| Discrimination and preference testing are two extremely important tests in the field of food sensory science. However, according to many current studies, for difference test, the disadvantage is that the balance of efficacy and practicability. Such tests as unspecified triangle test, duo-trio test are easily conducted, but in low power, while other specified 2-AFC test, 3-AFC test are powerful, but difficult to operate. Also, when it comes to preference test, the most commonly used are 9-point hedonic test and paired preference test and the defect is that lots of “false preference” are obtained in the test, resulting in the invalid data. Thus, in order to solve these problems, this research explored the reasons with theoritical analysis and experiments and then accordingly develop new methods for difference test and preference test, resp.Firstly, in order to validate that unspecified tetrad test has higher power than the most commonly used triangle test, in the experiment, tests were performed on different sample systems(standard system and beverage system), mainly using tetrad test and triangle test. The result showed that, reaching the same significance level, the minimum test sample size for tetrad and triangle test was 89 versus 301(standard system) and 25 versus 39(beverage system), respectively. It illustrated that tetrad is much more powerful than triangle test. However, further considering the power of discrimination test, specified tests are more powerful than unspecified tests, to fundamentally increase the practicality and sensitivity of discrimination test at the same time, this researched developed three “self-specified” 3-AFC tests, where subjects define the differences between samples by themseleves. The three self-specification procedure were that, consumers were presented two products, and then according to their own perception, they chose the one they preferred or they thought it was more noticeable(they could remembered better) or they described the overall difference between samples or any difference they noticed by themselves. After these, consumers would use the “attribute of difference” they devleoped to conduct 3-AFC tests based on this difference. Besides the three self-specified 3-AFC tests were compared with “warm-up” 3-AFC test that has low assessbility, but powerful and with triangle test that has low sensitivity. The results showed that the average numbers of correct answers for the three “self-specified” 3-AFC tests are 19.1(“preferred”), 20.2(“impressed”), 20.9(“self-described”), which was significantly different from that of triangle test, 15.1(p=0.001-0.000), but had no significant difference from that of traditional “warm-up” 3-AFC test, 19.8(p>0.05). This indicated that these three first-proposed “self-specified” 3-AFC test sufficiently combined the advantages of both “warm-up” 3-AFC test and triangle test, fundamentally increased the sensitivity and practicality of discrimination testing that applies to subtle differences between samples. Besides, with the aim of examining the practicality and sensitivity of this self-specified test, this paper established self-specified 2-AFC and 3-AFC test methods and compared the sensitivity between them and non-specified tetrad as well as triangle test by combining “preferred” and “self-described” ways of “self-specified”, When reaching the same level of significance, the minimum test sample size for self-specified 2-AFC, self-specified 3-AFC, tetrad and triangle test are 18, 15, 34 and 52, respectively. Thus, it can be seen that the sensitivity of four test methods from high to low is self-specified 3-AFC, self-specified 2-AFC, tetrad and triangle test, which has fully shown that the “self-specified” test methods are more accessible but also powerful.Secondly, via analyzing the reasons of invalid data from preference test, it was found that two main reasons of causing “false preference”:1) different test conditions affected the results, which were ignored commonly and occurred in hedonic test more often; 2)some effects hiding in the test influenced consumers’ preference choices and created “false preference’. Regardint the first problem, Feng et al has developed a new method to solve it, which is Hedonic ranking-rating test. Since it has not been applied in practice, this research confirmed its efficacy with using different products. However, as to the second problem,so far there was no proper methods. Therefore, this research is mainly focusing how to validate the data from paired preference test, for which, a new method,called Riversed Hidden Demand Characteristics preference test(RHDC) was develped. In this test, “screening” procedure was also adopted, in which two identical samples were used in a preference test to screen consumers and those who responded with “no preference”, would be considered not affected by test factors and passing the screen and their data would be kept for following analysis. Yet, different from previous tests, the instruction in RHDC is longer than traditional preference test, the experimenter added : I know you might like them just about the same, because everybody does. The results showed that RHDC test essentially reduces or eliminates the influence of the hidden “test factors” in the test by modifying “experiment instructions” of traditional test, which increased the proportion of “screened” consumers from 20-35% to 80-90% and remarkably improved data effective utilization ratio and result reliability.Nevertheless, under the consideration of the other factor affecting the results of preference test, different conditions influenced the results of preference tests. Thus, the efficacy of RHDC test must be verified under more different test conditions. Because consumers’ subjective preferences were involved in the tests, all types of test factors would affect results greatly; while the conclusion that RHDC test is superior to traditional preference test was drawn by comparing data reported in other papers. Therefore, in this paper, the effectiveness of RHDC test would be verified under conditions that were totally different from previous tests(different test samples, test subjects, test language, test environment and etc.) and each consumer would take both RHDC test and traditional preference test. The results showed that the proportion of passing screening of RHDC test was more than twice that of traditional preference test(88 versus 44%, 68-76.8% versus 22.7-30.5%), which fully illustrated that the changes of test conditions had no effect on the conclusion that RHDC test significantly increases data reliability and effectiveness of preference test. In addition, after this research, the latest paper reported another high efficient method--triadic preference test. Therefore, in order to prove the effectiveness of RHDC test, this paper compared these two methods and found out that passing rates for RHDC test and triadic preference test were 68.0-76.8% and 66.0-70.9%, respectively, showed no significant difference according to McNemar test(p>0.05), which indicates that these two methods are basically the same in the effectiveness of raising passing rate. For specific use, choice can be made according to practical needs.Last, considering that when using either discrimination or preference testing alone, only fixed and single information will be acquired; while combing the two, more useful information will be obtained. For this reason, this paper will discuss “ the attitudes of consumers towards the changes of differences between samples”, combing the new methods mentioned above. It is very meaningful for new product development, product storage, formulation improvement and etc. In the experiment, using original orange juice and orange juice with different concentration of potassium chloride(2, 4, 8, 16, 32, 64 and 128 mM) as test samples, tests were conducted to 50 selected consumers. Firstly, use self-specified 2-AFC test to inspect consumers’ ability to distinguish the difference between original orange juice and orange juice with different concentration of potassium chloride; secondly, use different preference test model to study consumers’ preferences between samples. It can be told from the result that consumers could distinguish between two samples but had no definite preference, when the concentration of potassium chloride was between 2-16 mM; while the concentration raised to 16-32 mM, consumers could not only tell the difference but also “prefer original orange juice” and the tendency got more obvious when the concentration of potassium chloride increased. When concentration raised to 32-64 mM, consumers started to reject samples with potassium chloride; when reached 64 mM or even higher concentration, no consumers would buy the product. In the end, this paper determined that there are three cognitive zones for consumers, when treating the differences between samples, which are ‘indifference zone’, ‘tolerance zone” and ‘rejection zone’. |
PDF Full Text Request