So close, yet so far - a methodological investigation of the potential of and optimal sample sizes for the application of napping as rapid sensory method

Rapid methods, like napping, with consumers are used more and more frequently. The aim of napping is a sensory comparison of several products with regard to their relative similarity to one another.
Napping studies generally involve 9-15 people, but it is largely unclear whether this sample size is sufficient for untrained test subjects to provide statistically reliable and reproducible results.
A napping with 15 strawberry yogurt samples and n = 104 untrained participants for the characteristic properties appearance, taste and texture was arranged, whereby the formed groups were provided with sensory attributes by the test persons.
For the evaluation of the napping data, random samples were drawn from the entire sample in order to determine the optimal sample size. The samples taken comprised 10, 15, 20 and 33 cohorts and were evaluated using multiple factor analysis (MFA) and generalized procrust analysis (GPA) in the Microsoft Excel add-on XLStat®.
Relative matches of the determined napping data between the random samples and the entire sample were calculated, expressed by the tucker´s congruence coefficient (rc). From this it can be deduced, that sample sizes of n ≤ 15 do not yet achieve statistically reliable results and a sample size from n = 20 already show a mediocre agreement, but all consensus indices for sample sizes of n = 33 are equal to or higher than 97 %, indicating exceptional high consensus of the separate sub-samples with the whole sample. According to literature, congruence coefficient values above 95 % can be considered as virtual identical factors. So finally from a sample size of n = 33, statistically reliable and reproducible results are achieved.
Based on the available results, a sample size of at least 20, ideally 33 untrained subjects for napping studies are recommended in order to obtain statistically reliable, reproducible results.