Use of the STATIS method to analyze time-intensity profiling data

Abstract

The sensory properties (sourness, pepper, oily, fatty/creamy and garlic) of nine Ranch salad dressing samples varying in fat and garlic flavor according to a 3 2 factorial design were evaluated by time-intensity profiling (TIP). The dynamic and complex processes involved in the perception of these attributes were analyzed with the STATIS method (Structuration des Tableaux A Trois Indices de la Statistique). STATIS allows for the simultaneous analysis of several data matrices, where each matrix consists of the data recorded by TIP at a given instant. STATIS analysis revealed the existence of a common sensory structure during the early and central portions of the time-intensity evaluations (from 6 to 50 s), i.e., the respective positions of the salad dressing samples changed very little during that time period. The compromise matrix obtained for all the instants in the evaluation provided a summary of the sensory properties of the dressing during that 6–50 s period. A different distribution of the samples was observed, however, at the beginning (3 s) and towards the end (75 and 120 s) of the evaluations, with lesser contributions of these instants to the compromise plot. The first two principal components of the compromise space, which accounted for 65 and 11% of the variance, respectively, contrasted the attribute fatty/creamy with the other four attributes. This is the first time the STATIS method has been used to analyze time-intensity curves. The main improvement of the method over other approaches is its ability to determine which portions (start, central portion, tail end) of a TI curve are most critical to the understanding of intensity and time course differences among products.