Purpose: The goal of this study was to use mathematical modeling to determine the impact of including a portion of the samples based on the sampler’s knowledge of risk factors. The performance characteristics of sampling plans that include such “samples of opportunity” (SOO) were compared to that of traditional preharvest sampling plans.
Methods: Computer simulations were performed to compare the relative effectiveness of random, stratified-random, and z-pattern vs. SOO sampling. The SOO sampling reserved two thirds of samples to be taken from identified high-risk areas within a field. These evaluations assumed the contamination in the field was nonrandom, with three contamination scenarios being evaluated: animal house nearby, power line above the field, and field partially exposed to floodwaters. The simulation modeling tool allowed a large number of field contamination scenarios to be generated and evaluated systemically.
Results: The detection probability for a nonrandomly contaminated preharvest field (five by six plots with nine subplots per plot (total of 270 subplots)) using random, stratified-random, and z-pattern sampling plans was 0.30±0.11, 0.32±0.11, 0.32±0.17, respectively. The SOO sampling plan had a detection probability of 0.61±0.25. The detection probability of SOO was 96% higher than other sampling plans (P<0.001). However, if the assumption of contamination source is incorrect, detection probability of SOO drops to 0.33±0.23, which is not significantly different than the other sampling plans.
Significance: This study provided a mathematical approach for evaluating the effectiveness four preharvest sampling plans, and suggested that having the knowledge of the contamination source in the field would improve effectiveness of sampling.