We analyze crowdsourcing communities by creating a detailed process for quantifying individual behaviour in online environments. The key feature of our communities is their social interactions so we call these social crowdsourcing communities (SCC). First, we derive factors based on actions captured about textual contributions. We interpret and name these factors. Then we demonstrate their utility in predicting the quality of team contributions. We capture the actions of members using measurable variables and perform factor analysis on these to produce factors of behaviour in SCCs (i.e. dimensions of behaviour). We derive factor scores for each member. An abstract notion of teams is used that is based on the social interactions. Team scores are then determined by the aggregation of the individual factor scores. The relationship between the team-level factor scores and the quality of contributions made by each team are then used as a proxy for team effectiveness. We found that member behaviour has three dimensions/factors: Impact, Activity, Policing/Rowdiness and there is a linear relationship between a team's contribution quality and their Impact scores. We also found a moderate negative linear relationship between the smallest Activity scores in each team with the quality of their individual contributions. This shows that teams that produce higher quality contributions tend to have higher total and maximum Impact score with lower levels of Activity. Thus, we demonstrate that properly aggregated behavioural factors can predict the quality of team-level contributions.