You are on page 1of 5

Food Quality and Preference 44 (2015) 70–74

Contents lists available at ScienceDirect

Food Quality and Preference


journal homepage: www.elsevier.com/locate/foodqual

Short Communication

The concurrent use of JAR and CATA questions in hedonic scaling is


unlikely to cause hedonic bias, but may increase product discrimination
Sara R. Jaeger a,⇑, Denise C. Hunter a, Karrie Kam a, Michelle K. Beresford a, David Jin a, Amy G. Paisley a,
Sok L. Chheang a, Christina M. Roigard a, Gastón Ares b
a
The New Zealand Institute for Plant & Food Research Limited, 120 Mt Albert Road, Private Bag 92169, Victoria Street West, Auckland, New Zealand
b
Departamento de Ciencia y Tecnología de Alimentos, Facultad de Química, Universidad de la República, Gral. Flores 2124, C.P. 11800 Montevideo, Uruguay

a r t i c l e i n f o a b s t r a c t

Article history: The concurrent use of CATA questions (check-all-that-apply) with hedonic scaling in product research
Received 26 March 2015 with consumers has gained acceptance in recent years. However, CATA questions do not enable product
Received in revised form 31 March 2015 developers to identify consumers’ preferred intensity of a sensory attribute directly, and for this reason
Accepted 1 April 2015
just-about-right (JAR) questions are sometimes used together with questions that quantify attribute pres-
Available online 8 April 2015
ence and/or intensity. Using JAR questions, consumers have to describe each of the samples in terms of
each specific attribute, typically: 1 = ‘Not enough. . .’, 3 = ‘Just about right’, and 5 = ‘Much too. . .’. Past
Keywords:
research has suggested that co-elicitation of hedonic and JAR responses bias the hedonic data. This unde-
Research methods
Consumer research
sired effect could extend to the case where JAR questions are used concurrently with hedonic and CATA
Just-about-right questions questions. In seven studies (N = 762) across seven product categories no such evidence was found.
Check-all-that-apply questions Additionally, there was some evidence pointing to an increase in product discrimination based on hedo-
nic and CATA responses. Preliminary data indicated that the inclusion of JAR questions in product tests
also eliciting hedonic and CATA responses did not negatively affect consumers’ perception of the task,
which, on average, was still regarded as easy and not tedious. Future research is required to validate
the findings across more varied experimental settings.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction identify consumers’ preferred intensity of a sensory attribute


directly, which is a limitation in terms of using the data to direct
In new product development, ability to identify offerings that product reformulation. For this reason, just-about-right (JAR) ques-
closely align to consumer preferences and match their product tions are popular with product developers in industry, and are
expectations is paramount (van Kleef, van Trijp, & Luning, 2006). often used together with questions that quantify attribute pres-
This requires a good understanding of consumers’ perception of ence and/or intensity (Popper, 2014).
products, from both a sensory and a hedonic point of view The JAR question format seeks to determine the optimum inten-
(Worch, 2012). For this reason, questions concerning the sensory sity of a sensory attribute by asking consumers if they consider
characteristics of products are often included in consumer tests, that a sensory attribute is too strong, too weak or just-about-right
along with overall liking questions (Lawless & Heymann, 2010). (Lawless & Heymann, 2010). According to Prescott, Lee, and Kim
One of the methods used for this purpose is CATA questions (2011), tasks that force consumers to focus their attention on
(check-all-that-apply), in which consumers have to select from a specific sensory characteristics can induce an analytical mind-set,
list of terms all those that they consider applicable to describing which can inhibit representation of synthetic characteristics, such
a focal sample. This question format has been reported to be easy as overall liking.
and intuitive for consumers, providing valid and repeatable infor- Evidence of hedonic bias due to the inclusion of sensory ques-
mation without influencing hedonic responses (Ares & Jaeger, tions is contradictory (Moskowitz, Muñoz, & Gacula, 2003). Some
2015a; Jaeger & Ares, 2014; Jaeger et al., 2013). Despite these authors have reported that JAR questions biased hedonic scores,
advantages, CATA questions do not enable product developers to leading to changes in conclusions regarding sample ranking based
on overall liking (Earthy, MacFie, & Hedderley, 1997; Popper,
Rosenstock, Schraidt, & Kroll, 2004). On the contrary, Popper,
⇑ Corresponding author. Tel.: +64 99257000; fax: +64 99257001. Schraidt, and Kroll. (2005) and Jaeger et al. (2013) reported lack
E-mail address: sara.jaeger@plantandfood.co.nz (S.R. Jaeger). of hedonic bias associated to these questions.

http://dx.doi.org/10.1016/j.foodqual.2015.04.001
0950-3293/Ó 2015 Elsevier Ltd. All rights reserved.
S.R. Jaeger et al. / Food Quality and Preference 44 (2015) 70–74 71

No extant research into the influence of JAR question on hedo- to allow 2–3 bites/sips per sample. Samples were always presented
nic bias has considered the case of using JAR questions in product in cups labelled with 3-digit random codes at room temperature.
tests where consumer perform hedonic scaling and also answer
CATA questions. The present research addresses this gap. It also 2.3. Procedure
addresses the concern that the inclusion of JAR questions in pro-
duct tests that already ask consumers to answer hedonic and The procedure for data collection in all studies was similar.
CATA questions could make them longer/more tedious. This could Between-subjects experimental designs were always used,
decrease consumers’ attention to the task and detrimentally affect comparing responses from two experimental treatments.
their ability to identify differences among samples. Based on Experimental Treatment A was always ‘Hedonic + CATA’, meaning
preliminary analyses, Ares and Jaeger (2015b) suggested such an that participants in this group provided hedonic responses to the
effect when CATA questions were used concurrently with hedonic presented samples and answered a CATA question. In
scaling. Overall, the aim of the present work was to explore the Experimental Treatment B, participants always provided hedonic
influence of JAR questions on hedonic and sensory product scores and answered a CATA question, followed by a JAR question.
discrimination. In all studies, hedonic responses were obtained on a 9-point
Sample discrimination was of particular interest in the present labelled category scale (1 = ‘dislike extremely’, 9 = ‘like extremely’).
work. Research has shown that the strategy used by consumers to The CATA terms used in each study were based on previous
evaluate samples can improve their ability to detect differences research using these product categories and/or pilot work with
among them. For example, Boutrolle, Delarue, Köster, Aranz, and staff from Plant & Food Research. The CATA lists contained 12–16
Danzart (2009) and Chae, Lee, and Lee (2010) reported that induc- terms and covered multiple sensory modalities (appearance,
ing affective strategies for sample evaluation can increase their aroma, flavour/taste, texture, after taste/mouth feel). Consumers
ability to detect differences among samples in authenticity and were asked to check the terms they considered appropriate for
same-different tests. Therefore, it could be hypothesized that describing samples.
attribute questions can modify the strategies used for sample Three JAR questions were included in each of the studies. The
evaluation and influence discriminative ability. However, to the specific attributes were different, as dictated by the focal product
authors’ knowledge no study has evaluated the influence of category, but pertained to appearance, flavour and/or texture char-
attribute questions, including JAR questions, on the discriminative acteristics. Consumers’ task was to describe each of the samples in
ability of hedonic scores. terms of each specific attribute, using a 5-point scale (1 = Not
enough. . ., 3 = Just about right, 5 = Much too. . .).
In Studies 1 and 3, participants answered two Likert questions
2. Materials and methods
immediately after completion of the task: (i) ‘‘It was easy to
answer the questions about these samples’’; and (ii) ‘‘It was tedious
Seven consumer studies involving seven different product
to answer the questions about these samples’’. The labelled 7-point
categories, with 2–5 samples per study, were conducted.
scale had 1 = ‘disagree extremely’ and 7 = ‘agree extremely’ as end-
Between-subjects experimental designs were used in all studies,
point anchors. These questions have previously been used to
with experimental treatments that compared ‘Hedonic + CATA’
explore consumers’ task perception of CATA questions and variants
product evaluation tasks with ‘Hedonic + CATA + JAR’ tasks.
hereof (Jaeger & Ares, 2014).
Table 1 provides an overview of the studies.
Participants attended research sessions in groups of 10–12 peo-
ple. Fifty-nine of the participants in Study 3 also took part in Study
2.1. Participants 6, but completed a different experimental treatment (i.e., A then B
or B then A).
A total of 762 consumers participated in the seven studies. The All data collection took place in standard sensory booths under
number of consumers in each study ranged from 105 to 135. white light. Temperature and airflow was regulated. Samples were
Participants were registered on a database maintained by a profes- presented in monadic sequence, following experimental designs
sional recruitment firm and were screened in accordance with that were balanced for order and carry-over effects. The order in
eligibility criteria for each of the studies. They attended research which the terms were listed in the CATA question was different
sessions at the Plant & Food Research Sensory Facility in for each product and each participant, following a design balanced
Auckland. Participants gave written informed consent to partici- for presentation order (Williams’ Latin Square), according to the
pants and were compensated in cash. recommendations put forward by Ares et al. (2014).
All participants lived in the greater Auckland region (New For classification purposes, participants’ ages, gender and con-
Zealand). All participants were aged 20–70 years old and the per- sumption frequency of target product were recorded. In all
centage of female participants ranged from 47% to 67%. instances, differences between the participant profiles (age, gender
Participants represented households in diverse socio-economic and frequency of product consumption) of the experimental treat-
strata, but none of the consumer samples was representative of ment groups were non-significant (p > 0.11), with the exception of
the general New Zealand population. In studies 1–3 and 5–7, par- age in Study 7 (p = 0.003).
ticipants were overwhelmingly of Caucasian ethnicity. Participants Data were always collected as part of sessions that featured
in Study 4 were of Han Chinese ethnicity and recent immigrants to multiple tasks including several product categories and research
New Zealand (less than 3 years) and all aspects of this study were methods. Only data relevant to the aims of this research are pre-
conducted in Mandarin. sented here.

2.2. Samples 2.4. Data analysis

All samples in Studies 1, 3 and 6 were commercially available in For each study, linear mixed modelling was performed to
New Zealand and purchased from local supermarkets. Samples in uncover significant differences in hedonic ratings across experi-
Studies 2, 4, 5 and 7 were either advanced selections grown under mental treatments. Treatments, samples and their interaction were
commercial conditions or commercially available cultivars (apple, specified as fixed effects, whereas consumer (within experimental
pear, kiwifruit and mussels). Serving sizes were always sufficient treatment) was specified as a random effect. A linear mixed model
72 S.R. Jaeger et al. / Food Quality and Preference 44 (2015) 70–74

Table 1
Overview of studies and summary of results.

Study ID Study description Experimental treatments Results summary of linear mixed model
Study 1 Fruit cakes A (n = 67): Hedonic + CATA FExp.Tr. = 0.04 p = 0.846
5 Samples B (n = 68): Hedonic + CATA + JAR FSample = 5.58 p < 0.0001
15 CATA terms FExp.Tr.*sample = 2.00 p = 0.093
3 JAR questions
Study 2 Mussels A (n = 54): Hedonic + CATA FExp.Tr. = 2.91 p = 0.104
2 Samples B (n = 51): Hedonic + CATA + JAR FSample = 6.55 p = 0.011
19 CATA terms FExp.Tr.*sample = 3.15 p = 0.078
3 JAR questions
Study 3 Milk chocolates A (n = 58): Hedonic + CATA FExp.Tr. = 0.15 p = 0.701
3 Sample B (n = 58): Hedonic + CATA + JAR FSample. = 11.50 p = 0.002
12 CATA terms FExp.Tr.*sample = 0.63 p = 0.534
3 JAR questions
Study 4 Pear A (n = 55): Hedonic + CATA FExp.Tr. = 5.00 p = 0.039
3 Samples B (n = 55): Hedonic + CATA + JAR FSample = 12.81 p < 0.0001
16 CATA terms FExp.Tr.*sample = 1.10 p = 0.334
3 JAR questions
Study 5 Apple A (n = 60): Hedonic + CATA FExp.Tr. = 0.03 p = 0.867
4 Samples B (n = 59): Hedonic + CATA + JAR FSample. = 13.02 p < 0.0001
16 CATA terms FExp.Tr.*sample = 2.60 p = 0.052
3 JAR questions
Study 6 Peanuts A (n = 59): Hedonic + CATA FExp.Tr. = 0.00 p = 0.973
3 Samples B (n = 59): Hedonic + CATA + JAR FSample = 34.17 p < 0.0001
12 CATA terms FExp.Tr.*sample = 1.00 p = 0.371
3 JAR questions
Study 7 Green kiwifruit A (n = 59): Hedonic + CATA FExp.Tr. = 0.05 p = 0.832
4 Samples B (n = 59): Hedonic + CATA + JAR FSample = 0.55 p = 0.645
12 CATA terms FExp.Tr.*sample = 1.06 p = 0.368
3 JAR questions

was used on data from each experimental treatment, considering ranking in terms of overall liking. This can be seen by comparing
sample as a fixed effect and consumer as a random effect. mean overall liking scores between experimental treatments in
Tukey’s test was used for post hoc pairwise comparisons of sample Table 3.
means, at a significance level of 5%. Taken together, these results suggest that the addition of JAR
For each study, the frequency of use of each CATA term was questions to a task involving hedonic and CATA questions is unli-
determined by counting the number of consumers who used that kely to affect overall liking scores. This is in agreement with the
term to describe each sample. Fisher’s exact test (Fisher, 1954) findings of Popper et al. (2005) and Jaeger et al. (2013), who
was used to determine significant differences in the frequency of reported lack of hedonic bias associated to JAR questions. It may
use of the terms between experimental treatments. Cochran’s Q be noteworthy that the only study in which JAR questions affected
test (Manoukian, 1986) was carried out separately on data from hedonic scores was carried out with Chinese consumers (Study 4).
each experimental treatment to identify significant differences Based on data collected with consumers in Korea, Prescott et al.
among samples on each of the sensory terms, for each of the (2011) reported that an analytical mind-set could affect hedonic
experimental treatments. ratings. Thus, it could be relevant to explore further if cultural fac-
Task perception data were analysed using analysis of variance tors modulate the occurrence of hedonic bias due to the inclusion
(ANOVA), with Experimental treatment as the fixed effect. of attribute questions.
All data analyses were carried out using R (R Core Team, 2014). When hedonic data from both experimental treatments were
considered, significant differences among samples were estab-
lished in all studies, except for Study 7 (Table 1). This could be
3. Results and discussion1
due to very small differences between the kiwifruit samples
included in this study, which were induced by different posthar-
3.1. Influence of JAR questions on hedonic scores
vest treatments. When analysing the data in greater detail, it was
found that conclusions regarding the existence of significant differ-
As shown in Table 1, in six of the seven studies the inclusion of
ences between samples, and the magnitude of any differences,
JAR questions did not significantly affect hedonic scores when con-
depended on the experimental treatment. In Studies 1, 2 and 5, sig-
currently collected with CATA questions (p > 0.10). However, in
nificant differences between samples were found in Experimental
Study 4 there was a significant effect of the JAR question on hedo-
Treatment B (‘Hedonic + CATA + JAR’) but not in Experimental
nic scores (p = 0.039). On average, hedonic scores were signifi-
Treatment A (‘Hedonic + CATA’) (Table 2). Besides, the F-values in
cantly lower when the JAR scale was included than when only a
Studies 3 and 4 were higher when consumers also answered
CATA question was considered (6.1 vs. 6.6, respectively). The
CATA and JAR questions than when only the CATA question was
interaction between experimental treatment and sample was non-
used concurrently with hedonic scaling (7.98 vs. 4.50 and 8.37
significant in all the studies (p > 0.078), suggesting that the inclu-
vs. 4.73, respectively). The sample groupings determined by
sion of JAR questions did not affect conclusions regarding sample
Tukey’s test are shown in Table 3 and additionally support the
interpretation that the discrimination of samples based on hedonic
1
It is beyond the scope of this paper to report the product-specific attribute scores tended to improve when the hedonic responses were con-
information generated in the reported studies. Interested readers may contact the
authors for further information.
currently elicited with a combination of CATA and JAR questions
S.R. Jaeger et al. / Food Quality and Preference 44 (2015) 70–74 73

Table 2
Results of the ANOVA performed on hedonic scores for each experimental treatment, average percentage of terms selected by consumers to describe samples and percentage of
CATA terms for which significant differences between samples were identified, for consumers using CATA and CATA + JAR for describing samples in the seven studies, in addition
to providing hedonic responses.

Study ID Experimental treatments F-value and p-value Average percentage of CATA terms Percentage of CATA terms for which
(in brackets) in ANOVA selected for describing significant differences between samples
samples (%) were identified according to Cochran’s Q test (p < 0.05) (%)
Study 1 A (n = 67): Hedonic + CATA 1.28 (0.277) 29 100
B (n = 68): 6.81 (<0.001) 31 100
Hedonic + CATA + JAR
Study 2 A (n = 54): Hedonic + CATA 0.85 (0.357) 34 47
B (n = 51): 7.84 (<0.001) 31 53
Hedonic + CATA + JAR
Study 3 A (n = 58): Hedonic + CATA 4.50 (0.013) 42 50
B (n = 58): 7.98 (<0.001) 41 83
Hedonic + CATA + JAR
Study 4 A (n = 55): Hedonic + CATA 4.73 (0.010) 28 38
B (n = 55): 8.37 (<0.001) 28 69
Hedonic + CATA + JAR
Study 5 A (n = 60): Hedonic + CATA 2.37 (0.071) 30 81
B (n = 59): 14.24 (<0.001) 28 100
Hedonic + CATA + JAR
Study 6 A (n = 59): Hedonic + CATA 18.54 (<0.001) 37 83
B (n = 59): 17.54 (<0.001) 35 83
Hedonic + CATA + JAR
Study 7 A (n = 59): Hedonic + CATA 0.31 (0.818) 35 67
B (n = 59): 1.28 (0.282) 38 67
Hedonic + CATA + JAR

(four out of six studies; ignoring Study 7, where sample differences Table 3
were always non-significant). The exception to this trend was Mean and standard deviations for overall liking ratings obtained across the seven
studies. Ratings were collected on a 9-point hedonic scale (1 = ‘dislike extremely’ and
Study 4, and for completeness it is noted that in Study 6 the F-value
9 = ‘like extremely’). Treatment A is the control condition (‘Hedonic + CATA’) and
for sample effect was incrementally larger when the JAR question Treatment B includes the JAR question (‘Hedonic + CATA + JAR’). Refer to Table 1 for
was not included (18.54 vs. 17.54). full details of experimental treatments.
Taken together, these results provide evidence that the addition
Study ID Sample Treatment A Treatment B
of JAR questions is likely to improve consumers’ discrimination of
samples based on hedonic scores, an outcome that could be attrib- Study 1 (fruit cakes) Sample 1 5.8a ± 1.6 6.0a ± 2.0
Sample 2 5.3a ± 1.6 4.7b ± 1.6
uted to an increase in consumers’ engagement and attention to the Sample 3 6.1a ± 2.0 5.8a ± 1.8
task. This explanation is supported by previous reports of a strong Sample 4 5.5a ± 1.6 6.1a ± 1.6
relationship between the ability to concentrate and trained asses- Sample 5 6.0a ± 2.0 6.2a ± 2.0
sors’ performance (Lesschaeve & Issanchou, 1996). Furthermore, Study 2 (mussels) Sample 1 6.5a ± 1.8 6.8a ± 1.6
affective strategies for sample evaluation have been reported to Sample 2 6.2a ± 1.6 5.8b ± 2.0
increase consumers’ engagement and improve their performance Study 3 (milk chocolate) Sample 1 6.6a ± 1.6 6.9a ± 2.0
in authenticity and same-different tests (Boutrolle et al., 2009; Sample 2 7.0a ± 1.6 6.9a ± 1.6
Chae et al., 2010). Sample 3 6.1a,b ± 2.0 5.9b ± 1.8
Study 4 (pear) Sample 1 7.0a ± 1.4 6.7a ± 1.5
Sample 2 6.5b ± 1.5 6.2a ± 2.0
3.2. Influence of JAR questions on consumers’ responses to CATA Sample 3 6.2b ± 1.6 5.3b ± 2.1
questions Study 5 (apple) Sample 1 6.6a ± 1.7 6.9a ± 1.4
Sample 2 5.7a ± 1.9 5.6b ± 1.9
The addition of JAR questions to the task did not influence con- Sample 3 6.2a ± 2.0 6.9a ± 1.5
sumers’ CATA responses. According to Fisher’s exact test, the aver- Sample 4 5.8a ± 2.1 5.2b ± 2.0
age percentage of CATA terms used by consumers for describing Study 6 (peanuts) Sample 1 5.5b ± 1.7 5.8b ± 1.7
samples was not significantly different between the two experi- Sample 2 7.3a ± 1.3 7.3a ± 1.2
Sample 3 5.9b ± 1.9 5.6b ± 2.1
mental treatments (‘Hedonic + CATA’ and ‘Hedonic + CATA + JAR’)
(Table 2). Study 7 (green kiwifruit) Sample 1 6.2a ± 1.8 5.7a ± 1.6
Sample 2 6.3a ± 1.6 6.2a ± 2.0
However, the discriminative ability of the CATA question was
Sample 3 6.1a ± 1.8 6.3a ± 1.8
improved by the concurrent use of JAR questions. As shown in Sample 4 6.0a ± 1.6 6.3a ± 1.6
Table 2, the percentage of terms for which significant differences
Mean overall liking ratings with different superscripts within a column, for each
between samples were established according to Cochran’s Q test
study, are significantly different according to Tukey’s test for a 5% significance level.
differed in four of the seven studies (Studies 2–5). In the remaining
three studies (Studies 1, 6 and 7), discriminative ability was iden-
tical. Thus, it appears that JAR questions have the potential to 3.3. Influence of JAR questions on task perceptions
improve consumers’ ability to identify differences among samples
in terms of sensory characteristics, and, moreover, that the concur- In Studies 1 and 3, participants rated their perceived task ease/
rent use of this question format is unlikely to affect sample tediousness, and the responses were closely aligned with those
discrimination negatively. previously reported by Jaeger and Ares (2014). The inclusion of a
74 S.R. Jaeger et al. / Food Quality and Preference 44 (2015) 70–74

JAR question did not significantly affect perceived task ease Ministry for Business, Innovation & Employment, and from Plant
(p > 0.26) or tediousness (p > 0.39), relative to when only hedonic & Food Research.
and CATA responses were elicited. On average, consumers agreed
strongly that the task was easy (5.6–5.8 on the 7-point scale) and References
disagreed strongly that the task was tedious (2.2–2.5 on the
7-point scale). These responses are in agreement with those Ares, G., Etchemendy, E., Antúnez, L., Vidal, L., Giménez, A., & Jaeger, S. R. (2014).
Visual attention by consumers to check-all-that-apply questions: Insights to
previously reported by Jaeger and Ares (2014), but should be repli- support methodological development. Food Quality and Preference, 32, 210–220.
cated to verify their robustness. Ares, G., & Jaeger, S. R. (2015a). Check-all-that-apply (CATA) questions with
consumers in practice. Experimental considerations and impact on outcome. In
J. Delarue, J. B. Lawlor, & M. Rogeaux (Eds.), Rapid sensory profiling techniques
4. Conclusions and related methods (pp. 227–245). Sawston, Cambridge: Woodhead Publishing.
Ares & Jaeger, S. R. (2015b). Examination of sensory product characterization bias
when check-all-that-apply (CATA) questions are used concurrently with
Contrary to previous reports, there was no evidence in this
hedonic assessments. Food Quality and Preference, 40, 199–208.
research that JAR questions were associated with hedonic bias. Boutrolle, I., Delarue, J., Köster, E., Aranz, D., & Danzart, M. (2009). Use of a test of
When JAR questions were included in product tests that also eli- perceived authenticity to trigger affective responses when testing food. Food
cited CATA responses, consumers’ hedonic product responses were Quality and Preference, 20, 418–426.
Chae, J. E., Lee, Y. M., & Lee, H. S. (2010). Affective same–different discrimination
not affected. This result was obtained in seven different studies, tests for assessing consumer discriminability between milks with subtle
but should be replicated in other experimental settings, including differences. Food Quality and Preference, 21, 427–438.
those where more than three JAR questions are used. There was Earthy, P. J., MacFie, H. J. H., & Hedderley, D. (1997). Effect of question order on
sensory perception and preference in central location trials. Journal of Sensory
some evidence that the inclusion of JAR questions increased pro- Studies, 12, 215–237.
duct discrimination based on hedonic and CATA responses. This Fisher, R. A. (1954). Statistical methods for research workers. Edinburgh: Oliver and
has not previously been reported and warrants further investiga- Boyd.
Jaeger, S. R., & Ares, G. (2014). Lack of evidence that concurrent sensory product
tion. In this regard, it would be interesting also to establish characterisation using CATA questions bias hedonic scores. Food Quality and
whether the order of JAR and CATA questions is consequential. Preference, 35, 1–5.
Also, further research should explore if changes in product discrim- Jaeger, S. R., Giacalone, D., Roigard, C. M., Pineau, B., Vidal, L., Gimenez, A., et al.
(2013). Investigation of bias of hedonic scores when co-eliciting product
ination occur when only JAR questions are concurrently used to attribute information using CATA questions. Food Quality and Preference, 30,
hedonic scores. In this research, hedonic responses were obtained 242–249.
first, followed by CATA questions and lastly JAR questions. Lawless, H. T., & Heymann, H. (2010). Sensory evaluation of food. Principles and
practices (2nd ed.). New York: Springer.
Despite the additional consumer burden, the inclusion of JAR
Lesschaeve, I., & Issanchou, S. (1996). Could selection tests detect the future
questions was not associated with an increase in perceived task performance of descriptive panellists? Food Quality and Preference, 7, 177–183.
difficulty and tediousness. However, it should be noted that in Manoukian, E. B. (1986). Mathematical nonparametric statistics. New York, NY:
the present work consumers evaluated only two to five samples. Gordon & Breach.
Moskowitz, H. R., Muñoz, A. M., & Gacula, M. C. (2003). Viewpoints and controversies
As the number of samples increases, the inclusion of JAR questions in sensory science and consumer product testing. Trumbull, Connecticut:
could increase task tediousness and decrease consumer discrim- Blackwell Publishing.
ination. In this sense, research extending results from the present Popper, R. (2014). Use of just-about-right scales in consumer research. In P. Varela &
G. Ares (Eds.), Novel techniques in sensory characterization and consumer profiling
work to studies with more samples is necessary. (pp. 365–389). Boca Raton: CRC Press.
Popper, R., Rosenstock, W., Schraidt, M., & Kroll, B. J. (2004). The effect of attribute
questions on overall liking ratings. Food Quality and Preference, 15, 853–858.
Author contributions
Popper, R., Schraidt, M., & Kroll., B. J. (2005). When do attribute ratings affect overall
liking ratings?. In Sixth pangborn sensory science symposium. Davis, DA:
S.R.J. and G.A. developed the research and wrote the paper. G.A. University of California [7–11 August].
Prescott, J., Lee, S. M., & Kim, K. (2011). Analytic approaches to evaluation modify
analysed the data. All other authors contributed to data collection.
hedonic responses. Food Quality and Preference, 22, 391–393.
R Core Team. (2014). R: A language and environment for statistical computing. Vienna,
Acknowledgements Austria: R Foundation for Statistical Computing.
van Kleef, E., van Trijp, H. C. M., & Luning, P. (2006). Internal versus external
preference analysis: An exploratory study on end-user evaluation. Food Quality
All members of the Sensory & Consumer Science team at Plant & and Preference, 17, 387–399.
Food Research are thanked for help in planning and collection of Worch, T. (2012). The ideal profile analysis: From the validation to the statistical
data. Financial support was received from The New Zealand analysis of ideal profile data [Ph.D. thesis]. Rennes, France: Agrocampus-Ouest.

You might also like