Number of trials necessary to achieve a representative performance of accuracy and timing during combat shooting

Abstract

Acquisition evaluations are expensive, have a high time liability, and tend to prioritize engineering requirements over human factors and good experimental design. Shooting serials usually consist of static prone shooting to minimize movement variability, increase reliability of accuracy and timing data or use a single data point to make acquisition decisions. To better understand the number of trials required to achieve representative performance of accuracy and timing, 60 shots from the standing unsupported position while cyclically moving the weapon from the low ready to shoot was utilized. Intra‐class correlations, standard error of measurement, minimal detectable change, and sequential averaging analysis (SAA) were used to evaluate the variables of radial error, shot interval, x‐bias and y‐bias over the 60 shots. The number of trials required to achieve an intraclass correlation of greater than 0.8 ranged from 2 (shot interval) to 58 (y‐bias), whereas SAA varied between 3 (x‐bias) and 43 (shot interval) trials. When averaging 10 shots at a time, the moving intraclass correlation remained above 0.8 for radial error and y‐bias between 7 and 15 shots, shot interval from the second shot, but x‐bias never reached 0.8. The number of trials required to satisfy each reliability method was inconsistent, in line with previous literature. Given the limitations identified in the literature as well as practical considerations such as the preference for prioritizing radial error reasonable performance stability can be achieved after 15 shots, and using the moving intraclass correlation results it is recommended that the first six shots are discarded with the following nine shots used for analysis.