Multiplicative decomposition of the scaling effect in the Mixed Assessor Model into a descriptor-specific and an overall coefficients

- Scaling effect can be decomposed into overall and corrected scaling.
- Overall scaling can represent a psychological scaling effect.
- Corrected scaling can represent a physiological scaling effect.
- Over 60 years old panelists and smokers spread less their scores than other panelists.
- Negativity issue of scaling disappears with overall scaling.

In 1994, Brockhoff and Skovgaard presented the so-called assessor model, including a “scaling coefficient” expressing the predisposition of a panelist to spread more or less his scores than the panel on a given sensory descriptor. This paper discusses potential scaling causes, and then proposes a decomposition of the scaling coefficient into two components: (i) an overall scaling coefficient, independent of the descriptors, expressing a psychological trend of the panelist towards the scoring task in general; (ii) a corrected scaling coefficient for each descriptor, expressing specific sensitivity of the panelist to the descriptor.Applied to 187 sensory datasets, this decomposition showed that panelists over sixty years old and smokers tend to spread less their scores than other panelists regardless of the sensory modalities; further, the corrected scaling coefficients were more heterogeneous among panelists for aroma and flavor compared to texture and appearance sensory modalities.