Conflict Is Good

Task and relationship conflict were evaluated with eight items (four for each type of conflict) from an intrateam conflict scale. Sample items for task conflict include: "How often are there differences of opinions regarding the task in your team", or "How often do people in your work group disagree about the work being done", for relationship conflict: "How often are personality clashes present in your group" or "How much emotional conflict is there in your work group". The answers were recorded on a 5 point Likert scale (from 1 = never to 5 = very often). The Cronbach’s alpha for task conflict scale was 0.73, and for relationship conflict was 0.80. The values obtained on our sample are consistent with previous studies, which reported slightly lower coefficients for task than for relationship conflict items.

Teamwork quality is a multidimensional construct comprising coordination, planning and quality of communication. Coordination was evaluated with five items adapted from Curşeu et al. (e.g., "The group members have synchronized their actions in order to reach the group goals"), planning was also evaluated with five items adapted from Curşeu et al. (e.g., "The group has developed its own strategy in approaching this project") and the quality of communication was evaluated using four items adapted from Eby et al. (e.g., "During the debates, each team member has been carefully listened to by the others"). The answers were recorded on a 5 point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree.

Previous research on teamwork quality and group processes showed that scales evaluating these constructs often have a unitary factor structure, therefore a principal component analysis with the three teamwork processes was performed. The results showed a unitary factor structure, the main factor accounting for 81% of the scores variance with the following factor loadings: coordination = 0.86, planning = 0.80 and communication = 0.87. Due to these results as well as to the small sample size (at the group level), the three teamwork processes were aggregated into a unitary variable. The Cronbach’s alpha for the scale was 0.84 and the descriptive statistics for this aggregated variable are presented in Table 1.

Leadership styles were evaluated using a scale developed based on the Leader Behavior Description Questionnaire (LBDQ). Relations oriented leadership was evaluated using four items ("The tutor listen to the group members", "The tutor stimulate the development of informal relationships within the group", "The tutor is friendly and approachable", and "The tutor treats all group members as equals"). Task oriented leadership was also evaluated using four items ("The tutor assigns clear tasks to each group member", "The tutor makes sure that the group project is done in a timely and efficient manner", "The tutor focuses attention on irregularities, mistakes, exceptions and deviations from what each group member is expected to do" and "The tutor coordinates the activities of the group members so that the final goal of the group is achieved"). Each respondent was asked to rate his/her leader on a five-point Likert scale with answer categories ranging from 1 = strongly disagree to 5 = strongly agree. The Cronbach’s alpha for relations oriented leadership was 0.88 and for task oriented leadership was 0.82.

To justify aggregation into group scores, we used the procedure introduced by James, Demaree, and Wolf to estimate the inter-rater reliability (the index of agreement). For multiple-item scales assuming a number of J parallel items the formula is:

$r_{w g(J)} \frac{J\left[1- \overline{s_{x_{j}}^{2}} / \sigma_{E}^{2}\right]}{J\left[1- \overline {s_{x_{j}}^{2}} / \sigma_{E}^{2}\right]+\left(\overline{s_{x_{j}}^{2}} / \sigma_{E}^{2}\right)}$

Where $\overline {s_{x_{j}}^{2}}$ is the mean of item variance and $\sigma_{EU}^{2}$ varies as a function of the assumed variance. For an assumed uniform distribution:  $\sigma_{E U}^{2}=\left(A^{2}-1\right) / 12$, and for an assumed normal distribution: $\sigma_{E N}^{2}=[(A-1) / 6]^{2}$ and A is the number of response options (e.g., intervals on the Likert scale). The within group agreement index (Rwg) can take values between zero and one, and generally, a value of 0.70 or higher is considered to reflect a reasonable amount of agreement within a team. Table 1 summarizes the Rwg for each variable.

Table 1. Within Group Agreement Indices (Rwg).

After the within-group agreement was computed and verified, the individual scores of the group members were aggregated into group scores by computing the group mean.