Publication Date

6-1990

Advisor(s) - Committee Chair

Raymond Mendel, Elizabeth Erffmeyer, Karlene Ball

Degree Program

Department of Psychology

Degree Type

Master of Arts

Abstract

A study was conducted to evaluate the construct validity of four measures of Armor combat performance derived through the Simulation Networking (SIMNET) system. Problems with field testing, such as high cost, low reliability, and lack of realism, has lead the Army to look for alternative methods for soldier evaluation. SIMNET's utility for supplemental training and hardware development has been documented and the device holds promise as a low-cost alternative for soldier evaluation. Performance by 120 M1 tank crews on a SIMNET test was compared to their performance on a field test measuring four critical domains of Armor combat performance: command and control (C2), communications, position location, and combat driving. Acceptable levels of internal consistency were found for the C2 and communications dimensions. Some evidence of convergent and discriminant validity were found for these two dimensions through the multitrait-multimethod matrix and analysis of variance procedures. However, more score variance was attributable to undesirable sources (method bias and error) than to desirable sources (convergent and discriminant validity). Comparing performance on a set of direct-analogue items from the two tests failed to produce greater evidence of convergent and discriminant validity. Soldiers reported performing tasks on SIMNET to be "mostly the same" as performing tasks on the Ml tank. The rank order of soldier's questionnaire responses on the four performance dimensions reflected the rank order of the four correlations of performance on the two tests. The results of this study do not support the construct validity of SIMNET as a performance testing device of critical combat skills. Future research on SIMNET's construct validity should use a SIMNET test and field test with the exact same items and scenarios surrounding the performance of those items. If SIMNET's construct valid:ty still is not supported, better criterion measures should be sought against which SIMNET can be evaluated.

Disciplines

Industrial and Organizational Psychology | Psychology | Social and Behavioral Sciences

Download

Share

COinS