Validation of boar taint detection by sensory quality control: Relationship between sample size and uncertainty of performance indicators

To prevent impaired consumer acceptance due to insensitive sensory quality control, it is of primary importance to periodically validate the performance of the assessors. This communication showcases how the uncertainty of sensitivity and specificity estimates is influenced by the total number of assessed samples and the prevalence of positive (here: boar tainted) samples. Furthermore, a statistically sound approach to determining the sample size that is necessary for performance validation is provided. Results show that a small sample size is associated with large uncertainty, i.e., confidence intervals and thus compromising the point estimates for assessor sensitivity. In turn, to reliably identify sensitive assessors with sufficient test power, a large sample size is needed given a certain level of confidence. Easy-to-use tables for sample size estimations are provided. (C) 2014 Elsevier Ltd. All rights reserved.