Use of multi-stratum design and analysis in a sensory trial

Experimental designs are widely used in sensory and consumer research. Incomplete block designs and balanced crossover designs are common. Split-plot designs with multiple strata can also be used. This paper discusses the design and analysis of a trained panel experiment with multiple strata used to assess the "overall flavour" of five sweeteners each at six concentration levels. On each of five tasting days, 10 assessors evaluated 18 products, constituting six levels of three of the five sweetener series. A split-plot experimental design arrangement was used to assign series to the mainplots and the levels within series were assigned to the subplots. This design gave greater precision for the comparison of different levels within each sweetener series. The analysis investigated the existence of treatment carryover effects and correlations between repeated evaluations by assessors. In doing so, it was possible to adjust for the effect of day, order of presentation and assessor in the estimation of sweetener series and levels within series.