Investigation of Test Performance Over Repeated Sessions Using Signal Detection Theory: Comparison of Three Nonattribute‐specified Difference Tests 2‐afcr, A‐not a and 2‐afc

Abstract

To investigate more flexible methods for measuring overall sensory differences, the performance of three nonattribute‐specified difference test methods was compared using Signal Detection Theory. A‐Not A, 2‐AFC, and 2‐AFCR tests were performed with experienced subjects over repeated sessions. Learning effects were investigated to determine how much training would be needed before subjects could perform these tests consistently. Results in d′ showed that in early sessions, the 2‐AFC and 2‐AFCR performed better (higher d′) than A‐Not A and that after two sessions, the test performance stabilized on that level. Learning effects for the A‐Not A test were poorer in the earlier sessions, with performance levels increasing, even after six sessions. In business situations where the tested products frequently vary, the use of 2‐AFC and 2‐AFCR would be recommended because they are more sensitive in the earlier sessions, requiring less training time than A‐Not A. PRACTICAL APPLICATIONS Sensory difference tests are important tools for business decision making. Food companies may save a lot of money by employing more effective methodologies. For sensory difference tests to be used accurately and efficiently, test performances need to be investigated with consideration for food discrimination constraints and objectives. This article explored the performance of nonattribute‐specified 2‐AFC and 2‐AFCR tests over several sessions in comparison to the results of a standard signal detection measure, A‐Not A, to use the differences between these methods as a basis for inference concerning the cognitive strategies and relative test effectiveness of nonattribute‐specified 2‐AFC and 2‐AFCR for food discrimination, in practical situations when both the reference product and test product(s) change frequently. This article suggested the applicability of a nonattribute‐specified 2‐AFCR procedure for discriminating small differences without much prior training.

Sensory discrimination by consumers of multiple stimuli from a reference: Stimulus configuration in A-Not AR and constant-ref. duo-trio superior to triangle and unspecified tetrad?

“…Yet, in recent publications, the relative efficiencies of different overall discrimination test methods have been actively discussed using only a fixed design where the stimuli to be http://dx.doi.org/10.1016/j.foodqual.2015.06.021 0950-3293/Ó 2015 Elsevier Ltd. All rights reserved. discriminated from each other include only a pair of stimuli (2 stimulus types) (Choi, Kim, Christensen, van Hout, & Lee, 2014;Ennis & Christensen, 2014a, 2014bKim, Chae, van Hout, & Lee, 2014;McClure & Lawless, 2010;Olivas, Lopez-Malo, Angulo, & O'Mahony, 2014;Stocks, van Hout, & Hautus, 2013, 2014van Hout, Hautus, & Lee, 2011). Also, the relative test power of the various discrimination test methods have been only investigated based on the same numbers of tests (Bi, 2011;Ennis, 1993;Ennis & Jesionka, 2011;Jesionka, Rousseau, & Ennis, 2014;Kim et al, 2014;Schlich, 1993).…”

“…This test power or precision of a measure has been theorized based on the precondition of perfect memory and no reduction in the clarity of the sensory information during a subject's sensory and cognitive processes involved in performing the discrimination test (Kim, Kim, Cho, & Lee, 2015). Different from theoretical test power based on predicted variance for a given size of d 0 , operational test power refers to operational capability; that is, the test not being unfavorably interfered with by physiological or cognitive perceptual processes, leading to measure correctly the true sensory differences between stimuli (Kim & Lee, 2012;van Hout et al, 2011). In SDT and Thurstonian modeling, when studying a certain size of sensory difference, a more operationally powerful discrimination test method will result in a larger d 0 with a higher lower confidence limit.…”

“…When the number of tests increases for the reminder methods, their relative theoretical power deficiency can be compromised. Namely, in a sensory discrimination test, as mentioned earlier, the number of test stimuli required often determines the operational test power and applied efficiency (van Hout et al, 2011), although the test power has typically been demonstrated in terms of the number of tests required to detect the given sensory difference (Bi, 2011;Ennis, 1993;Ennis & Jesionka, 2011;Jesionka et al, 2014;Kim et al, 2014;Schlich, 1993). Fig.…”

Investigation of operationally more powerful duo-trio test protocols: Effects of different reference schemes

“…doi:10.1016/j.foodqual.2012.02.016 that between the test protocols that have the same statistical power, one protocol can be more resistible to the above mentioned bias sources and thus operationally more powerful than the other, leading to higher test performance. This has also been referred to as the applied power (van Hout, Hautus, & Lee, 2011) of a sensory difference test. Regarding on the duo-trio test, assuming that the various versions of the protocols have the equivalent statistical power, the question becomes which version of the protocol would have better operational power.…”

The Impact of Wine Style and Sugar Addition in liqueur d’expedition (dosage) Solutions on Traditional Method Sparkling Wine Composition

“…The A-Not A difference test can be superior to the triangle test, duo-trio test, and same-difference test in terms of statistical power [17]. It is used to measure the overall sensory difference of one or more products from a reference sample A (OC-control) using a sureness rating [17,18]. Participants were asked to state on ballot papers whether the sample poured was A or, not A.…”