Hedonic scale with reference: performance in obtaining predictive models Selma Bergara-Almeida, Maria Aparecida A.P. da Silva* Department of Alimentary Planning and Nutrition — DEPAN, Food Engineering Faculty, UNICAMP, caixa postal 6121, Campinas (SP), Brazil Received 28 January 2000; received in revised form 19 May 2001; accepted 20 July 2001 Abstract Although the hedonic scale can be used to optimise product acceptability, an alternative procedure allowing for direct compar- ison between the target product and the prototypes may be advantageous. The use of the magnitude estimation scale for hedonic responses has been exhaustively studied, but the use of the hedonic scale, adjusted to allow for the comparison between various samples and a reference, has not been reported. The objective of this research was to evaluate the performance of a hedonic scale with a reference to generate predictive models for product acceptability. Eleven formulations of reconstituted orange juice were prepared using a 2 2 central composite design. The independent variables studied were  Brix and acidity. Overall acceptability was evaluated using both the 9-point hedonic scale and a hedonic scale with reference. The data were analysed using Response Surface Methodology, residual analysis and Dunnett’s test. The models generated by the two scales were similar with respect to: the adjusted R 2 (P < 0.001), the non-significance of the lack of fit (P40.05), the appropriateness of residuals to the statistical assump- tions inherent in the regression analysis and, the optimised regions. However, it should be noted that the optimised region obtained by utilizing the hedonic scale with reference refer to samples whose acceptability are likely to be greater than that of the reference sample, as by using this scale consumers are forced to evaluate each sample in relation to the reference. For researchers working in product development this type of information can be a valuable tool. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Sensory evaluation; Consumer test; Hedonic scale; Reference scale; Response Surface Methodology 1. Introduction Frequently, food industries need to optimise a market product acceptability. In such situations, the industries usually apply affective tests, principally acceptance tests (Meilgaard, Civille, & Carr, 1988). When hedonic scales are used with this objective, the preference of one pro- totype in relation to the others samples is inferred from the acceptability means: products presenting means sig- nificantly greater than the others (P40.05) are con- sidered to be preferred. However, in these cases, a direct comparison with the objective of identifying the con- sumer preference/choice is not, in fact, carried out. On the other hand, preference tests, such as the paired or ranking test, which really do direct comparison among samples, are less used in consumer tests as they do not provide the degree of product acceptability, only indi- cating the choice of the consumer. Thus, a scale which allows the direct comparison between one or more samples and which also allows for the evaluation of the degree of product acceptability could be of great advantage in consumer tests. In addition, a scale which allows the consumers to express their hedonic responses in terms of positive or negative deviation from a refer- ence, can increase the accuracy of the individual responses, contributing to an increase in the dis- criminative power of the test. The magnitude estimation scale for hedonic responses presents the above cited characteristics, as it allows for the evaluation of one or more samples in relation to a reference, for example, a market leader. Vickers (1983) and Moskowitz and Sidel (1971) showed similar perfor- mance of this scale to that of the category scale in con- sumer tests, and McDaniel and Sawyer (1981) concluded that the magnitude estimation is more sensi- tive than the 9-point hedonic scale as a discriminative tool in consumer tests. On the other hand, this scale was considered inadequate for hedonic responses by 0950-3293/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0950-3293(01)00057-X Food Quality and Preference 13 (2002) 57–64 www.elsevier.com/locate/foodqual * Corresponding author. Tel.: +55-19-3788-4074; fax: +55-19- 3788-4060. E-mail address: mariasi@fea.unicamp.br (M.A.A.P. da Silva).