Resampling analysis of participant variance to improve the efficiency of sensor modeling perception experiments

Abstract

Night Vision and Electronic Sensors Directorate (NVESD) Modeling and Simulation Division (MSD) sensor models, such as NV Therm IP, are developed through perception experiments that investigate phenomena associated with sensor performance (e.g. sampling, noise, sensitivity). A standardized laboratory perception testing method developed in the mid-1990's has been responsible for advances in sensor modeling that are supported by field sensor performance experiments.1 The number of participants required to yield dependable results for these experiments could not be estimated because the variance in performance due to participant differences was not known. NVESD and George Mason University (GMU) scientists measured the contribution of participant variance within the overall experimental variance for 22 individuals each exposed to 1008 stimuli. Results of the analysis indicate that the total participant contribution to overall experimental variance was between 1% and 2%.