Metrological properties of Flash Profile

Although FP data may be relatively unusual to the user of conventional descriptive profile, a series of metrological properties (accuracy, repeatability, sensitivity, robustness and level of consensus) may provide a good indication of the quality of the results. These properties, how to estimate them and their pertinence in the case of FP are discussed in this section. 

The accuracy or validity of FP results can first be estimated by comparison with a reference technique, i.e. a conventional profile such as QDA. Note that when conducting an FP with consumers from a market research perspective, this comparison criterion may not hold and following discussion on validity takes on a different meaning. 

In addition to this, one may want to assess the quality of FP measurements in terms of panel performance. In this regard, it should first be mentioned that panel performance cannot be understood as in conventional consensus-based methods, since panellists do not use the same attributes, and even if they do use the same words these words may not be assumed to convey the exact same sensory notion. Also, one should bear in mind that FP results are dependent on subjects descriptive skills, levels and fields of expertise. As already mentioned, should one wish to get the most comprehensive description of a product set, it is advisable to recruit experienced subjects with complementary fields of expertise. 

As a result, most panel monitoring techniques that have given rise to a significant body of literature are not really applicable here. Nevertheless, a number of criteria can be reviewed and discussed briefly. 

Repeatability in FP could be assessed by means of a full repetition of the evaluation task, or by the use of duplicates. A full repetition would imply repeated evaluation 

---