How many assessors are necessary for the Optimized Descriptive Profile when associated with training?

Abstract

The number of assessors composing a descriptive analysis panel directly affects the costs for executing the sensory methodology. For the Optimized Descriptive Profile (ODP) method which uses semi-trained assessors, at least sixteen assessors are necessary to make up the panel, where the criteria magnitude of the experimental error estimate is most robust and demanding for determining the ideal number of assessors. Thus, the inclusion of training may result in reduced random variability in the data (experimental error), and therefore require fewer assessors to obtain reliable results. The present study therefore sought to evaluate the influence of training the panel to determine the ideal number of assessors for the ODP evaluation protocol. The study on determination of the number of assessors was conducted via a computer simulation, using the re-sampling with replacement technique from original data obtained by a panel of 26 trained assessors. A total of 10,000 subgroups were simulated for each number of assessors. The criteria for determining the number of assessors were: obtaining an experimental error less than or equal to the effect verified in the reference methodology (Conventional Profile); size of the effect of the interaction between the formulations and assessors less than or equal to the full panel; discrimination of samples similar to that obtained by the complete panel; and minimum loss of information in the sensory map. The discrimination criterion of the samples showed to be a more robust measure for determining the number of assessors required by the ODP technique when associated with training of the panel, with recommendation of at least eight assessors. Training of the panel had a great influence on data quality, providing a 50% reduction in the number of assessors required for the ODP.