Doctor of Philosophy (PhD)
School of Nutrition and Food Sciences
Ranking tests are important preference and attribute difference tools for sensory evaluation. Replicated testing is used widely to reduce the number of panelists required in other sensory methods such as discrimination. The information regarding replications sensory ranking is limited. This research evaluated important statistical and technical aspects for the development of the foundation for duplicated sensory ranking tests. Three studies were accomplished: 1) A study of nonparametric analyses on real preference ranked data; 2) a sensitivity study of two samples serving protocols for duplicated visual ranking, and 3) protocols comparison in taste. In study 1, 125 panelists ranked in duplicates each of two sets of three orange juice samples. One set contained very different samples and the other similar samples. Five methods of data analysis were evaluated. With similar samples, analyzing duplicates separately yielded inconsistent conclusions across sample sizes. The Mack-Skillings test was more sensitive than the Friedman test and is more appropriate for analyzing duplicated rank data.
Study 2 compared the sensitivity of duplicated yellow color intensity ranking served either in one or two sessions. Panelists (n=75) ranked both similar and different orange juice sets. For each set, rank sum data were obtained from (1) intermediate ranks from jointly re-ranked scores of two separate duplicates for each panelist, (2) joint ranked data of all panelists from the two replications in one serving session, and (3) median rank data of each panelist from two replications. Rank data (3) were analyzed by the Friedman test, while those from 1 and 2 by the M-S test. The similar-samples set had higher variation and inconsistency with one serving session, producing higher P-values than two serving sessions. Both M-S ranking protocols were more sensitive to color differences than Friedman on the medians.
For study 3, an identical design was used to evaluate both serving protocols of duplicated sweetness ranking tests. Separate duplicates were more sensitive for color but not in sweetness, especially with confusable samples. This showed that the conducting duplicated ranking in a single session can be beneficial, but it should be tested for the products and attributes of interest before standardizing testing.
Carabante Ordonez, Kennet Mariano, "Statistical and Technical Methodologies for Duplicated Multiple-Samples Preference and Attribute Intensity Sensory Ranking Test" (2017). LSU Doctoral Dissertations. 4141.