CONFIDENCE ELLIPSES APPLIED TO THE COMPARISON OF SENSORY PROFILES

Our experiment involved seven panels and six chocolates – five dark chocolates and one milk chocolate. The aim of the study was to compare the sensory profiles of the chocolates. A natural question to ask is “Did the panelists detect any differences among the five dark chocolates or did they systematically contrast them with the milk chocolate?” The scatter plot of the chocolates obtained by principal component analysis was useless to answer that question, because of the proximity of the points. To overcome that, we used confidence ellipses calculated using bootstrap. The originality of the study lies in the fact that we applied those ellipses to hierarchical multiple factor analysis (HMFA): among the seven panels, six were composed of trained professionals and the last one was composed of untrained students, and through that method, we managed to compare the two types of panels and balance the role of each trained panel. HMFA provides in a single scatter plot a representation of the six chocolates for each panel, the trained panels and all the panels. Confidence ellipses around each chocolate show that the combined panels – the six trained panels and also the untrained panel – differentiate the five dark chocolates. They also show how much larger the untrained panel's variability is than that of the trained panels, and how comparable are the trained panels' variability to each other.     

COMPARISON OF SENSORY PROFILES DONE BY TRAINED AND UNTRAINED JURIES: METHODOLOGY AND RESULTS

We compare the sensory profiles of six dark chocolates done by two types of juries: some trained juries and an untrained jury. Six laboratories, each one made up of 10 to 15 judges, are regarded as trained juries since the assessors were well trained before the evaluation of the products. The second type of jury is composed of only one panel of 29 untrained assessors and this jury is named the untrained jury. Fourteen attributes were evaluated and analyses of variance have been carried out by attribute to compare the sensory profiles of the six chocolates done by the trained juries and by the untrained jury. These analyses of variance show that the two types of juries give similar sensory profiles and that the few differences are mainly due to different ways of using the scale.     

CHEESE HARDNESS ASSESSMENT BY EXPERTS AND UNTRAINED JUDGES

Although expert assessment of food characteristics is recognized as a key step in product development, the use of consumer based measurements is sometimes recommended as an equivalent to the experts. From cognitive psychology, support of the role of perceptual learning is found in some instances, although this could not be relevant in others. To address this point performance analysis of experts and untrained panelists in cheese texture evaluation was carried out. Neither the untrained panelists nor the experts were familiar with either the scales or the kind of cheese. The same Cheddar cheese was given to 44 untrained subjects in three trials to assess hardness. The results showed that their judgment has a 29% average random error variance; the interrater reliability being low. The same experiment gave a random error variance of 2% for three highly skilled judges (experts). The difference in variance was linked to training. Untrained panelists also showed an adaptation error. Nevertheless, there was no significant difference between the average ratings of both groups, whether untrained or experts.     

THE RELATIONSHIP BETWEEN CONSUMER PREFERENCES AND TRAINED PANEL RESPONSES

Eight samples of dried tomato soups were profiled by a trained panel, who also provided ratings of overall liking and liking for particular attributes on relative-to-ideal rating scales. A consumer panel tested a subset of four of the soups in home trials, assessing them for overall liking and for particular attributes on either hedonic or relative-to-ideal rating scales. The pattern of overall preferences differed between the trained and untrained panelists, demonstrating the inappropriateness of using trained panelists to provide measures of preference or acceptance. The conclusions regarding consumer preferences would differ depending on which rating scale was used; these differences disappeared when the relative-to-ideal ratings were converted into values of the same form as the hedonic ratings. The overall liking was best predicted by flavor rather than color or thickness. A principal components analysis (PCA) of the profile data compared well with a plot based on the trained panel preference data using MDPREF. The preference data were also fitted to the PCA dimensions using the PREMAP vector model, which gave a good fit for only six of the 15 trained panelists; the PREMAP ideal point model failed to show a better fit. In order to test these models adequately more than eight samples would need to be tested.     

COMPARISON OF PROJECTIVE MAPPING AND SORTING DATA COLLECTION AND MULTIVARIATE METHODOLOGIES FOR IDENTIFICATION OF SIMILARITY-OF-USE OF SNACK BARS1

Three data collection procedures, sorting and two forms of projective mapping (PM), were compared for ease-of-use and the ability to produce meaningful spatial maps when analyzed using Multidimensional Scaling (MDS), Generalized Procrustes Analysis (GPA) or Coordinate Averaging (CA). Eighteen commercially available snack bars were evaluated for similarity-of-use by two panels of 24. MDS of the sorting data and Procrustes analyses of PM data collected on unlabeled axis grouped the bars according to function and provided a meaningful spatial relationship in one dimension. However, MDS analysis of these PM data grouped the bars by similarity-of-use and provided a meaningful spatial interpretation in two dimensions. The CA analysis was not effective in separating the bars by similarity-of-use but did provide an indication of liking. A comparison of spatial configurations using RV coefficients showed moderate correlations between the methods. A panelist survey showed no significant differences in the ease-of-use, task interest or level-of-satisfaction with the final arrangement between the sorting and the PM data collection methods, but panelists did find it easier to change their minds using the PM procedure.     

COMPARISON OF SINGLE PRODUCT SCALING AND RELATIVE-TO-REFERENCE SCALING IN SENSORY EVALUATION OF DAIRY PRODUCTS

Single product scaling and relative-to-reference scaling were compared on the basis of numbers of significant differences among chocolate milks and among vanilla yogurts using both trained and untrained panels. The study involved 920 comparisons among product means with appearance, flavor and mouthfeel attributes. Although there were a greater number of significant t-tests when the relative-to-reference scales were used by both trained and untrained panels, the differences between scale types were minor. Panels showed no advantage with the type of scale they used first. The two methods may be used with equal efficiency for sensory evaluations.     

A COMPARISON OF METHODOLOGY APPLIED TO THE SELECTION OF A PANEL FOR SENSORY EVALUATION OF INSTANT COFFEE

Twenty judges performed a variety of chemosensory tasks in order to select the best scores to form a panel for coffee evaluation. An average of correct responses (P%), one-way analysis of variance (ANOVA) and principal components analysis (PCA) were compared. The tests involved: ability to recognize the four basic tastes, identification and matching of odors, taste intensity evaluation and perception of small differences in taste. P% accounted for 71.17 ± 4.34% and 10 of the judges had scores greater than the final average. ANOVA and PCA resulted in 2 different panels consisting of 9 and 12 judges, respectively. The panel was composed by the nine panelists selected by the three methods. The other three panelists that were doubtful could improve to the point of acceptance with additional training. These methods should be used simultaneously to have more security in the acceptance or rejection of panelists.