| Summary: | The tolerance level of 0 and the confidence level of 0.95 are widely applied in current investigation and sampling strategies for the crime, food, environment, and biodiversity fields. Although some researchers recommend the non-zero tolerance level, few relevant sampling plans have been proposed. In this report, I used the binomial distribution as an example to show the estimate of sampling size, considering different detection efficacies, proportion of concerned units, and confidence level. The results indicate that the required sampling size based on non-zero tolerance is larger than that based on zero tolerance when other parameters are the same. High detection efficacy and high proportion of concerned units can decrease the sampling size. Especially, large sampling size can not only increase the confidence level, but also decrease the proportion of concerned regulated units with effective policy measurement.
|
|---|