Creation of teams in an academic learning environment can be a challenge. Team Machine is a tested, twenty-first century decision-support tool that is purpose built to meet this challenge by optimising the selection of members of teams in order to meet education goals.

Teams have become an important element of the workplace and classroom. The high performance of such groups is thus important to teachers and employers. Researchers have found detractors to performance may be in communication, free riding and doing it all yourself attitudes. Heterogeneity of team members can have competing effects. A diverse membership can provide a diversity of knowledge and perspectives useful for innovation and problem solving. However, a homogeneous membership may have less intra-group conflict.

In the classroom, the student project team is becoming more utilised as a learning method. Perhaps unlike the business world, however, assigning members to teams so that the teams are fair and balanced is often, if not always, critical. Thus, the objective is to assign members to teams in such a way that the learning objectives are accomplished with teams that are similar to each other (in performance potential) but where the membership within each team reflects the diversity of the student body.

The Team Machine Model
Team Machine is based upon the bin-packing approach which seeks to put n items (students) with individual “weights” into k bins (teams) so that each bin has the same totals in weights and number of items. The “weight” is a vector of attributes (e.g., undergraduate grade point average, work experience, GMAT score, or ethnic background) representing a student’s academic performance potential and demographics. To favour balanced teams, the student assignments are chosen to minimise the squared deviation of the average student potential in a team to the average student potential of the student population.  Slight adjustments may be made if the attributes include categorical items such as gender or personality type. The importance of the attributes can also be weighted.

The Team Machine Tool
Team Machine has been operationalised as a Microsoft Excel add-in application. Its graphical user interface and spreadsheet layout are designed to allow the novice user to utilise, with ease, the functions and features of Team Machine – even though the underlying functionalities are mathematically complex. Once relevant data is selected (from spreadsheets), Team Machine automatically computes the necessary equations to utilise the bin-packing approach.  The user may also modify the model and make adjustments based upon their own judgment.

In attempting to make the optimal selection of members for each team, the Team Machine can employ several algorithms such as the People Sequential Heuristic (PSH) – a sorting approach for one or two criteria; the greedy randomised adaptive search procedure (GRASP) for more than two criteria; and the genetic algorithms (GA) where elements of potential solutions are exchanged with elements of other potential solutions or are ‘mutated’ (randomly altered) to eventually arrive at a solution through a ‘natural selection’ process, based upon a mathematical objective function GA uses GRASP results as a starting point while GRASP, in turn, uses PSH results as a starting point. Computing power and time may be factors to consider in selecting algorithms. PSH is simple and fast, while GA, generally the best of the three, requires the most computing power and time.

TM in the field and testing
From the Fall of 2004 through 2012, TM was exclusively used to form teams for incoming classes of the Master of Business Administration (MBA) program at a large US university with positive effects upon students, administrators and faculty.

In addition, team selection by TM was compared to team selection by a team research subject matter expert. Given TM’s superior processing capability, it was hypothesised that TM would be better at creating optimally balanced teams. The performance of the teams for three pre-TM entering MBA classes, where the teams were chosen by a subject matter expert, was compared to the performance of the TM-selected teams for three entering MBA classes. Performance data was based upon team assignments from eight core subjects. The results of a two sample t-test indicated that the higher mean score of the TM selected teams was statistically significant (p ≤ 0.5). To determine if grade inflation was a factor, the overall GPA of the students in the pre-TM and post-TM classes were compared. The research showed that there were no significant effects related to grade inflation. In the classes where TM was applied, there were favorable student comments, both solicited and unsolicited, about the team selection process and team experience.

The results above demonstrated that computer assisted mathematical modeling, more specifically TM, delivers benefits to both students and the school. For the students, team performance increased and qualitative feedback suggested that the student experience improved. As manual team selection is time intensive, there was significant time savings for the faculty and staff members responsible for team selection. TM also allowed for more factors to be considered than what was done by a subject matter expert.

TM is not a decision maker but a decision support tool. To arrive at a satisfactory result, human judgment is still needed to determine the factors to consider and their relative importance. These determinations are often done through an iterative process that involves evaluating each run of TM and understanding trade-offs that occur when focusing on improving the results in a particular dimension.  

While Team Machine has been tested and utilised at the tertiary (primarily MBA) education level, its ability to accept different factors means it may have the potential to be adapted to other educational settings. For instance, for team formation at the primary or secondary school level, TM could utilise factors such as gender, age, academic performance scores and/or socio-economic indicators. TM also offers the potential to further research in teams, such as to develop a better understanding of what factors are the strongest indicators of team effectiveness.

To obtain a copy of TM, interested users may contact Dr Paul K Bergey at: