Evaluation of the Labeled Hedonic Scale under different experimental conditions

Hedonic scaling: A review of methods and theory

“…Third, because of both its small number of available categories and the general tendency of subjects to avoid using extreme categories (Hollingworth, 1910;Moskowitz, 1982;O'Mahony, 1982), the scale is highly vulnerable to ceiling effects (Schutz & Cardello, 2001;Stevens & Galanter, 1957), one of the context effects that was described above (Section 2.3.4). The avoidance of the end categories effectively reduces the 9-point scale to a 7-point scale (Moskowitz, 1982;Moskowitz & Sidel, 1971) and limits its ability to discriminate among very well liked or very disliked stimuli (Lim & Fujimaru, 2010;Schutz & Cardello, 2001;Villanueva & Da Silva, 2009). Lastly, from a statistical standpoint, because the data it yields are categorical and discrete without a true zero point, the type of statistical analyses that can be applied with confidence is limited, i.e., nonparametric statistics.…”

“…In addition, as recognized in one of the original publications of the scale (Peryam & Pilgrim, 1957), some of the assumptions for parametric analyses (e.g., normality, homogeneity (Peryam & Girardot, 1952); and (b) a sample ballot for a common consumer test used in a laboratory setting. of variance) are often violated (Gay & Mead, 1992;Giovanni & Pangborn, 1983;O'Mahony, 1982;Villanueva, Petenate, & Da Silva, 2000), particularly the data for extremely liked or disliked stimuli (Lim & Fujimaru, 2010;Lim et al, 2009). Accordingly, a large sample size, commonly over 75 responses per stimulus, is necessary to approximate normality in order to make valid statistical inferences.…”

“…Fourth, the hedonic gLMS and the LHS, because they are bounded by all-inclusive end anchors (e.g., 'Most liked or disliked sensation imaginable'), enable comparison of individual and group differences within the context of the full range of perceptual experiences (Bartoshuk et al, 2002). Fifth, unlike the intensity ratings obtained by magnitude estimation and the gLMS that are typically distributed log-normally across subjects (Green et al, 1993), it has been shown that the hedonic ratings obtained by category-ratio scaling are normally distributed (Lim & Fujimaru, 2010;Lim et al, 2009). Therefore, parametric analyses (e.g., ANOVA) can be readily applied without any data transformation, i.e., there is no need to log-transform the data obtained from the LHS.…”

Methodology for constructing perceptual maps incorporating measuring error in sensory acceptance tests

“…Individual responses are certainly featured in a descriptive study summarized in numerical data. (Lim & Fujimaru, 2010). As to interference of the contextual effect in stimulus-response model, consider a situation where sensory perception comes from a trained panel with the ability to detect small differences between samples.…”

“…This suggests that parametric statistics such as analysis of variance may return incoherent results (Peryam & Pilgrim, 1957), because the assumptions are generally violated. See Gay & Mead (1992), Giovanni & Pangborn (1983), Lim, Wood & Green (2009), Lim & Fujimaru (2010, O'Mahony (1982), and Villanueva, Petenate and Silva (2000).…”