## Info

"I = instrumental, 0 = objective, 3 = high 2 = medium 1 = low. 4C = consumer. dQC = quality control. dE = expert panel.

"I = instrumental, 0 = objective, 3 = high 2 = medium 1 = low. 4C = consumer. dQC = quality control. dE = expert panel.

inverted U shape (32,33). One can plot the relation between sensory attribute level (abscissa) and liking (ordinate). The data will scatter around the curve but the underlying data should follow a pattern that can be extracted. Fitting a parabola [liking = A + B(Sensory Level) + C(Sensory Level2)] to the data, uncovers this relation. The fitted curve shows the probable nature of the relation, as Figure 3 illustrates (schematically).

Developing Relations Between Variables (Modeling)

Modeling creates a mathematical relation between the systematically varied independent variables and dependent

30 35 40 45 50 55 60 65 70 75 Attribute liking

Figure 2. Relation between attribute liking and overall liking. The steeper the slope of the straight line the more important is the attribute.

30 35 40 45 50 55 60 65 70 75 Attribute liking ceived sweetness) it is sin adequate and useful representation.

We express the relation by the simple linear equation:

Dependent Variable = A + B(Independent Variable)

Depending upon how closely the data points fall to the line, the equation may fit the data well or poorly. Of critical importance is the value of slope, B, which shows how unit changes in the independent variable correspond to changes in the dependent variable. When B is high there is great sensitivity to the independent variable. When B is low there is little sensitivity. Large changes in the independent variable correspond to small changes in the dependent variable.

Figure 2. Relation between attribute liking and overall liking. The steeper the slope of the straight line the more important is the attribute.