Witoon Prinyawiwatkul, kcarabante | 8/7/2017 9:25:44 PM
Kennet Carabante and Witoon Prinyawitwatkul
In the food industry, the study of consumer sensory perception of foods is a key step in new product development. Consumer preference and acceptance evaluation of food products may appear overwhelming because a large number of consumers are often required. However, simplifying some practices may lead to inaccurate results and interpretation when some scientific considerations are avoided. With stiff market competition, measuring preference or difference among food products against the competitors’ products becomes important for product success.
Asking consumers to rank a set of food products based on their preference or intensity of a specific characteristic, such as sweetness, is one of the simplest ways to obtain information for product development or improvement. An example of a ranking test would be recruiting 50 consumers and asking each consumer to rank five ice cream samples, assigning a score of “1” to the least sweet and “5” to the sweetest sample. This testing can be done at a supermarket, a farmers market or in the sensory laboratory without complications. The data analysis to investigate significant differences for a ranking test with multiple samples and consumers is carried out with the classic Friedman test, which is a test to determine if statistical differences exist among the rank scores of three or more food samples. It is easy to compute, and its instructions are widely available. The test requires that each consumer rank all samples independently from other consumers.
When it comes to consumer studies, recruiting consumers is always a challenge and may be costly. Replicating a preference or a difference ranking test by making each consumer perform the test more than once can be done to reduce the number of consumers and cut down on cost. However, the data analysis is often more complicated and has not received formal investigation. Unless the ranked scores from the replications are averaged or used individually, the classic Friedman test cannot be used on replicated ranking. LSU AgCenter sensory scientists at the Sensory Analysis Laboratory have carried out multiple studies to determine the best practice to handle a duplicated sensory ranking test along with its data analysis. The studies compared the performance of correctly detecting differences from the Mack Skillings (M-S) test against other analysis alternatives, including the classic Friedman test, on averaged scores. The M-S test is an extension of the Friedman test that includes an adjustment in its analysis to accommodate replications from the same consumers.
In studies at the AgCenter lab, consumers ranked orange juice sample sets, based on personal preference, in duplicates without knowing they were repeating the exact same test. Both a very different sample set (100 percent, 70 percent and 40 percent orange juice) and a sample set containing similar samples (100 percent, 95 percent and 90 percent orange juice) were evaluated to compare performance of the analysis depending on degree of difficulty of the test. The number of consumers used in a single test was also varied from 10 to 125 to evaluate the impact of increased sample size on the analysis.
With very different samples, all statistical analyses found significant differences in all consumer sample sizes; however, with similar samples the M-S test was the most sensitive and consistent analysis to detect significant differences. The M-S test could consistently detect differences in preference using only 20 panelists, whereas the classic Friedman test required at least 50 consumers. Although the reduction of the number of consumers needed could vary depending on the product or degree of difference among samples, the M-S test is capable of rewarding agreement between the responses from the same consumers, thus saving the cost of consumer testing. Researchers at the AgCenter laboratory have used ranking tests to screen various products, including the development of a low-sodium Cheddar cheese with up to 70 percent less sodium and preference comparisons among steaks from grass-fed steers.
Kennet Carabante is a doctoral student under the supervision of Witoon Prinyawiwatkul, Horace J. Davis Endowed Professor in Food Science and Technology, in the School of Nutrition and Food Sciences.
Kennet Carabante, left, and Amber Jack, both graduate students in food sciences, participate in a food demonstration. Photo by Tobie Blanchard