Holthuysen (2017)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Received: 13 July 2016 | Revised: 5 December 2016 | Accepted: 19 January 2017

DOI 10.1111/joss.12254

ORIGINAL ARTICLE

“Welcome on board”: Overall liking and just-about-right ratings


of airplane meals in three different consumption contexts—
laboratory, re-created airplane, and actual airplane

 A. de Wijk | Stefanie Kremer


Nancy T. E. Holthuysen | Milou N. Vrijhof | Rene

Consumer Science and Health, Wageningen


University and Research Centre,
Abstract
Wageningen Food & Biobased Research, The main objective of this study was to evaluate the effect of contexts on overall liking and just-
Wageningen, The Netherlands about-right (JAR) ratings of airplane meals. A rice dish (meal type A) and a pasta dish (meal type B)
Correspondence were assessed. Per meal type, two variants were produced (variant 1 and 2). Two hundred forty-
N. Holthuysen, Consumer Science and
two consumers were randomly allocated to evaluate one of the four meals, first in a laboratory
Health, Wageningen University and
Research Centre, Wageningen Food & setting and then in a re-created airplane environment. In addition, 222 passengers did the same
Biobased Research, Bornse Weilanden 9, assessments during an actual flight. Specific meals (A1, B1) were less liked in the laboratory than in
6708 PD Wageningen, The Netherlands. the re-created airplane. In general, no differentiation in overall liking occurred per meal type
Email: nancy.holthuysen@wur.nl
between the two tested variants in the laboratory, whereas these two variants were significantly
Funding information
differentiated in liking in the re-created airplane and the actual airplane. Mean overall liking ratings
DLO/TNO 2013 Topsector Agri&Food, the
Dutch Ministry of Economic Affairs, in the re-created airplane did not significantly differ from the mean overall liking ratings in the
Agriculture, and Innovation actual airplane. The observed JAR ratings did not differ much between the re-created airplane and
the actual airplane. In summary, the re-created airplane as a testing location produced more similar
test results to the actual airplane than the traditional laboratory.

Practical applications
Sensory consumer testing in re-created contexts may produce results with a higher external valid-
ity than laboratory testing and therefore offer a cost-efficient alternative to extensive sensory
consumer testing in real-life contexts.

1 | INTRODUCTION contextual setting to increase the external validity of the results of


nez, Vidal, & Ares, 2014; Meisel-
these consumer tests (Machín, Gime
Real-life human eating behavior does not take place in an environmen- man, 2013). However, real-life testing as an integral part of the new
tal vacuum; rather, it happens in context. The context in which con- product development cycle comes at a price, as these tests are more
sumers eat foods influences their hedonic appraisal of the consumed burdensome and costly to perform than classical central location tests
foods (de Graaf et al., 2005; Edwards, Meiselman, Edwards, & Lesher, —even more so in the case of foods that are intended for consumption
2003; King, Meiselman, Hottenstein, Work, & Cronk, 2007). Conse- in remote markets such as airplanes.
quently, consumers’ hedonic ratings elicited in a natural consumption Both situational and social context have previously been reported
context have been observed to differ from those elicited under con- to impact human eating behavior (Stroebele & de Castro, 2004). Situa-
trolled sensory laboratory conditions (Boutrolle, Delarue, Arranz, tional context comprises mainly the physical surroundings of the con-
€ster, 2007; Petit & Sieffermann, 2007). In line with these
Rogeaux, & Ko sumer, that is, the situation in which the food is typically consumed
empirical observations, a close link between the consumption context (Carvalho et al., 2015). Social context comprises mainly the people
and how consumers feel has been postulated (Desmet & Schifferstein, involved, that is, whether foods are consumed alone or in a group, or
2008), and these feelings are thought to be underlying modulators of whether they are consumed together with family or with strangers (de
food perception, food liking, and overall enjoyment of human eating Castro, Brewer, Elmore, & Orozco, 1990; de Castro & de Castro, 1989).
experiences (Hartwell, Edwards, & Brown, 2013). Consequently, it is Although social context is considered to be equally important as physi-
strongly recommended to assess foods and drinks in the appropriate cal context (Jaeger & Rose, 2008), only few attempts have been made

J Sens Stud. 2017;e12254. wileyonlinelibrary.com/journal/joss V


C 2017 Wiley Periodicals, Inc. | 1 of 8
https://doi.org/10.1111/joss.12254
2 of 8 | HOLTHUYSEN ET AL.

to incorporate it on a structural basis into sensory consumer testing. Weber, Meiselman, & Lv, 2004; Meiselman, Johnson, Reeve, & Crouch,
Recently, the incorporation of situational context into sensory con- 2000), it was hypothesized that higher overall liking ratings would gen-
sumer testing has received considerable scientific attention. For exam- erally be observed in the more realistic testing contexts (REC and ACT)
ple, situational simulation methods have tried to evoke a consumption and that the hedonic discrimination within a meal type would be
situation either by means of providing a descriptive text (Hein, Hamid, improved in these contexts.
Jaeger, & Delahunty, 2010, 2012) or a text in combination with a situa- A second aim was to explore possible differences in consumers’
tional image (Piqueras-Fiszman & Jaeger, 2014a,b,c). Other recent JAR ratings after consumption of the complete meal under either the
studies applied immersive environments to consumer food choice test- normal pressure condition (REC) or the low pressure condition (ACT). It
ing. For example, a virtual coffeehouse—replete with visual, auditory, was hypothesized that the majority of the observed JAR ratings would
and olfactory cues found regularly in this setting—was created (Bang- be significantly different between those two conditions because of
cuyo et al., 2015) or a bar-like environment—with different furniture changes in taste perception that are thought to be caused by low pres-
and clips with visual and music stimuli projected on walls to change the sure conditions.
overall warmth of the ambience (Sester, Deroy, Sutan, & Galia, 2013).
Currently, different results on the impact of testing environments on 2 | METHODS
sample ratings are reported. Significant differences in mean hedonic
ratings per sample and/or a greater sample discrimination have been 2.1 | Participants
reported by some (Bangcuyo et al., 2015; Hein et al., 2010), whereas
For the laboratory and the re-created airplane context study, consum-
others did not find these differences (Hersleth, Ueland, Allain, & Næs,
ers were recruited via the database of Wageningen Food and Biobased
2005).
Research. They were invited to participate in a consumer study with
Flying in an aircraft generally has a great physiological and psycho-
warm, ready-made meals. All participants met the following criteria:
logical impact on passengers, and this impact will most likely influence
minimum age of 18 years, not suffering from food allergies, not on a
their eating behavior. Physiological influences felt by humans in air-
prescribed diet, having traveled on an airplane at least once within the
planes are the reduced atmospheric pressure, low humidity, back-
last 10 years, and with a good command of the Dutch language. Of the
ground noise, cosmic radiation, and magnetic-field exposure (Butler,
242 consumers that participated, 35.6% were male and 64.4% were
Nicholas, Lackland, & Friedberg, 2000). In addition, limited space to
female (mean age 56, range 18–79 years). All the consumers who took
move, stress, and boredom, and the close vicinity of strangers are fac-
part in the LAB and REC studies signed an informed consent and
tors that can psychologically influence humans traveling in airplanes.
received a monetary fee for participation after completion of the stud-
Higher taste thresholds have been reported under lower pressure con-
ies. For the actual airplane study, passengers were recruited on board
ditions (Burdack-Freitag, Bullinger, Mayer, & Breuer, 2011; Maga &
an airplane. After they signed the informed consent, they received the
Lorenz, 1972), and background noise has been reported to suppress
onboard meal free of charge. Two hundred twenty-two passengers
saltiness, sweetness, and overall enjoyment of food (Spence, 2012;
participated; they were a minimum of 18 years old, not suffering from
Woods et al., 2011) and hypothesized to not effect umami taste per-
food allergies, and not on a prescribed diet. Of these, 46.6% were male
ception (Spence, Michel, & Smith, 2014). If all these influences are
and 53.4% female (mean age 51, range 19–75 years). Table 1 provides
combined, it seems that optimizing airplane meals would ideally require
more details on their characteristics. The study was approved by the
extended consumer testing in real-life airplanes. However, this is not
Social Science Ethics Committee of Wageningen University and
feasible within airlines’ day-to-day practices.
Research Centre, The Netherlands.
Therefore, this study aims to evaluate the effect of various testing
contexts on overall liking ratings of airplane meals. Whereas most pre-
vious studies only compare results from lab situations with results from 2.2 | Food samples
semi-real-life contexts, this study includes results from testing in the Two commercially available top sellers were selected from a broader
real-life situation. Therefore, testing was conducted in three different range of airplane meals: a chicken curry rice dish (meal type A) and a
contexts: (1) a classical laboratory context (LAB); (2) a semi-real-life pasta bolognese dish (meal type B). For each meal type, two variants
context that re-created as much as possible the situational and social were selected that were produced by two different commercial air-
context of the consumption of airplane meals on airplanes (re-created plane meal producers according to their own recipe (variant 1 and vari-
airplane (REC); and (3) a real-life test on board an airplane during a real ant 2). The meals were heated in a pre-heated oven at 1008C for 45
flight (ACT). The test procedure used in the laboratory reflects the cur- min.
rent standard procedures in airplane meal testing and thus liking ratings
are assessed based on one-bite evaluations only, whereas in the two
2.3 | Procedure
other contexts both liking and just-about-right (JAR) ratings are given
after consumption of the complete meal. On the basis of previously 2.3.1 | Study 1: Laboratory context (LAB)
reported observations (Boutrolle, Arranz, Rogeaux, & Delarue, 2005; Study 1 is the standard method used by companies to study the overall
Boutrolle et al., 2007; Hersleth, Mevik, Næs, & Guinard, 2003; King, liking of meals. The study in the LAB context lasted approximately 25
HOLTHUYSEN ET AL. | 3 of 8

TA BL E 1 General characteristics of the participants in both studies: means (SD) or %

LAB and REC ACT p value

Sample size (n) 242 222

Male 35.6% 46.6% .018

Female 64.4% 53.4%

Age (year) 56.3 (16.3) 51.4 (13.4) .001

Travel frequency .001


A few times a year 36.1% 44.5%
Once a year 39.5% 37.3%
Once every 2–3 years 23.1% 14.5%
Less than once every 3 years 1.3% 3.6%

Buy meal on board .161


Every flight 6.7% 8.6%
Almost every flight 13.8% 13.6%
Occasionally 34.3% 38.6%
I have never bought a meal on board 45.2% 39.1%

min and was conducted at the Rijn IJssel vocational college, Wagenin- were asked to rate the same hedonic question as before, in addition
gen, The Netherlands. At the beginning of the laboratory session, the they had to rate a few sensory attribute questions on a JAR scale (i.e.,
consumers were asked to indicate whether they would prefer to eat a satiated assessment [REC2]).
rice or a pasta dish. Consumers with a preference for one of the meal
2.3.3 | Study 3: Actual airplane context (ACT)
types (pasta vs. rice dish) received the meal type of their choice. If con-
sumers expressed no preference between the two meal types, the The study in the ACT context was conducted on board a Dutch airliner
researchers made the selection on the basis of availability. Whether during a regular, round-trip Amsterdam–Tenerife flight (March 15,
the consumers received variant 1 or variant 2 of the selected meal 2014). Normally, during this flight, a buy on board service is in opera-
type was randomly allocated and not communicated to the consumer. tion. Passengers who were both willing and eligible to participate in the
The consumers were individually seated in a laboratory setting and study were offered one of the airplane meals free of charge. The pas-
were asked to assess a few bites of one of the four meals presented in sengers were asked to indicate whether they would prefer to eat a rice
a small polystyrene cup with a random three-digit number. The con- or a pasta dish and, as long as both meal types were available, their
sumers were asked to taste the meal and to evaluate overall liking, on a wishes were met. If consumers expressed no preference, the research-
100 mm horizontal visual analogue scale (VAS) with the verbal anchors ers made the selection on the basis of availability. After completion of
“very little” and “very much” placed at 10% and 90% on the scale. the meal, the passengers received the same questionnaire as the con-
sumers in REC2 had received.
2.3.2 | Study 2: Re-created airplane context (REC1 and
Table 2 provides a schematic overview of the three studies and
REC2)
gives an overview of the number of completed evaluations per meal
The REC study lasted approximately 50 min and was conducted at Rijn type in the different contexts.
IJssel, Wageningen, The Netherlands. The consumers were told that a
flight to Tenerife was going to be re-created as part of the research on
ready-made airplane meals and, each individual received a boarding
pass and hand luggage. Shortly after, they entered a special room with
a re-created part of an airplane with 16 passenger seats (see Figure 1).
A flight attendant assisted them during the boarding. After they were
seated and their luggage was stored, a voice tape was switched on.
The tape started with the usual safety instructions (recorded by a real
flight attendant) and continued with airplane noises throughout the
test. After 20 min, the participants received a complete portion of the
same meal they had in study 1, in its original package. The meals were
served by a flight attendant on airplane trays together with a cup of
water and airplane cutlery. The consumers were asked to take a few
bites from their meal and to evaluate overall liking, (i.e., assessment
after first bite [REC1]). After answering this initial question, they were
asked to eat their meal until satiated. On completion of the meal, they FIGURE 1 Photo re-created airplane setting
4 of 8 | HOLTHUYSEN ET AL.

TA BL E 2 A schematic overview of the three studies, including an


overview of the number of completed evaluations per meal type in
the different contexts

Meal type LAB REC1 REC2 ACT

A1 n 5 60 n 5 60 n 5 60 n 5 80
OL OL OL 1 JAR OL 1 JAR

A2 n 5 61 n 5 61 n 5 61 n 5 54
OL OL OL 1 JAR OL 1 JAR

B1 n 5 61 n 5 61 n 5 61 n 5 51
OL OL OL 1 JAR OL 1 JAR

B2 n 5 60 n 5 60 n 5 60 n 5 37
OL OL OL 1 JAR OL 1 JAR F I G U R E 2 Mean (within-subjects) overall liking ratings (6SEM) in

Note. OL 5 overall liking rating; JAR 5 just-about-right rating.


different consumption contexts (LAB, REC1, and REC2)
Note. Overall liking was assessed on a VAS scale running from 1 to
100 mm; different letters indicate a significant difference at p  .05
2.4 | Data analysis
tically differ in their mean overall liking for either meal type A (59.67
First, a within-subject analysis was conducted with a general linear
vs. 55.15) (t(21.540)5119, p 5 .126) or meal type B (55.85 vs. 51.07)
mixed model analysis taking into account a repeated measure. Per meal
(t(21.526)5119, p 5 .130). In the re-created airplane, the two meal
type, significant differences in hedonic rating between LAB, REC1, and
variants significantly differed for both meal types in the initial assess-
REC2, with overall liking entered as dependent variable and situation
ment (REC1): meal type A: 64.2 versus 57.4 (t(22.609)5119, p 5 .010)
entered as fixed factor, were assessed. Consumers’ individual data
and meal type B: 64.3 versus 53.6 (t(23.994)5119, p  .001). In the
were indicated as random factors.
Once the F test indicated a significant difference between the
overall liking ratings, post hoc LSD comparisons were performed to
determine the consumption context in which the ratings differed signif-
icantly from one another.
Second, per consumption context (LAB, REC1, REC2, and ACT),
the significance of hedonic differences between the two variants (1 vs.
2) within one meal type (i.e., rice (A) or pasta (B)) were assessed by
independent sample t tests.
Third, independent sample t tests were conducted per meal variant
to reveal the significance of differences in the hedonic and/or JAR rat-
ings between the satiated meals assessments in REC2 and ACT. A Bon-
ferroni correction was applied, as the statistical analyses were
performed on each of the 28 JAR ratings separately (Armstrong, 2014).
These results were considered significant at a level of p < .0018.
All other significant differences were defined as p < .05. A signifi-
cance level between 0.05 and 0.10 is described as a trend. The data
analyses were performed using SPSS v. 19.0 (SPSS Inc., Chicago, IL,
USA).

3 | RESULTS

The mean overall liking scores for the different meals are presented
graphically in Figure 2. For meals A1 (F 5 5.95; df 5 2; p 5 .003) and B1
(F 5 10.47; df 5 2; p  .001), the overall liking was significantly higher
in both re-created contexts (REC1 and REC2) than in the laboratory. F I G U R E 3 Differences in mean overall liking (6SEM) between
No such difference was observed for meals A2 and B2. variant 1 and variant 2 of meal A (a) or meal B (b) in the different
consumption contexts
Differences in mean overall liking between variant 1 and variant 2
Note. Overall liking was assessed on a VAS scale running from 1 to
of meal A or meal B in the different consumption contexts are dis- 100 mm; n.s. 5 not significant; ~p  .10 *p  .05, ** p  .01,
played in Figure 3a,b. In the laboratory, the meal variants did not statis- *** p  .001
HOLTHUYSEN ET AL. | 5 of 8

were observed. Per variant, the scores were either normally distributed
or slightly skewed, ranging from 3 to 9 for the more liked variants and
from 1 to 9 for the lesser liked variants.
The mean overall liking scores for the different meals in the plane
are also presented graphically in Figure 3. The two meal variants dif-
fered significantly in their mean overall liking for both meal type A
(66.7 vs. 57.4) (t(22.488)585.2, p 5 .015) and meal type B (62.1 vs.
45.8) (t(23.657)558.3, p  .001).
Differences in mean overall liking ratings between the re-created
and the actual consumption context are displayed per airplane meal in
Figure 4. No significant difference was observed.
No significant differences were observed between the JAR ratings
in the re-created and the actual airplane context (see Table 3).

FIGURE 4 Difference in mean overall liking ratings (6SEM) per 4 | DISCUSSION


airplane meal between the re-created and the actual consumption
context In the current study, overall liking and JAR ratings of airplane meals
Note. Evaluation took place after consumption of the whole meal; were explored in different consumption contexts. Specific meals were
overall liking was assessed on a VAS scale running from 1 to
100 mm; n.s. 5 not significant less liked in the laboratory context than in the re-created consumption
context. In general, no differentiation on overall liking occurred
satiated assessment (REC2), for the two meal variants of meal type A, a between the two tested variants in the laboratory context, whereas
trend was observed (63.3 vs. 58.8) (t(21.678)5119, p 5 .096), and a they were significantly differentiated in both the re-created and the
significant difference was observed for meal type B (62.0 vs. 51.9) (t actual airplane context. Mean overall liking ratings in the re-created
(23.172)5108, p 5 .002). On visual inspection of the individual liking context did not differ significantly from the mean overall liking ratings
scores, no systematic differences in distribution between the variants in the actual airplane context. Surprisingly, no significant differences in

TA BL E 3 Mean JAR ratings (6 SD) between re-created and actual airplane

Meal type A Variant 1 Variant 2

REC2 ACT p value REC2 ACT p value

Taste intensity 20.77 (0.77) 20.44 (0.78) .014 20.25 (0.86) 20.35 (0.73) .498

Fruity 20.57 (0.85) 20.42 (0.90) .346 20.64 (0.80) 20.53 (0.82) .498

Spicy 20.9 (0.86) 20.6 (0.76) .030 20.26 (1.0) 20.22 (0.93) .824

Sweet 20.05 (0.53) 20.09 (0.54) .674 20.03 (0.55) 20.04 (0.44) .958

Bitter 20.15 (0.36) 20.08 (0.45) .305 20.03 (0.37) 20.06 (0.24) .682

Sour 20.27 (0.61) 20.18 (0.47) .347 20.20 (0.55) 20.17 (0.43) .775

Salt 20.22 (0.76) 20.16 (0.61) .640 20.02 (0.74) 20.09 (0.56) .539

Meal type B Variant 1 Variant 2

REC2 ACT p value REC2 ACT p value

Taste intensity 20.36 (0.82) 20.08 (0.60) .037 20.59 (0.93) 20.72 (0.70) .446

Spicy 20.44 (0.83) 20.29 (0.54) .256 20.73 (0.91) 20.57 (0.70) .380

Bitter 20.02 (0.23) 20.06 (0.24) .322 20.08 (0.34) 20.09 (0.28) .989

Sour 20.07 (0.31) 20.06 (0.24) .936 20.07 (0.52) 20.09 (0.29) .834

Salt 0.30 (0.82) 0.14 (0.75) .295 0.14 (0.78) 20.09 (0.56) .114

Mushroom 21.11 (0.92) 20.85 (0.86) .130 21.20 (0.74) 20.91 (0.95) .103

Tomato 20.34 (0.79) 20.33 (0.71) .940 20.58 (0.91) 20.74 (0.78) .370

Note. Significant, p  .0018.


6 of 8 | HOLTHUYSEN ET AL.

JAR ratings were observed between the re-created and the actual air- € ster & Mojet, 2007). For example, in the
(Kinnear & de Kock, 2011; Ko
plane context. current study’s laboratory test, the hedonic ratings were derived after
Food evaluations in natural eating situations may differ from evalu- one brief product exposure to a small amount of the meal. However, in
ations in controlled eating situations (Meiselman, 1992). The current the two other consumption contexts, consumers were either visually
study’s observed lower hedonic ratings for some of the meals in the lab- exposed to (initial assessment in the re-created airplane context) or had
oratory context are in line with a number of previous studies (Boutrolle repeated taste exposures while consuming (satiated assessment in the
et al., 2005, 2007; Hersleth et al., 2003; King et al., 2004; Meiselman re-created and the actual airplane context) a full serving size and evalu-
et al., 2000). However, there are also studies in which a reverse relation ating the meal. Second, de Graaf et al. (2005) reported that the predic-
has been observed (Daillant-Spinnler & Issanchou, 1995). Several possi- tive ability of laboratory ratings also depends on the type of food being
ble explanations for these observed context-dependent differences in assessed. They assumed that lab ratings were more relevant for snacks
the size of hedonic ratings have been suggested. First, clear differences than for complex products such as meals. Therefore, the airplane meals
often exist in the amount of food sampled between lab and field con- in the current study might require presentation and/or consumption in
texts, and this is deemed especially relevant in the assessment of main a realistic portion size to produce relevant hedonic ratings. Third, it
dishes (de Graaf et al., 2005). At first sight, this difference in consumed may also be plausible that, during a laboratory test, participants feel
sample size might also have contributed to the difference in the ratings somewhat bored and less engaged (Bangcuyo et al., 2015) and that this
observed between the laboratory context and both the re-created and in turn contributes to the poor differentiability and predictability of this
the actual airplane context. However, in the re-created airplane context, €ster, 2009). Fourth, since group
kind of consumer-generated data (Ko
two measures were taken (REC1 vs. REC2), and the differences in sizes are quite small in the current study we cannot completely rule out
ratings were already manifest after the initial assessment (REC1). Thus, that between group differences (or non-differences) observed between
it seems unlikely that the actual amount consumed is the driving factor re-created airplane and actual airplane may have to be attributed to
behind this difference. Second, it has previously not been completely variability in the population samples and perhaps lack of power. Over-
ruled out that differences in ratings in different testing contexts are at all, these various explanations could be underlying reasons for the
least in part caused by actual differences in the populations from which differences observed in the current study in hedonic product discrimi-
the consumer sample is drawn (de Graaf et al., 2005). Lab tests, for nation between the different consumption contexts. The observed JAR
example, might draw on more experienced assessors who have actively ratings did not differ between the re-created and the actual airplane
enrolled in a database or responded to an advertisement, whereas sam- context. This finding is somewhat surprising. Higher taste thresholds
pling in field contexts often occurs on the spot and might draw on and reduced odor and taste intensity perceptions have previously been
assessors with a lesser general food involvement. reported under lower pressure conditions (Burdack-Freitag et al.,
Third, Boutrolle et al. (2007) assumed that more naturalistic eval- 2011). However, it has clearly not been possible to mimic the physio-
uation conditions might favorably improve the mood of the partici- logical factors that are thought to be responsible for raising taste
pants and that therefore their overall liking ratings might increase. In thresholds—such as reduced atmospheric pressure, low humidity, and/
the current study, the increase in hedonic ratings was already or magnetic-field exposure—in the re-created airplane context. One
observed in the re-created airplane context, implying that this semi- possible explanation could be attributable to methodological differen-
real-life context must already have been beneficial for the general ces. In Burdack-Freitag’s (2011) study, assessors rated four similar ver-
mood of the study participants. In line with this argument, during the sions of a menu semi-monadically; this is in itself a somewhat unnatural
debriefing, several participants spontaneously mentioned a “holiday task for consumers that in turn might lead to an overestimation of the
feeling” (personal communication, Milou Vrijhof). Finally, social inter- hedonic implications of perceptual differences between foods. In con-
action is known to play an important role in the consumption of foods trast, in the current study—in line with actual consumer behavior during
(Berry, Beatty, & Klesges, 1985; de Castro et al., 1990). In the current a flight—each consumer assessed only one meal and was thus unable
laboratory test, social interaction was minimized, whereas, in the re- to focus on perceptual differences between several meals. Thus, a
created and the actual airplane context, social interaction took place, change in sensory perception due to lower pressure conditions might
and this social interaction might in turn have influenced the expressed not lead immediately to a preference for flavor-enhanced meals—at
overall likings. least not on short distance flights were physiological impacts such as
Consumption-context-dependent differences were also observed actual dehydration and consequently a craving for salty foods is limited.
in the hedonic discrimination between the two variants of the same As no differences in overall liking were observed between REC2 and
meal type. In both the re-created and the actual airplane context, the ACT in the current study, it seems possible that, previously, the impor-
two variants were hedonically discriminated, whereas they were not in tance of actual sensory perception in the mouth for human food liking
the laboratory context. Several possible explanations for this observa- on board an airplane has been overestimated. This observation war-
tion come to mind. It has been reported that the discrimination of rants further research, ideally with the same participants in the same
products may be strongly dependent on the chosen methodology airplane—both on the ground and in the air.
(Calvin & Sather, 1959; McDaniel & Sawyer, 1981) and that first The study has strengths as well as weaknesses. A strength is that
impressions do not predict preferences after repeated exposures the subjects could choose between a chicken curry rice dish (meal
HOLTHUYSEN ET AL. | 7 of 8

type A) and pasta bolognese dish (meal type B), hence in the current the fruitful discussions we had during the writing of this article; and,
study, individual choices are taken into account which adds to the last but not least, Jim Groot and Joost Snels for their collaboration
naturalistic nature of the experiment. In the design of the experiment in this project. The DLO/TNO 2013 Topsector Agri&Food, the
it was decided that the same subjects who participated in the REC Dutch Ministry of Economic Affairs, Agriculture, and Innovation
test, first participated in the LAB test and that not counterbalancing funded the research.
was done between these two tests among the participants. On one All authors of this research article have directly participated in
hand, this avoided that participants that first conducted the test in the planning, execution, or analysis of this study. All authors of this
REC could transfer their contextual experience into the LAB test. On article have read and approved the final version submitted. The con-
the other hand, it cannot completely ruled out that this might have tents of this manuscript have not been copyrighted or published
systematically biased the results of the LAB test due to the occur- previously. There are no directly related manuscripts or abstracts,
rence of first-order effects. This demonstrates a clear dilemma inher- published or unpublished, by any authors of this article.
ent to context research and should be looked at systematically from
a methodological point of view in future research to define a stand- RE FE RE NCE S
ard “way of working” in context testing. The absence of an additional Armstrong, R. A. (2014). When to use the Bonferroni correction.
LAB measure after consumption of a complete meal can be seen by Ophthalmic and Physiological Optics, 34, 502–508.
some as a missed opportunity in the current study, as it would have Bangcuyo, R. G., Smith, K. J., Zumach, J. L., Pierce, A. M., Guttman, G. A.,
enabled us to also assess potential validity enhancing effects of & Simons, C. T. (2015). The use of immersive technologies to
improve consumer testing: The role of ecological validity, context
repeated measures in a lab setting (Kremer, Shimojo, Holthuysen,
and engagement in evaluating coffee. Food Quality and Preference,
€ ster, & Mojet, 2013). From a statistical point of view, a within-
Ko 41, 84–95.
persons design that extended to the actual airplane context would Berry, L. S., Beatty, W. W., & Klesges, R. C. (1985). Sensory and social
have been even more powerful. However, for practical reasons, this influences on ice cream consumption by males and females in a labo-
was not possible. Unfortunately, on the return flight, a school group ratory setting. Appetite, 6, 41–45.

with children under the age of 18 was on board, and these were not Boutrolle, I., Arranz, D., Rogeaux, M., & Delarue, J. (2005). Comparing
central location test and home use test results: Application of a new
eligible to participate in the research for ethical reasons. Conse-
criterion. Food Quality and Preference, 16, 704–713.
quently, this reduced somewhat the observations for meal type B,
€ ster, E. P. (2007).
Boutrolle, I., Delarue, J., Arranz, D., Rogeaux, M., & Ko
variant 2. Still, it was possible to collect in all three studies enough Central location test vs. home use test: Contrasting results depending
observations with an adequate sample size. Between the two con- on product type. Food Quality and Preference, 18, 490–499.
sumer groups (group 1: LAB and REC; group 2: ACT), some statisti- Burdack-Freitag, A., Bullinger, D., Mayer, F., & Breuer, K. (2011). Odour
cally significant differences existed (i.e., gender distribution, average and taste perception at normal and low atmospheric pressure in a
€r Verbraucherschutz und Lebensmit-
simulated aircraft cabin. Journal fu
age, and travel frequency). Therefore, it cannot be completely ruled
telsicherheit, 6, 95–109.
out that these differences in general characteristics of the partici-
Butler, G. C., Nicholas, J., Lackland, D. T., & Friedberg, W. (2000). Per-
pants might have had some impact on the results of the current
spectives of those impacted: Airline pilot’s perspective. Health
study. However, as in the current study fairly similar results were Physics, 79, 602–607.
observed for both participant groups, one could also question the Calvin, L. D., & Sather, L. (1959). A comparison of student preference
ecological validity of the observed differences in general characteris- panels with a household consumer panel. Food Technology, 13, 469–
tics between the two participant groups. 472.
Carvalho, F. R., Van Ee, R., Rychtarikova, M., Touhafi, A., Steenhaut, K.,
Persoone, D., Spence, C., & Leman, M. (2015). Does music influence
5 | CONCLUSION the multisensory tasting experience? Journal of Sensory Studies, 30,
404–412.
In summary, the re-created airplane context produced liking results Daillant-Spinnler, B., & Issanchou, S. (1995). Influence of label and loca-
more similar to the actual airplane context than the traditional labora- tion of testing on acceptability of cream cheese varying in fat con-
tory context did. Supra-threshold differences in sensory perception tent. Appetite, 24, 101–106.

under normal and lower pressure conditions and their relevance for de Castro, J. M., Brewer, E. M., Elmore, D. E., & Orozco, S. (1990). Social
facilitation of the spontaneous meal patterns of humans is independ-
actual food/meal liking on board an airplane might have previously
ent of time, place, alcohol or snacks. Appetite, 15, 89–101.
been overestimated. Consequently, sensory consumer testing in re-
de Castro, J. M., & de Castro, E. S. (1989). Spontaneous meal patterns in
created airplane contexts may offer a cost-efficient alternative to humans: Influence of the presence of other people. American Journal
extensive sensory consumer testing in real-life contexts. of Clinical Nutrition, 50, 237–247.
de Graaf, C., Cardello, A. V., Kramer, F. M., Lesher, L. L., Meiselman, H.
ACKNOWLEDG MENTS L., & Schutz, H. G. (2005). A comparison between liking ratings
obtained under laboratory and field conditions: The role of choice.
The authors would like to thank the Rijn IJsel vocational college in Appetite, 44, 15–22.
Wageningen for giving us the opportunity to conduct the study in Desmet, P. M. A., & Schifferstein, H. N. J. (2008). Sources of positive
their training facility. We would also like to thank Louise den Uijl for and negative emotions in food experience. Appetite, 50, 290–301.
8 of 8 | HOLTHUYSEN ET AL.

Edwards, J. S. A., Meiselman, H. L., Edwards, A., & Lesher, L. (2003). The Mcdaniel, M., & Sawyer, F. (1981). Preference testing of whiskey sour
influence of eating location on the acceptability of identically pre- formulations: Magnitude estimation versus the 9-point hedonic. Jour-
pared foods. Food Quality and Preference, 14, 647–652. nal of Food Science, 46, 182–185.
Hartwell, H. J., Edwards, J. S., & Brown, L. (2013). The relationship Meiselman, H. L. (1992). Obstacles to studying real people eating real
between emotions and food consumption (macronutrient) in a food- meals in real situations. Appetite, 19, 84–86.
service college setting – A preliminary study. International Journal of Meiselman, H. L. (2013). The future in sensory/consumer research:
Food Sciences and Nutrition, 64, 261–268. Evolving to a better science. Food Quality and Preference, 27, 208–
Hein, K. A., Hamid, N., Jaeger, S. R., & Delahunty, C. M. (2010). Applica- 214.
tion of a written scenario to evoke a consumption context in a labo- Meiselman, H. L., Johnson, J. L., Reeve, W., & Crouch, J. E. (2000). Dem-
ratory setting: Effects on hedonic ratings. Food Quality and onstrations of the influence of the eating environment on food
Preference, 21, 410–416. acceptance. Appetite, 35, 231–237.
Hein, K. A., Hamid, N., Jaeger, S. R., & Delahunty, C. M. (2012). Effects Petit, C., & Sieffermann, J. M. (2007). Testing consumer preferences for
of evoked consumption contexts on hedonic ratings: A case study iced-coffee: Does the drinking environment have any influence?.
with two fruit beverages. Food Quality and Preference, 10, 35–44. Food Quality and Preference, 18, 161–172.
Hersleth, M., Mevik, B.-H., Næs, T., & Guinard, J.-X. (2003). Effect of Piqueras-Fiszman, B., & Jaeger, S. R. (2014a). The impact of evoked con-
contextual factors on liking for wine – Use of robust design method- sumption contexts and appropriateness on emotion responses. Food
ology. Food Quality and Preference, 14, 615–622. Quality and Preference, 32, 277–288.
Hersleth, M., Ueland, O., Allain, H., & Næs, T. (2005). Consumer accep- Piqueras-Fiszman, B., & Jaeger, S. R. (2014b). Emotion responses under
tance of cheese, influence of different testing conditions. Food Qual- evoked consumption contexts: A focus on the consumers’ frequency
ity and Preference, 16, 103–110. of product consumption and the stability of responses. Food Quality
Jaeger, S. R., & Rose, J. M. (2008). Stated choice experimentation, con- and Preference, 35, 24–31.
textual influences and food choice: A case study. Food Quality and Piqueras-Fiszman, B., & Jaeger, S. R. (2014c). The impact of the means
Preference, 19, 539–564. of context evocation on consumers’ emotion associations towards
King, S. C., Meiselman, H. L., Hottenstein, A. W., Work, T. M., & Cronk, eating occasions. Food Quality and Preference, 37, 61–70.
V. (2007). The effects of contextual variables on food acceptability: Sester, C., Deroy, O., Sutan, A., & Galia, F. (2013). “Having a drink in a
A confirmatory study. Food Quality and Preference, 18, 58–65. bar”: An immersive approach to explore the effects of context on
King, S. C., Weber, A. J., Meiselman, H. L., & Lv, N. (2004). The effect of drink choice. Food Quality and Preference, 28, 23–31.
meal situation, social interaction, physical environment and choice on Spence, C. (2012). Auditory contributions to flavour perception and
food acceptability. Food Quality and Preference, 15, 645–653. feeding behaviour. Physiology & Behavior, 107, 505–515.
Kinnear, M., & de Kock, R. (2011). Would repeated consumption of Spence, C., Michel, C., & Smith, B. (2014). Airplane noise and the taste
sports drinks with different acidulants lead to hedonic adjustment?. of umami. Flavour, 3, 2.
Food Quality and Preference, 22, 340–345. Stroebele, N. M. A., & de Castro, J. M. (2004). Effect of ambience on
€ster, E. P. (2009). Diversity in the determinants of food choice: A psy-
Ko food intake and food choice. Nutrition, 20, 821–838.
chological perspective. Food Quality and Preference, 20, 70–82. Woods, A. T., Poliakoff, E., Lloyd, D. M., Kuenzel, J., Hodson, R., Gonda,
€ster, E. P., & Mojet, J. (2007). Boredom and the reasons why
Ko H., & Thomas, A. (2011). Effect of background noise on food percep-
some new products fail. In H. J. H. MacFie (Ed.), Consumer-led food tion. Food Quality and Preference, 22, 42–47.
product development (pp. 262–280). Cambridge, UK: Woodhead
Publishing.
€ster, E. P., & Mojet, J.
Kremer, S., Shimojo, R., Holthuysen, N. T. E., Ko How to cite this article: Holthuysen NTE, Vrijhof MN, de Wijk
(2013). Consumer acceptance of salt-reduced “soy sauce” foods over
RA, Kremer S. “Welcome on board”: Overall liking and just-
rapidly repeated exposure. Food Quality and Preference, 27, 179–190.
about-right ratings of airplane meals in three different consump-
nez, A., Vidal, L., & Ares, G. (2014). Influence of context
Machín, L., Gime
tion contexts—laboratory, re-created airplane, and actual air-
on motives underlying food choice. Journal of Sensory Studies, 29,
313–324. plane. J Sens Stud. 2017;00:e12254. https://doi.org/10.1111/

Maga, J. A., & Lorenz, K. (1972). Effect of altitude on taste thresholds. joss.12254
Perceptual and Motor Skills, 34, 667–670.

You might also like