning. A comparison of 33 questionnaire observations, distributed over 10 groups in two research conditions (role and nonrole) revealed no main effect for performance (grade). A latent variable was interpreted as perceived group efficiency (PGE). Multilevel modeling yielded a positive, marginal effect for PGE. Groups in the role condition appear to be more aware of their efficiency as compared to groups in the nonrole condition, regardless of whether they performed well or poorly. Content analysis reveals more taskcontent focused statements in the role condition; however, this was not as we hypothesized (i.e., the premise that roles decrease coordination). In fact, roles appear to stimulate coordination that simultaneously increases the amount of taskcontent focused statements.