Power Analysis of Gunnery Performance Measures: Differences between Means of TwoIndependent Groups

摘要

Determination of sample size is a problem that has both practical and statisticalimplications for gunnery performance research. This report examines these implications in the context of statistical tests that compare means from two independent samples of armor crews. Performance variability estimates of gunnery data were derived from Table VIII qualifications at Grafenwohr and from published research on the Unit-Conduct of Fire Trainer (U-COFT). These data were used in examples to describe power analysis procedures developed by Wlkowitz, Ewan, and Cohen (1982) for determining power and sample size and to calculate minimum detectable differences (MDDs) between two samples of crews assuming a two-tailed test of significance with a standard significance criterion of .05 and power set to .80. One of the more notable findings from this analysis was that statistical comparisons of company-sized samples of crews (i.e., N = 14) are relatively insensitive to mean differences in speed and accuracy of performance. The limitations of the proposed method for other tests significance are also discussed. Keywords: Statistical power analysis; Gunnery performance; Table VIII; U-COFT. (KT)