Performance and Process Measure Relationships in Transitioning from a Low to High Fidelity Simulation Environment

Abstract

Entin (1999) reported that after sufficient training with a six-node non-traditional optimized C2 architecture, teams performed higher with it than with a six-node traditional non-optimized C2 architecture. Moreover, teams performing with a four-node non-traditional optimized C2 architecture performed as well as the six-node traditional non-optimized C2 architecture. Entin (1999) also noted that the various team process measures supported the performance outcome and predicted performance.

The prior work was performed employing the low fidelity Distributed Dynamic Decision making III (DDD-III) research simulator. The DDD offers a high degree of experimental control, on-line data collection, as well as a high level of abstraction. Part of the charge of the Adaptive Architecture for Command and Control project is to investigate whether findings ascertained within the low fidelity DDD environment can be replicated in a high fidelity simulation environment. The current study was a partial response to this charge.

The scenario and forces used with the DDD experiment were adapted to the marine's MAGTIF Tactical Warfare Simulation (MTWS) environment. Because of the watershed nature of this study several technical issues were examined (e.g., the effects of the presence or absence of trained operators and the benefit of operator training). Further resource limitations precluded replicating the earlier design, but it was possible to examine the team process and performance relationship for the four-node — six-node comparison.

Nineteen officers enrolled at the Naval Postgraduate School were partitioned into five four-node teams that participated in eight trials and three six-node teams that participated in six trials. Twelve participants received intensive MTWS operator training and all received general MTWS familiarization training.

Results showed that as in the previous experiment, teams employing the four-node optimized architecture where able to perform at the same level as teams using the six-node non-optimized architecture. It would seem that the optimization of the four-node architecture is able to compensate for the 33% reduction in team staff.

In the previous experiment the team process measures significantly predicted performance outcome. The regression analysis revealed a multiple R of .86 (p<.02), that accounts for 73% of the variance. Findings for the current experiment revealed a similar relationship and pattern. The multiple R is .82 (p<.03), accounting for 68% of the variance. Moreover, the process measures in the equations were quite similar across the two experiments. We conclude that we were successful in transitioning the scenario and mission from a low to high fidelity simulation environment.