Modelling Relations between Instrumental and Sensory Measurements in Factorial Experiments Modelling Relations between Instrumental and Sensory Measurements in Factorial Experiments

In connection to sensory evaluations of foods, chemical or physical measurements of the relevant sensory properties are often searched for. The data available for testing whether some such instrumental measurement is linearly related to a sensory variable usually stem from a factorial experiment rather than from some random sub-population. Thus, to calibrate the two variables properly and to test whether a linear relationship is compatible with the data, we need a statistical model that reeects the design. The ultrastructural model suggested by Dolby (1976) is such a model. This model further allows for random variation in both of the variables considered. It is shown how testing and estimation may be carried out by use of this model although the methods suggested are not optimal. Finally, it is argued that the use of factorial designs in this connection is sensible, or even necessary, despite the complications it leads to in terms of statistical modelling and analysis.