About WTS Prof. Clair Brown Faculty, Students and Affiliates Research Areas Online Research Reports Working Papers


Second Interim Report
Clair Brown, Editor

13. A Systems View of Work Group Performance: An Example from Semiconductor Manufacturing
David M. Bowen

4.0 Strategies for Improving Work Group Performance

The exact variables to include in the model and the specific R2, F-ratios and p-values associated with each model provide an indication as to the relative importance of the factors represented by each variable. However, a greater potential contribution of this research lies in the intelligent use of the multi-level systems perspective as a blueprint for decision-making to facilitate the design and management of work groups. We now discuss use of the model in this capacity, first on a general basis, then as applied to results for the subject work group.

Utilizing the model as a guide for improving group performance leads to five broad categories

of improvement strategies, corresponding to factors at the four levels as shown in Table 1. To determine which category provides the greatest leverage for improvement for a given group requires an analysis whereby the performance of the group is plotted alongside that predicted by the model. The preferred improvement strategy then depends on where the actual performance is located with respect to the predicted performance.

For example, consider the hypothetical group performance depicted in Figure 6. If actual group performance occurs at point A, then an 'increasing KSA' strategy is supported. By contrast, if actual performance where at point D, additional expenditures aimed at increasing KSA would not result in significant improvement in performance. Given the latter (point D), attention to increasing the upper limit by altering the system and/or external environment would likely be more beneficial.

Extending the analysis to the rest of Figure 6, performance at point E3 indicates the need to identify and remove inhibitors to performance, i.e., to determine what is causing the group's KSA to be underutilized and/or misapplied and take corrective action. Performance at point C indicates opportunities for identifying, facilitating and (if there is more than one work group) transferring group performance catalysts, i.e., to determine what is allowing the group to perform beyond expectations, then nurture and institutionalize it.


From the site visit interviews it is clear that the subject fab is committed to a policy of continually increasing the KSA of its manufacturing employees through training and certification. Over time, there is a significant increase in the KSA per person (see Figure 7). The downward fluctuations in Figure 7 are the result of hiring new employees with no documented KSA. If there were only minimal training for the employees, we would expect the graph in Figure 7 to show these dips followed by a transition period while the new employee is trained to the level of the other group members. As seen in the graph, there is instead an increasing average over time.

Comparison of actual group performance to that predicted by the model (see Figure 3B), suggests that increasing group KSA is an appropriate strategy for improving group performance given the current work system. This strategy appears to work well, as substantial 'continuous improvements' in performance occurred simultaneously with increases in group KSA over the twenty-three week study period. However, focus on this improvement strategy exclusively has long term limitations which need to be monitored due to the diminishing returns to additional investment in training for a given work system.

It appears that the strategy of increasing group KSA will be viable in the present work system at the subject fab for some time. The illustrative sample calculation of the" saturation point," though utilizing a number of rough estimates for parameters, calculates a saturation point more than triple the current level of training.

The data indicates that other strategies may provide attractive opportunities for improvement. Analysis of group level factors indicates opportunities for performance improvement via investigation of the causes driving the differences in performance in three versus four day work weeks. Similarly, the dynamics of adding new members to the group warrants investigation. Such investigations may reveal relatively simple, cost effective opportunities for improving group performance by institutionalizing performance catalysts.


Making group performance data public to all the work groups in a given organization can serve as the starting point for allowing the groups themselves to identify performance catalysts and transfer them within the organization. When the performance of one group is consistently above the performance of the others, the lower performing groups will know; a) their performance is less than what it could be, and b) which group to go to for guidance.

Comparison of groups from different organizations can also provide insight. Such information is commonly shared among different organizations in the form of 'benchmarking studies.' Some of the specific catalysts observed in work groups in different contexts may not be transferable to work groups in other organizations, e.g., because of their reliance on the specific technologies in use. However, there are likely to be practices (e.g., running an efficient meeting, a
suggestion system, organization of committees, intra and inter -group communication tools, etc.) that are transferable.


When group-members are performing near the limit of the system, they are in a unique and advantageous position to understand what is limiting performance, and therefore able to suggest and implement system changes which will result in performance improvements. Consequently groups performing near the system's upper performance limit indicate a situation where an increase in croup autonomy over task and work system could prove beneficial. The performance of the work group under investigation is not necessarily close enough to the upper limit to warrant chances in group autonomy at present. Presumably, with increased training and experience, performance will eventually approach the upper limit. When this occurs, consideration of increasing, group autonomy as a mechanism for altering the system limit will be appropriate.


Organizations need to constantly identify and eliminate barriers to performance to ensure continuous improvement. Lapses in continuous improvement, or reductions in rate of improvement, can arise when managers come to the conclusion that performance is constrained by an element of the system other than that which is truly constraining performance (e.g., when a lack of training is holding back performance and the manager surmises that it is a problem with facilities, equipment, government regulation, etc.) Such mismatches between what is thought to be constraining performance and what actually is constraining performance will result in misguided investment of critical resources.

The systems view with a focus on group-member KSA presented in this report supplies a good descriptive model for the performance of the work group being studied. Mathematically speaking, its predictive capability is satisfactory. However, we argue that the primary benefit of the KSA model is that it provides a superior tool for identifying which elements of the work system are actually inhibiting performance improvement. Such identification can enhance an organization's decision making capabilities for managing the performance of work groups and work systems.

Minimal data, specifically the history of the performance and KSA acquisition of group, proves sufficient for constructing the model of group performance. Given only this minimal data, the KSA model can explain 80% of the variance in work group performance, and also provides:

1) an estimate of the upper limit of group performance given the current work system,

2) a basis for estimating current and past system efficiency (utilizing 1),

3) a method for estimating changes in performance expected from changes in group member KSA,

4) a method for determining what level factors are constraining system performance, and

5) a method for determining the break-even or 'saturation point' of croup KSA in a given system

Applied to the work group at the subject fab, the model confirms the company's policy of continually increasing the KSA of manufacturing personnel as a viable strategy for improving performance in the current work system. This is a general result in that the model does not specify who should receive what training when, only that on a group-wide level, training leads to improved group performance. Questions of whether cross training in skills already represented in the group would provide as great a benefit as training to bring new skills into the group are not addressed in the current level of analysis.

Our results suggest that opportunities for performance improvement exist via increasing group member KSA, investigation of three- versus four-day-work-week practices, and investigation of the dynamics of adding members to the group.

End of Chapter 13

Go to References for this Chapter
Go to Next Chapter
Go to Table of Contents for the CSM-HR Interim Report


© 2005 Institute for Research on Labor and Employment. 
2521 Channing Way # 5555 
Berkeley, CA 94720-5555