Abstract:

Productive social interaction in computer-supported collaborative learning (CSCL) groups is important as only then groups will perform well and achieve their intended goals. However, putting students in groups and telling them to collaborate does not guarantee that productive social interaction is taking place. This is even more true in online CSCL: even when the virtual learning environment (VLE) environment provide an abundance of computer-mediated communication (CMC) tools to communicate, it may not necessarily lead to increased social interaction. For that reason, CSCL researchers propose to structure the group learning processes by means of specific pedagogical techniques to ensure productive social interaction. The question arises what is to be understood under productive social interaction and how can it be measured? The paper will outline the characteristics of productive social interaction and present a preliminary Productivity Scale using a polytomous scale with six rating scale steps (absolutely untrue, quite untrue, untrue, true, quite true, absolutely true). Sample items are: ‘In our group we criticize each other’s ideas,’ and ‘In our group, we ask questions to each other.’ The Productivity scale was validated using Winsteps version 4.8.0 that implements the Rasch Measurement Model. Sixty-one participants were recruited from distance courses at the Open Universiteit of the Netherlands. the Rasch literature pointed out that we need at least 50 participants to perform a pilot study. Two participants were removed from the Rasch analyses because of extreme scoring or because of not fitting the Rasch Model. Five other persons had also to be removed as they did not score any item. Person separation index was 5.62, which exceeds the minimum value of 2.00 meaning that the items could excellently separate persons (i.e., participants). In classical terms: Cronbach alpha was .97. Item separation index was 2.92, which means that persons could separate items well. The separation index was near to 3.00; in general, item separation values exceeding 3.0 are considered good. All items were in the ‘safe’ zone meaning that all item infit and outfit MNSQ values were between .5 and 1.5. The scale also proved to be unidimensional; the unexplained variance in the first contrast was 4.15 and the Pearson correlation and disattenuated correlation between the first and third item cluster were .70 and .80 respectively. The Rasch analyses showed that the six-step polytomous scale had to be collapsed into a five-step polytomous scale as participants could not differentiate between ‘quite untrue’ and ‘untrue.’ The rating scale steps are now: almost always untrue, regularly untrue, neither true or untrue, regularly true, almost always true. We can conclude that the pilot study demonstrated that the preliminary productivity scale already is an instrument that can be used to assess group productivity.

Keywords:

CSCL, CMC, Rasch measurement model, productive social interaction.