Professional Documents
Culture Documents
Scale Basic operation Number usage Permissible statistics Example hedonic scales
The numerical values for the semantic labels can be found in the original papers for each scaling method (Lim et al., 2009 [LHS], Schultz and Cardello, 2001 [LAM], Guest et al. (2007)
[OPUS], and Green et al. (1993) [gLMS]).
Source: Adapted from Lim, J. (2011). Hedonic scaling: a review of methods and theory. Food Quality and Preference 22, 733–747.
Scoring Methods This scale can also be displayed vertically and be adapted to
a seven-point scale or present ‘faces’ representing the degree of
Scoring methods are those in which one can observe the accep- like/dislike, to facilitate its use with children or people who
tance differences and in some cases the size of such differences. cannot read. This type of scale is often used for comparison
Furthermore, in many cases, it is possible to evaluate differ- among samples in Food Science, considering that if consumers
ences in acceptance among groups of consumers. ‘liked extremely’ sample ‘A’ and they ‘liked slightly’ sample ‘B,’
then sample ‘A’ is more accepted and preferred than sample ‘B.’
Thus, this scale infers the preference.
Category Scales When using this scale, products must be presented in a
Category scale is a measurement method in which the assessor monadic way, that is, one at a time, with no new assessment
is asked to ‘measure’ the intensity of a stimulus assigning a of the same sample or reference to the scores given to previous
value (category) on a limited scale, which is usually numerical. samples. The idea is that in purchasing situations, consumers
Category scales are widely used in sensory analysis for affective do not test several products and then select the preferred one;
responses (related to the preference or acceptance of products). they evaluate the products, one at a time, and select one based
These are scales where the degree of acceptance is evaluated on their comparative memory of previous evaluations of prod-
through verbal (nine-point hedonic scale) or numerical labels ucts, that is, comparing the products to their internal and
(numerical nine-point hedonic scale). individual reference. The aim when evaluating samples mon-
adically is to promote absolute evaluations that are not affected
by comparisons with other products, thus removing contextual
Nine-point hedonic scale
effects. Ideally, the samples should be evaluated on different
Nine-point hedonic scale is widely used for acceptance testing
days or weeks, but for practical conditions, products are
when the goal is to determine how much a specific product is
evaluated one after the other and the evaluation sheet with
liked by the consumer. It consists of a balanced bipolar scale
the scores and samples are removed after each assessment. The
around the neutral category (5 ¼ neither liked nor disliked),
order of samples presentation must be balanced to avoid posi-
with four positive categories (6, 7, 8, and 9 ¼ liked slightly,
tioning errors or carryover effects, and for this, designs are
liked moderately, liked very much, and liked extremely, respec-
already available to facilitate the implementation of tests.
tively) and four negative categories (1, 2, 3, and 4 ¼ disliked
The main reason the nine-point hedonic scale is the most
extremely, disliked very much, disliked moderately, and
used in acceptance testing is that, compared to other scales
disliked slightly, respectively) (Figure 2).
(e.g., magnitude estimation), it has a limited number of cate-
The categories are composed of sentences representing sev-
gories, making it easy to use by different segments of the
eral degrees of affection and are arranged successively, suggest-
population and requiring no prior training. For researchers,
ing continuity of liking and disliking of products. The
the evaluation of the generated data is easier when compared
sentences help consumers to respond adequately and
to other techniques, which require measurement of lines or
researchers to interpret the responses regarding the acceptance
estimation of ratios that may include fractions. In addition, it
of products.
was demonstrated that scales with simple categories are as
sensitive as the other scales (magnitude estimation) in relation
Table 2 Example of data of six judges evaluating four samples using to the power of sample discrimination. So, if the main goal of a
best–worst scale
study is to measure differences in acceptance among samples
B–W B–W B–W B–W or predict the acceptability, the nine-point hedonic scale has
Participant (sample A) (sample B) (sample C) (sample D) proved to be simple and effective. Many companies have stored
data based on this kind of scale, which facilitates the compar-
1 3 1 3 1 ison of products under development to those already available
2 0 0 3 3 on the market.
3 3 1 2 2 However, although it is widely used in Sensory Science,
4 1 3 3 1
several limitations are attributed to the nine-point hedonic
5 3 3 1 1
scale. First, by having unequally spaced categories, the result
6 3 2 1 2
Aggregate 7 0 5 2 indicates which samples are more accepted but does not quan-
tify how much a sample is more accepted than another sample
Source: Jaeger, S. R., Jorgensen, A. S., Aaslyng, M. D. and Bredie, W. L. P. (2008). (because if a product obtained score 8, it does not mean that
Best–worst scale: an introduction and initial comparison with monadic rating for it is twice more accepted than a product obtaining score 4),
preference elicitation with food products. Food Quality and Preference 19, 579–588. nor compares the acceptance among individuals or groups.
DISLIKE DISLIKE DISLIKE DISLIKE NEITHER LIKE LIKE LIKE LIKE LIKE
EXTREMELY VERY MUCH MODERATELY SLIGHTLY NOR DISLIKE SLIGHTLY MODERATELY VERY MUCH EXTREMELY
1 2 3 4 5 6 7 8 9
Figure 2 Nine-point hedonic scale with the numbers used in the statistical assessments. Reproduced from Wichchukit, S. and O’Mahony, M.
(2014). The 9-point hedonic scale ranking in food science: some reappraisals and alternatives. Journal of the Science of Food and Agriculture. http://dx.
doi.org/10.1002/jsfa.6993.
Sensory Evaluation: Sensory Rating and Scoring Methods 747
1 2 3 4 5 6 7 8 9
judges, and therefore, comparisons among the values assigned adaptation (repeating the answers when there are a number of
by individuals have no meaning; (2) the absence of a label products), encouraging the judge to conduct the evaluation in a
(e.g., ‘disliked very much’) makes difficult the translation of more conscious and discerning manner; (3) it reduces or elim-
numbers into perceptions of acceptance; (3) it does not pro- inates the numerical and contextual effects, allowing its use in
vide ordinal data on the acceptance or rejection of products; cross-cultural studies; and (4) it generates continuous data,
(4) the numerical nature of the method, using numbers to allowing a safer use of parametric or nonparametric analyses.
estimate the magnitude of the differences in acceptance, can
be a difficult task for the general population, so the quality of
the data obtained often depends on the judges’ experience or
Category-Ratio Scales
training level; (5) there is a tendency among judges to use
rounded values, such as 5, 10, 20, and 100, casting doubt on Discussions focusing on the validity and superiority of category
the validity of the method; and (6) data analysis is complex, or magnitude scales fostered the proposition of a new type of
requiring normalization and standardization even before the scale containing the strengths of both. The category-ratio scale
statistical analysis begins. Therefore, the magnitude estimation is linear and has verbal descriptors positioned according to the
scale requires very detailed instructions and practice with the semantic magnitude, with the positions determined empiri-
method, which cannot be achieved in experiments involving cally by magnitude estimation.
consumers. For this reason, it is not widely adopted in applied The first category-ratio scale was developed for the evalua-
sensory research. tion of the intensity of oral sensations, which was called oral
labeled magnitude scale (LMS). This scale had the term
‘without sensation’ at the lower end and ‘strongest imaginable
Hybrid Hedonic Scale
oral sensation of any kind’ at the upper end, with five addi-
It is a linear scale (0–10) resulting from the combination of tional descriptors spaced in a quasi-logarithmic manner. After
structured and unstructured scales (Figure 5). The hybrid hedonic a few years, there were discussions about the top of the scale, as
scale is anchored with verbal labels on middle regions it was considered that oral sensations were different among
(5 ¼ neither liked nor disliked) and extremes of scale individuals due to differences in the tongue anatomy and that
(0 ¼ disliked extremely and 10 ¼ liked extremely), which makes the correct action should be anchoring the top end more
it easier to be used by the consumer than the unstructured scale. generally. Thus, the upper anchor was replaced by ‘strongest
The other portions are marked with equidistant points in order to imaginable sensation of any kind,’ and this scale was called
better define the degree and direction of hedonic continuity. general version (gLMS) (Figure 6).
Advantages attributed to this scale are the following: (1) One of the first scales developed for purposes of product
Since it is not restricted to a limited number of categories, it acceptance was LAM, in which the labels were derived from the
provides better discrimination power; (2) it reduces errors by nine-point hedonic scale, with two additional anchors:
0 5 10
disliked neither liked liked
extremely nor disliked extremely
Figure 5 Hybrid hedonic scale. Reproduced from Villanueva, N. D. M., Petenate, A. J. and Silva, M. A. A. P. (2005). Performance of the hybrid hedonic
scale as compared to the traditional hedonic, self-adjusting and ranking scales. Food Quality and Preference 16, 691–703.
Extremely
Very strong
Like
Very much
Strong
Moderately Moderate
Slightly Weak
Neutral
Slightly Weak
Moderately Moderate
Strong
Dislike
Very much
Very strong
Extremely
‘greatest imaginable like’ and ‘greatest imaginable dislike.’ The From demands of users of sensory analysis, it is expected
development of the scale followed the procedures used in the that researches are conducted for the development of new
creation of LMS. methodologies with scales to evaluate the acceptance of prod-
After that, other scales have been developed ((hedonic) ucts by consumers, in order to quantify the magnitude of
gLMS, OPUS, etc.), with advantages and disadvantages, such differences among samples and segment consumers according
as incorrect quantification of the magnitude of acceptance, to the preference. One has to keep in mind, however, that not
inappropriate descriptors, and inappropriate spacing among always what is new is the best. On the other hand, it is not
descriptors. because a methodology has been used for years that it has no
A few years ago, a group of researchers developed the LHS, disadvantages.
using a procedure similar to that used in the development of
LMS and comparing the magnitude estimation and nine-point
hedonic scales. The LHS generated data identical to those
obtained by the magnitude estimation, validating the position- Further Reading
ing of descriptors and assuming that the magnitude of accep-
Almli VL, Ovrum A, Hersleth M, Almoy T, and Naes T (2015) Investigating individual
tance can be quantified. In addition, it had the power of sample preferences in rating and ranking conjoint experiments. A case study on semi-hard
discrimination slightly higher than that of hedonic scale and cheese. Food Quality and Preference 39: 28–39.
greater resistance to the effects of nonuse of the scale extremes, Cardello AV and Schutz HG (2006) Sensory science: measuring consumer acceptance.
resulting in data with a more normal distribution. In: Hui YH (ed.) Handbook of, food science and technology, vol. 2, pp. 1–8. CRC
Taylor & Francis. Chapter 56.
The advantages of category-ratio scales, especially LHS, are
Feng Y-H, Gutiérrez-Salomón AL, Angulo O, O’mahony M, and Wichchukiti S (2014)
as follows: (1) They generate magnitude data, making them Data from ‘words only’ and ‘numbers only’ 9-point hedonic scales are not
possible, for example, to evaluate the relationship between the interchangeable for serial monadic as well as rank-rating protocols: aspects of
acceptance and the amount of a given ingredient and how memory and culture. Food Quality and Preference 41: 12–19. http://dx.doi.org/
much a sample is more accepted than another one (product 10.1016/j.foodqual.2014.11.004.
Green BG, Schaffer GS, and Gilmore MM (1993) Derivation and evaluation of a
A is three times more accepted than product B); (2) it is semantic scale of oral sensation magnitude with apparent ratio properties. Chemical
possible to translate the numerical data into perceptions of Senses 18: 683–702.
acceptance, because the positions of the category labels were Guest S, Essick G, Patel A, Prajapati R, and McGlone F (2007) Labeled magnitude
determined; (3) since they are continuous scales, the judges are scales for oral sensations of sweetness, dryness, pleasantness and unpleasantness.
Food Quality and Preference 18: 342–352.
not confined to a limited number of categories; (4) by present-
Gutiérrez-Salomón AL, Gámbaro A, and Ângulo O (2014) Influence of sample
ing more spaced anchors (most liked/disliked sensations imag- presentation protocol on the results of consumer tests. Journal of Sensory Studies
inable), scale sensitivity is increased when discriminating 29: 219–232.
samples, especially those that are very or little accepted; (5) Jaeger SR, Jorgensen AS, Aaslyng MD, and Bredie WLP (2008) Best-worst scale: an
data on the acceptance are normally distributed, so parametric introduction and initial comparison with monadic rating for preference elicitation
with food products. Food Quality and Preference 19: 579–588.
analyses can be applied requiring no preliminary processing of Land DG and Sheperd R (1988) Scaling and ranking methods. In: Piggott JR (ed.)
data; and (6) the scales as well as the nine-point hedonic scale Sensory analysis of foods. London: Elsevier.
are easy to use, although they require instructions and practice Lawless HT, Popper R, and Kroll BJ (2010) A comparison of the labeled magnitude
so that the data obtained are reliable. (LAM) scale, an 11-point category scale and the traditional 9-point hedonic scale.
Food Quality and Preference 21: 4–12.
The disadvantages of this type of scale would be as follows:
Lim J (2011) Hedonic scaling: a review of methods and theory. Food Quality and
(1) As the anchors are far away from the other categories, there Preference 22: 733–747.
could be a compression of ratings in the center of the scale, Lim J, Wood A, and Gren BG (2009) Derivation and evaluation of a labeled hedonic
reducing its power of discrimination, and (2) it can be misused scale. Chemical Senses 34: 739–751.
by some groups of judges such as those with lower education McEwan, J. A. and Lyon, D. H. (2003). Sensory evaluation | Sensory rating and scoring
methods. In: Caballero, B. (ed.) Encyclopedia of food sciences and nutrition, pp.
or who are already used to other scales. For example, it was 5148–5152. Elsevier.
observed that many judges performed evaluations of samples Meilgaard MC, Carr BT, and Civille GV (2006) Sensory evaluation techniques, 4th ed.
only at the positions where the labels were located, making it a Boca Raton, FL: CRC Press.
category scale. Therefore, instructions on the use of the scale Moskowitz HR, Dravnieks AL, and Klarman L (1976) Odor intensity and pleasantness:
evidence for different psychological laws. Science 184: 583–585.
are of vital importance for obtaining reliable data.
Nicolas L, Marquilly C, and O´Mahony M (2010) The 9-point hedonic scale:
are words and numbers compatible? Food Quality and Preference
21: 1008–1015.
Peryam DR and Pilgrimm FJ (1957) Hedonic scale method of measuring food
Conclusions preferences. Food Technology 11: 9–14.
Schiffersteinm HNJ (2012) Labeled magnitude scales: a critical review. Food Quality
In this article, the four basic types of scales used in sensory and Preference 26: 151–158.
Schutz HG and Cardello AV (2001) A labeled affective magnitude (LAM) scale for
evaluation of food were presented: nominal, ordinal, interval,
assessing food liking/disliking. Journal of Sensory Studies 16: 117–159.
and ratio. The main properties of the methods with scale were Stevens SS (1946) On the theory of scales of measurement. Science
listed, emphasizing the use in acceptance testing. By describing 103: 677–680.
the properties, advantages, and limitations of the methods Villanueva NDM, Petenate AJ, and Silva MAAP (2005) Performance of the hybrid
with scales used in acceptance testing, we do not intend to hedonic scale as compared to the traditional hedonic, self-adjusting and ranking
scales. Food Quality and Preference 16: 691–703.
show the superiority of a single method compared to others, Wichchhukit S and O’Mahony M (2014) The 9-point hedonic scale ranking in food
but direct the researcher’s choice of the appropriate and valid science: some reappraisals and alternatives. Journal of the Science of Food and
method, according to the study objective. Agriculture. http://dx.doi.org/10.1002/jsfa.6993.