To study the extent to which group development can reduce spreadsheet errors, an experiment compared error rates in spreadsheet development by subjects working alone (monads) and by subjects working in groups of three (triads). Impressively, triads made 78% fewer errors than monads. However, this was not as large a reduction as nominal group analysis suggests was possible. Members of triads were satisfied with group development. However, triads whose work went most smoothly, whose members were most satisfied with group interactions, and that had the loosest leadership structure were significantly more likely to make errors than other triads.
Panko, Raymond R. and Halverson, Jr, Richard P.
"An Experiment In Collaborative Spreadsheet Development,"
Journal of the Association for Information Systems: Vol. 2
, Article 4.
Available at: http://aisel.aisnet.org/jais/vol2/iss1/4