IMPROVING TEAM PERFORMANCE IN
SOFTWARE ENGINEERING
Robert Lingard and Elizabeth Berry
California State University,
Northridge
INTRODUCTION
Both educators and industry representatives seem to agree that to be successful in today's workplace high levels of teamwork and communication skills are needed by engineering and computer science graduates (Lewis, 1998). In employer surveys one of the most frequently mentioned skills lacking among CSU, Northridge computer science graduates was the ability to work effectively in teams. Many other universities have made similar observations and have begun efforts to improve the computer science curriculum in this regard. At the University of Massachusetts, Dartmouth curriculum changes have been outlined which place a greater emphasis on the development of large software systems using student teams (University of Massachusetts, No date). At the University of Virginia the lack of an emphasis on team projects has been identified as a major deficiency in their computer science program (University of Virginia, No date). Cal Poly, San Luis Obispo, the University of Texas, Austin and Georgia Tech have also specified teamwork and group projects as essential components of their computer science programs (Upchurch, 1994).
However, even when team projects are assigned in software engineering courses, students seldom receive any training on how to function collaboratively before such assignments are given, and little attention is given to how teams are formed. Consequently, teams often fail to function effectively. According to a study at Brigham Young University (BYU), team process effectiveness was the major factor accounting for the success of group projects (Swan, 1994). Furthermore, students do not learn much from participating on dysfunctional teams and often develop negative views about the value of teamwork (Aldridge, 1996).
Simply assigning more team projects is not sufficient in addressing the need for students to learn teamwork skills. In order for students to benefit from these team projects, efforts must be made to ensure that the teams are well formed and given the knowledge and tools necessary to operate effectively.
DESCRIPTION OF THE STUDY
The main purpose of this effort was to develop techniques for improving team performance within group projects assigned to software engineering students. The first objective was to provide some specific instruction in group communication skills as part of the regular course of instruction, and the second was to improve the process of forming teams in order to maximize group effectiveness.
Last year as a joint effort with the Communication Studies Department at CSU, Northridge, some specific instruction in group communication was given to two of the four sections of Computer Science 380, "Introduction to Software Engineering." The other sections received no such instruction. The instruction consisted of discussing the group process and how teams function. Students participated in an exercise that demonstrated the value of the group process, discussed group members' roles, analyzed their individual talents and those of their group, and assessed the group process at various points in the semester. Students wrote about the problems they were having in completing their group projects and the instructors tried to help them solve those problems.
As for the second objective, groups were evaluated for group synergy using the Kolbe Conative Index (KCI-A®). According to Kolbe (1990), group synergy contributes positively toward group productivity. If true this would suggest that forming groups to maximize synergy would result in groups that worked together more effectively and would, therefore, provide students with a more beneficial group experience.
The KCI-A® is an instrument that measures conation or a person's inherent talent or natural way of doing things and predicts what a person will or will not do, given the freedom to act. Whereas intelligence tests measure I.Q. and personality tests measure values and preferences, the Kolbe index measures the conative, the way people act while trying to achieve goals. It identifies four modes or striving instincts -- Fact Finder, Follow Thru, Quick Start, and Implementor -- each prompting people to act in a certain way. Synergy is a productive balance of instincts within a team. It is derived from a mixture of complementary, conative talents. Ideal synergy involves not only the right mix of instincts to initiate solutions, but the same amount of energy to avoid problems as well. It was this measure of group synergy that was used in this study.
FINDINGS
A survey was given at the end of each semester to all four sections of the course asking students to assess their group experiences. In one of the questions, students were asked to rate the effectiveness of their team on a scale of 1 to 5, where 5 was "extremely effective" and 1 was "extremely ineffective."
Table 1 shows the average response for each team along with the average score on the group projects. The teams F-A through F-F and S-A through S-E were those that received specific group communication instruction, and the teams F-1 through F-5 and S-1 though S-7 were those that received none. The results show a significant correlation between the teams' ratings of their effectiveness and the scores on the projects [r(21) = 0.451, p < 0.025].
One of the main objectives of this study was to examine team composition as a factor in determining team effectiveness. The fourth column in the table below shows the measure of team synergy based on the "conative" assessments of all team members using the Kolbe Conative Index. Each team member was assessed using this instrument, and from that information a measure of group synergy was determined for each team. The synergy is expressed as a percentage where 100% indicates ideal synergy. One hypotheses of this experiment was that higher synergy would result in higher team scores. At first examination the data suggest no such correlation.
Table 1 Team Results
|
Team ID |
Project Scores |
Effectiveness Rating |
Team Synergy |
Test Scores |
|
F-1 |
81.00% |
3.67 |
80.00% |
81.01% |
|
F-2 |
79.00% |
4.00 |
75.00% |
72.93% |
|
F-3 |
66.00% |
2.33 |
60.00% |
76.06% |
|
F-4 |
78.50% |
2.75 |
50.00% |
77.71% |
|
F-5 |
90.50% |
4.50 |
87.00% |
81.77% |
|
F-A |
80.00% |
4.50 |
100.00% |
81.66% |
|
F-B |
81.50% |
3.00 |
60.00% |
77.77% |
|
F-C |
95.50% |
4.00 |
80.00% |
85.25% |
|
F-D |
73.50% |
3.33 |
90.00% |
76.80% |
|
F-E |
90.00% |
3.00 |
83.00% |
78.19% |
|
F-F |
90.50% |
4.00 |
37.00% |
88.07% |
|
S-1 |
78.00% |
4.00 |
80.00% |
76.45% |
|
S-2 |
88.00% |
4.25 |
50.00% |
88.19% |
|
S-3 |
82.50% |
4.00 |
80.00% |
83.30% |
|
S-4 |
79.50% |
3.60 |
40.00% |
79.20% |
|
S-5 |
84.00% |
3.80 |
80.00% |
84.25% |
|
S-6 |
72.50% |
3.50 |
42.00% |
79.63% |
|
S-7 |
75.00% |
4.00 |
80.00% |
81.10% |
|
S-A |
81.00% |
4.40 |
80.00% |
75.10% |
|
S-B |
80.50% |
2.33 |
80.00% |
74.86% |
|
S-C |
73.50% |
3.50 |
75.00% |
77.94% |
|
S-D |
85.50% |
4.40 |
90.00% |
78.55% |
|
S-E |
86.50% |
4.20 |
80.00% |
78.85% |
There was, however, a very significant correlation between Project Scores and the combined test scores of the of the team members. Column five of Table 1 shows the average test scores from the mid-term and final exams for the members of each team. This correlation is significant at the 0.005 level [r(21) = 0.564, p < 0.005]. This result is hardly surprising. It suggests that teams made up of students who do well on tests do well on group projects. However, this correlation is so high that it might be obscuring any effect of synergy on group achievement. If we consider the population of teams without exceptional high or low test scores, in particular, only those teams within two standard deviations of the mean, there is a significant correlation between group synergy and project scores [r(19) = 0.380, p < 0.05]. With these more cognitively balanced teams there is also a high correlation between each team's rating of its own effectiveness and team synergy [r(19) = 0.478, p < 0.025]. Further study using cognitively balanced teams is necessary to validate this result.
In addition to studying group formation as a means to improve team performance, efforts were made to provide specific instruction in group communication as part of the regular software engineering curriculum. Another hypothesis of the experiment was that the classes receiving this instruction would do better on group projects and rate their group experiences more positively. Although the data, as summarized in Table 2, show higher project scores for the classes receiving the instruction, the difference is not statistically significant for this relatively small sample. The results are encouraging, but more study is needed to accurately determine the effect of such instruction on group performance.
Table 2 Effect of Group Communication Instruction
|
Class |
Project Scores |
Effectiveness |
Team Synergy |
Test Scores |
|
F98 - No Instr. |
79.00% |
3.45 |
70.40% |
77.90% |
|
F98 - w/Instr. |
85.17% |
3.64 |
75.00% |
81.29% |
|
S99 - No Instr. |
79.93% |
3.88 |
64.57% |
81.73% |
|
S99 - w/Instr. |
81.40% |
3.77 |
81.00% |
77.06% |
|
Total - No Instr. |
79.68% |
3.67 |
65.82% |
80.47% |
|
Total - w/Instr. |
83.45% |
3.70 |
77.73% |
79.37% |
CONCLUSIONS AND RECOMMENDATIONS
The results of this experiment support the notion that there is a direct relationship between group effectiveness and project success. The results also suggest that group project achievement can be improved by taking more care in the way teams are formed. High team synergy appeared to contribute to group achievement. While the results of this initial experiment are encouraging with respect to providing instruction in group communication, further work is needed to establish whether this correlates significantly with group performance.
The experiment clearly showed a strong relationship between achievement on tests by team members and group project success. This suggests that to be fair in forming groups, the groups should be balanced with respect to knowledge of software engineering. This would help to ensure that all teams would have an equal chance to do well and would improve the likelihood that all students could participate in a successful team experience.
In summary, the results of this experimental effort strongly suggest that more attention should be given to the process of forming groups. Rather than forming groups in some arbitrary fashion or just allowing students to form their own groups, groups should be formed carefully, taking into account both the cognitive skills and the conative talents of the individual group members. In this way the best possible group performance can be assured which will provide students with the greatest opportunity to learn software engineering.
REFERENCES
Aldridge, M. D., et al. (1996). Introduction to team-based design for students in engineering, business, and industrial design. Final report to the National Science Foundation.
Kolbe, K. (1990). The conative connection. Reading, MA: Addison Wesley.
Lewis, P., et al. (1998). Assessing teaming skills acquisition on undergraduate project teams. Journal of Engineering Education, 87(2), 149-155.
Swan, B. R., et al. (1994). A preliminary analysis of factors affecting engineering design team performance. Proceedings, 1994 ASEE Annual Conference (pp. 2572-2589).
University of Massachusetts, Dartmouth. (No date). Putting software process in computer science education. Proposal to the National Science Foundation, Award No. DUE-9555042 [On-line]. Available: http://www2.umassd.edu/SWPI/nsfccd/nsfccd.html [1999, November 30].
University of Virginia, Charlottesville, Virginia, Department of Computer Science, School of Engineering. (No date). Curriculum [On-line]. Available: http://www.cs.virginia.edu/curriculum/ [1999, November 30].
Upchurch, R. (1996). Software process education [On-line]. Available: http://www2.umassd.edu/SWPI/1docs/SPEd.html