Food Quality and Preference 21 (2010) 410–416

Contents lists available at ScienceDirect

Food Quality and Preference

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

Application of a written scenario to evoke a consumption context in alaboratory setting: Effects on hedonic ratings

Karen A. Hein a,*, Nazimah Hamid a, Sara R. Jaeger b, Conor M. Delahunty c

a Department of Food Science, University of Otago, P.O. Box 56, Dunedin 9054, New Zealandb Plant and Food Research, Private Bag 92169, Auckland Mail Centre, Auckland 1142, New Zealandc CSIRO Division of Food and Nutritional Sciences, P.O. Box 52, North Ryde, NSW 1670, Australia

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

Article history:Received 6 March 2009Received in revised form 14 October 2009Accepted 14 October 2009Available online 20 October 2009

Keywords:ContextConsumerHedonic ratingDiscriminationEvokedLaboratorySensory

0950-3293/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.foodqual.2009.10.003

* Corresponding author. Fax: +64 3 479 7567.E-mail address: karen.hein@stonebow.otago.ac.nz

Consumer hedonic testing of products is typically conducted under controlled sensory laboratory condi-tions. This setting does not accurately represent how food and drink are consumed. Literature demon-strates that consumer hedonic ratings elicited in the natural consumption context differ to thoseelicited under controlled conditions. This suggests that when removing a product from its natural con-sumption context, accurate hedonic ratings may not be obtained. The interest of this research was todevelop an approach that evokes a consumption context in the sensory laboratory and study its impacton hedonic ratings. A written scenario was developed that was effective at making participants imaginean occasion when they desired a refreshing beverage. Consumer hedonic ratings of four apple juice sam-ples elicited using the evoked consumption context (context condition) were compared to those elicitedin a control condition (i.e., no evoked context). Differences in mean hedonic ratings of the samples wereobserved between the two conditions with greater sample discrimination observed for the evoked con-text condition. Consumers using the evoked context found it easy to indicate their product liking/dislik-ing, and felt that the liking information they provided was accurate, more so than consumers in thecontrol setting. Sensory practitioners need to be aware of the potential use of an evoked context in a con-trol setting for eliciting product hedonic ratings, and understand its impact on mean hedonic ratings.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. Why seek to evoke a consumption context in a controlledlaboratory setting?

Consumer sensory testing is carried out to understand con-sumers’ hedonic appreciation toward products. While traditionalconsumer sensory testing takes place in a controlled sensory lab-oratory setting (Lawless & Heymann, 1998; Meilgaard, Civille, &Carr, 1999; Stone & Sidel, 2004), application of these testing pro-tocols has been criticized for not taking into account the products’consumption context. Consumer sensory testing performed in theabsence of a context (e.g., control setting) has been described as a‘situational fallacy’ (Köster, 2003). Though context is often re-ferred to as a location/place with physical characteristics, otherassociations (e.g., feelings, events, activities, etc.) regarding whenthe product is consumed may define a product and may also de-fine its consumption context (Lyman, 1989). In this way, different

ll rights reserved.

(K.A. Hein).

types of context may be explicitly defined. While sensory qualitiesof food may allow for recognition and identification, mental con-texts (associations) derived from the product, give the productmeaning (Lyman, 1989). This would suggest that by removingthe product from the context when it is normally consumed, aconsumer may have less involvement with the product and accu-rate hedonic ratings may not be obtained. While a body of recentliterature has demonstrated that differences in consumer hedonicratings are observed when products are evaluated in natural usetype conditions, compared to a control-type setting (e.g., Bou-trolle, Delarue, Arranz, Rogeaux, & Köster, 2007; Kozlowskaet al., 2003; McEwan, 1997), product testing under natural con-sumption contexts is expensive and time consuming. In order tomore accurately elicit hedonic ratings, an attempt should be madeto make consumer testing protocols more realistic in terms ofhow the products are consumed. To achieve this, it is necessaryto explore whether a consumption context can be successfullyevoked under a controlled laboratory setting and to understandhow hedonic ratings are affected. We consider issues associatedwith: (i) developing and applying an evoked consumption context,and (ii) the influences of contextual information on hedonicratings.

K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416 411

1.2. Developing and applying a method to evoke a consumption context

Different approaches have been taken to evoke consumptioncontexts. Some studies have used physical means (Bell & Meisel-man, 1994; Hersleth, Mevik, Naes, & Guinard, 2003; Petit & Sieffer-mann, 2007). Physical aspects have been said to provide limitedmeaning to a consumption context and may be a poor approachto evoke a consumption context (Köster, 2009). A consumptioncontext is more than only physical aspects (e.g., décor). Associa-tions such as emotions, people and weather/time are among thosemore commonly related to consuming food items (Lyman, 1989).An approach taken by Petit and Sieffermann (2007) used visual,olfactory and auditory cues to induce a context. Authors reportedthat the physical means (i.e., photo, curtains, lights) may have ‘sur-prised’ consumers. Written scenarios have successfully been usedto evoke or create different contexts (Belk, 1975; Hansen, 2005;Jaeger & Meiselman, 2004). Written scenarios are statements orbrief texts that describe a particular situation meant to evoke asense of presence in a real situation. Unlike the use of physicalmeans, written scenarios do not require major modifications tothe physical environment and depending on how the text is writ-ten, allow consumers to personalize the context being evoked.

As recommended by Petit and Sieffermann (2007), ‘pre-valida-tion’ of the context being evoked is necessary to ensure the con-texts’ effectiveness. Without accurate portrayal of the contextand ability of the consumer to become engaged with the simula-tion, realism is minimized (Runkel & McGrath, 1972). Some studieshave taken an additional step towards ensuring context effective-ness by having consumers provide a written response towardsthe written scenario used to evoke a context. A study looking atthe influence of mood on product evaluation was carried out byQiu and Yeung (2008). Consumers were asked to think of a recentevent that made them feel either happy or unhappy and were thenasked to provide a written description of that time. Providing awritten response, has been said to increase the availability of suchevents that exist in the memory (Schwarz & Clore, 1983). The writ-ten responses can also provide qualitative data regarding aspects ofindividual consumer’s contexts that would allow further under-standing on how the context was created.

1.3. Influences contextual information may exert on hedonic ratings

An evoked consumption context may influence hedonic ratingsrelative to when no context is evoked (i.e., when hedonic ratingsare elicited in controlled laboratory setting). Differences in meanhedonic ratings observed under various testing conditions can begeneralized into three categories of effects: span, level, and order.A level effect would be observed if samples were rated equallymore/less in different contexts (Boutrolle, Arranz, Rogeaux, & Del-arue, 2005; Kozlowska et al., 2003). This would be observed whentwo samples are rated 5 and 6 in one condition, and 7 and 8,respectively, in another. A span effect would be observed if sam-ples are liked/disliked similarly, but by a greater relative magni-tude. This is demonstrated if two samples are rated 5 and 6 inone condition, and are rated 4 and 7 in another condition. As ob-served by McEwan (1997), the order effect may be observed if sam-ples are liked differently. This could take the form of samples beingrated in the same order in both the control and context conditions,but different in their relative magnitudes; or if samples are liked ina different order. For example, sample A is liked more than B in onecondition, while B is more liked than A in another condition.

1.4. Aims of study

The first aim of this study was to explore if a consumption con-text could be effectively evoked in a controlled laboratory setting

using a written scenario. The second aim of this study was tounderstand how hedonic ratings are affected when elicited usingan evoked context in a laboratory setting. Specifically the con-trolled laboratory setting and evoked context in the laboratory set-ting were compared for differences in aggregate sample hedonicratings, sample discrimination and consumers perceptions ofhow they performed product evaluations.

2. Materials and methods

In this study it was essential to establish that the consumptioncontext being evoked matched the product type under evaluation.Knowing that consumers would really consume the product in thecontext being evoked was critical in order to obtain relevant hedo-nic ratings. As indicated by Boutrolle and colleagues, ‘‘the usualcontext of consumption has a leading role in the hedonic evalua-tion of food products” (Boutrolle et al., 2007). Consumers wouldnot be able to relate to the consumption context if it is not typicalof when the product would be normally consumed, making theevoked context ineffective. Following extensive pilot work, applejuice was determined to be an appropriate beverage for the‘refreshing’ context; consumers indicated they would consume ap-ple juice when desiring a refreshing beverage. Other products per-ceived as ‘refreshing,’ could have been used in this study (e.g.,cola). However it is important to point out that while it was criticalto establish that the product and context matched, the specificcontext and product was not the focus of this study. Rather thisstudy was a methodological investigation to explore if a contextcan be evoked and to study its effect on hedonic ratings.

2.1. Samples

Four apple juice samples were created to vary subtly in two sen-sory dimensions, such that samples would be similar in mean he-donic ratings but different in sensory character when presented ina controlled laboratory setting. Establishing this would mean thatany difference in mean hedonic ratings observed under an evokedrefreshing context would be attributed to a difference in the per-ception of underlying sensory character. The present study testedthe hypothesis that an interaction between context and hedonicratings will be observed when sensory properties are varied subtlywithin product type. An alternative design might test the effect ofcontext on products that vary to a relatively large extent in sensoryproperties, either within product type or across different products.For example, the effect of an evoked context on the hedonic ratingsfor a set of different products (e.g., apple juice, orange juice, cola)could be investigated. For both study designs, interactions betweenthe effects of context and product differences can be estimated.

Using a base apple juice (Fresh-Up Crisp Apple Juice, FrucorBeverages Ltd., Auckland, New Zealand), four samples were createdusing a 2 � 2 design consisting of citric acid and strawberry es-sence (Hansells Natural Strawberry Essence, Old Fashioned FoodsLtd., Auckland, New Zealand). These samples will be referred toas A (0.12 g/100 ml citric acid, 0 ml strawberry essence), B(0.12 g/100 ml citric acid, 0.05 ml/100 ml strawberry essence), C(0 g citric acid, 0 ml strawberry essence) and D (0 g citric acid,0.05 ml/100 ml strawberry essence). A pilot test carried out withconsumers (n = 21) established that the four apple juice sampleswere not different in hedonic ratings when evaluated in a con-trolled laboratory setting. However were perceived as having sub-tle differences. Acidity was modified as previous studies havedemonstrated its influence on perceived beverage refreshment(Labbe, Gilbert, Antille, & Martin, 2009; McEwan & Colwill,1995). Strawberry essence was added to modify the flavor profile

412 K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416

of the beverage. The added taste and flavor were such that all sam-ples could be categorized as the product type, apple juice.

Samples were prepared the day before evaluation and storedunder refrigeration (�4 �C). Prior to evaluation, 30 ml of sampleswere poured into plastic portion cups with lids. Small sampleswere used, as is typical in central location tests. Samples werecoded with three digit random numbers and randomized to ac-count for sample order and carry-over effects (MacFie, Bratchell,Greenhoff, & Vallis, 1989). The randomized sets of four sampleswere placed in an insulated container with crushed ice, and storedrefrigerated until evaluation.

2.2. Participants

Consumers involved in the pilot tests for the selection of applejuice samples, development of the context, and the hedonic ratingtasks in the main study met the same selection criteria. Consumerswere between 18 and 65 years of age, lived in New Zealand for aminimum of 5 years and consumed apple juice or a juice beverageat least once during the month prior to their participation. All con-sumers were given a movie voucher as a token of appreciation fortheir participation. Ethical approval to perform this study wasgranted by the University of Otago Human Ethics Committee.

Consumers were assigned to the control setting condition(n = 72: 26% males) and the evoked context condition (n = 70:30% males). Conditions were created to have consumers with sim-ilar demographics with respect to age (v2

4 ¼ 2:171, p = 0.704) andgender (v2

1 ¼ 0:229, p = 0.632). By demonstrating homogeneity inkey demographics, subsequent differences between conditionscan be inferred to be a result of the experimental conditions.

2.3. Development of written scenario to evoked consumption context

A preliminary inquiry was carried out to determine beveragesthat are perceived as refreshing and to establish associations thatconsumers have with wanting a refreshing drink. This involvedadministering a questionnaire to consumers (n = 19). The question-naire asked consumers what beverages they considered ‘refresh-ing’ and to describe in words the last time they had a refreshingbeverage. Consumer responses indicated that the apple juice bev-erages were perceived as refreshing, confirming this as the productfor the study. The descriptions provided by consumers aided in thewording and writing of the written scenario. A second pilot testwas then carried out where consumers (n = 17) were presentedwith the written scenario and asked to provide a written responsedescribing the occasion that they were imagining. A questionnairewas also administered to gauge the effectiveness of simulating thecontext from the consumers’ perspective and to verify the appro-priateness of consuming apple juice in this context. The results ob-tained established that (1) the context evoked by the writtenscenario was an occasion when a refreshing beverage was desired,(2) the context evoked by the written scenario was an occasionappropriate for consuming apple juice and, (3) consumers wereable to understand and carryout the tasks requested of them. Addi-tionally, the pilot test determined that it was important to empha-size the ‘‘occasion,” so as not to confuse the consumer by havingthem consider the ‘‘type” of refreshing drink. The written textwas consequently modified.

The following written scenario was provided to consumers toevoke the consumption context, when a refreshing drink wasdesired.

Think about an occasion when you want something refreshing todrink. Clearly imagine you are experiencing this occasion. Now, writedown a detailed description of the occasion you are imagining. Pleasetake your time and provide a description that is as complete aspossible.

2.4. Procedure

2.4.1. Control conditionConsumers were asked to attend one 60 min evaluation session.

Consumers were provided with the four juice samples and were in-structed to taste each sample and indicate their overall liking on 9-point hedonic labeled category scales labeled from ‘like extremely’to ‘disliked extremely,’ with all standard incremental labels (Pery-man & Pilgrim, 1957). Consumers were instructed to taste the sam-ples by consuming as much or as little as desired. They were theninstructed to rinse their palate between samples with water andtake a 1 min break before continuing with subsequent samples.After sample evaluation, consumers completed a questionnaireregarding how they felt about their evaluations. Consumers repliedto two questions on 9-point category scales: ‘how easy/difficult didyou find it to rate your liking/dislike of the apple juice samples’(1 = ‘very difficult,’ 9 = ‘very easy’), and ‘to what extent do you feelthat the liking information you have given is accurate’ (1 = ‘not atall accurate’, 9 = ‘very accurate’). These questions were also askedin the evoked context condition. A comparison of responses fromthe two conditions provided insight into the consumers’ perspec-tive as to whether evoking a context impacted on the demandsof the task. An open-ended question was included asking consum-ers what they thought the study was about. This question was in-cluded to ensure that consumers were unaware that the study wasintended to compare hedonic ratings elicited in a controlled settingto that elicited when an evoked context was used.

2.4.2. Evoked context conditionIn the evoked context condition, the same evaluation procedure

as the controlled laboratory condition was followed with the fol-lowing exceptions. Before sample evaluation instructions were gi-ven, consumers were presented with the written scenario (seeSection 2.3). The scenario was projected on a screen and read aloudtwice before consumers were permitted to provide a written re-sponse describing the occasion they were imagining. Once all con-sumers present in the session had written their response,instructions for product evaluation were given. Consumers wereinstructed to keep in mind the occasion they had described,throughout evaluation and were allowed to reread their descrip-tion at anytime. The written scenario was provided at the top ofeach page where consumers indicated their hedonic ratings towardthe samples. Two further questions were included in the question-naire to serve as a manipulation check that the context ‘when arefreshing drink is desired,’ was effectively evoked. The questionswere rated on 9-point category scales and read: ‘to what extentdid the occasion you imagined compel you to desire a refreshingdrink?’ (1 = ‘not at all,’ 9 = ‘very’), and ‘while you tasted the applejuice samples, how vivid in your mind was the occasion you imag-ined?’ (1 = ‘not at all vivid,’ 9 = ‘very vivid’).

Consumer testing was carried out in individual booths in theSensory Science Laboratory, at the Department of Food Science,University of Otago. All sessions were held at 12:00 pm. Testingof each condition occurred once within each testing week.

2.5. Statistical analysis

Hedonic ratings collected in the controlled laboratory settingwhen the context was evoked were compared in terms of: (i)aggregate sample hedonic ratings to investigate level and order ef-fects, (ii) discrimination of hedonic ratings through the investiga-tion of pair-wise sample comparisons, and (iii) how consumersfelt when they completed product evaluations.

Mean hedonic ratings for each of the four samples was tabu-lated for the two conditions. Sample significance (a = 10%) wastested by two-way (sample and judge effects) analysis of variance

3.0

4.0

5.0

6.0

7.0

8.0

9.0

Mea

n A

ccep

tanc

e

Evoked Context (n=70)*

Control Setting (n=72)

a

bab

ab

* samples sharing the same letter are not different at p<0.1

1.0

2.0

A B C DSample

*

Fig. 1. Mean hedonic ratings and standard errors for four apple juice samplesevaluated using the evoked context and evaluated without the evoked context inthe controlled laboratory setting.

K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416 413

without interaction for hedonic rating data. Separate analysis wasperformed on data elicited in the control condition and in theevoked context condition. This was to determine whether therewas an overall difference in the hedonic ratings of samples whenthe context was evoked, when compared to the control condition.Tukey’s post hoc testing was carried out on significant sample ef-fects (a = 10%) to identify samples that were statistically differentfrom one another.

Paired samples t-tests were performed to compare the sixpossible sample pairs within each condition. Resulting t-valueswere compared between the two conditions, as the relative sizeof t is an indicator of the magnitude of sample discrimination. Alarger value of t would suggest a greater magnitude of discrimina-tion between conditions. Authors do acknowledge that the samplesize between context and control settings differed by two consum-ers. Comparison of t-values was made as this difference wasnegligible.

Consumers’ perception of their ability to complete product eval-uations was measured in the questionnaire. Comparison of the twotest conditions for elicited hedonic ratings were performed usingan independent samples t-test (a = 10%).

3. Results

3.1. Manipulation check to ensure that context was effectively evoked

In order to compare between hedonic ratings elicited when awritten scenario was used to evoke the context ‘when a refreshingdrink is desired,’ and a control setting, it was imperative to verifythat the context was effectively evoked. Two manipulation checkswere performed.

The first manipulation check was performed as part of the ques-tionnaire administered to consumers following product evaluation.Consumers in the evoked context condition reported that theywere compelled to desire a refreshing drink (�x ¼ 7:8� 1:8,n = 70) and that the occasion they described was vivid in theirmind while evaluating samples (�x ¼ 8:1� 1:1, n = 70). As re-sponses were made where the maximum scale response was equalto 9, these data suggest that the written scenario successfullyevoked a refreshing context.

The second manipulation check was performed using the writ-ten descriptions provided by the consumers in response to thewritten scenario. Qualitative analysis was carried out on thedescriptions to establish that consumers were describing an occa-sion when they desired a refreshing drink as they would in real life.Analysis of the written descriptions showed that all consumersconveyed in words, an occasion when they desired a refreshingdrink. Representative verbatim examples of these descriptionsare shown in Table 1. These descriptions are typical responses fromconsumers. Consumer responses described occasions of physicalactivity or when present in a hot conditions (83%), and occasions

Table 1Verbatim examples of consumer responses to the written scenario used to evoke the cont

Consumer Response

A ‘On a clear, hot, wind free day. Sun is beating down. I have done all thtools away. Sitting on the deck relaxing and admiring my handy work

B ‘I have just been mowing the lawns on a hot summer’s day. It has beenperfect.’

C ‘A sunny summer day, late afternoon a gathering of friends, conversaD ‘Afternoon tea at my Nanas house, with all my cousins and we’re sitting

with my nana and cousins. It’s been a very hot day.’E It’s a hot, humid summer afternoon and I have just biked home from a

quench my thirst, cool me down, and give me some post-exercise ene

of relaxation (17%). Results from the manipulation checks substan-tiated that the written scenario effectively evoked the contextwhen a refreshing drink was desired.

Response to the question ‘what do you think this study wasabout?’ indicated that consumers were unaware that the studysought to determine the influence of context on their hedonic rat-ings. Twenty-four percent of consumers in the control conditionresponded that specific sensory attributes (i.e., sweeteners, differ-ent additives, levels of added sugar, sweet–sour balance, and flavorstrength) were being investigated in the study. On the other hand,no consumer in the context condition indicated a specific sensoryattribute was being investigated. Consumer data from both condi-tions were used for analysis.

3.2. Comparison between hedonic ratings of four apple juice sampleselicited in a control setting and when using an evoked context

3.2.1. Aggregate sample hedonic ratingsA significant difference in hedonic ratings of the four apple juice

samples was observed when elicited under the evoked context(F3,207 = 2.19, p = 0.091) but not in the control setting(F3,213 = 0.92, p = 0.830). When evaluated under the evoked con-text, sample C was the most liked, and was significantly differentfrom sample B (Fig. 1). Under the evoked context, the two samplescontaining citric acid (A and B) were the least liked (�x ¼ 6:6 and�x ¼ 6:5, respectively, n = 70). Samples C (�x ¼ 7:1) and D (�x ¼ 6:8)were the most liked. The addition of citric acid decreased samplehedonic ratings. Although sample C is most liked in the evokedcontext condition overall mean hedonic ratings were not consis-tently higher compared to the control condition.

extual occasion, when a refreshing drink is desired.

e weeding, mowing of lawns, vegetable garden and have just cleaned and put all.’hard work and taken about an hour and something cold and refreshing would be

tion, music, smell of food being barbequed in the air.’outside in the sun enjoying a refreshing drink after just playing a game of cricket

game of tennis at the local courts. I’m thirsty and hot, and want a cold drink torgy boosting.’

Fig. 2. t-Values comparing hedonic ratings for all possible pairs of four apple juicesamples evaluated using the evoked context and evaluated without the evokedcontext in the controlled laboratory setting.

414 K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416

3.2.2. Pair-wise sample discriminationOverall, a greater discrimination of individual sample pairs for

hedonic ratings was observed when elicited under the evoked con-text compared to the control condition (Fig. 2). This is demon-strated by greater magnitudes of t in the evoked contextcondition for four of the six possible sample pairs (A–C, B–C, B–D, and C–D).

In the evoked context condition, the comparison of two samplepairs resulted in significant t-values, while no sample pairs weresignificantly discriminated in the control setting condition(p > 0.1). In the context condition, sample C was significantly moreliked than both of the samples that contained citric acid (p 6 0.10).Results also suggest that sample C was more liked than sample Dwhen evaluated using an evoked context (t69 = 1.5, p = 0.13). Theseresults demonstrate improved discrimination for hedonic ratingswhen presented under an evoked context condition compared toa control setting.

A tendency for greater sample discrimination of individual sam-ple pairs was observed when evaluated using the evoked contextcompared to the control condition (Fig. 2). While significancewas not attained between samples, sample D was more liked thansamples B or A sample. This was seen as a negative t-value for boththe B–D and A–D sample pairs. In the control setting however,sample D was found to be less liked than either sample A or B.

3.3. Comparison of consumers perception on how they performedduring sample evaluation when elicited in a control setting and whenusing an evoked context

A comparison of the two evaluation conditions based on con-sumers’ perception of how they performed during sample evalua-tion was achieved using the questionnaire. It was important toensure that evoking a context in a controlled laboratory settingdid not make consumers feel that their task was any more difficultthan when no context was evoked. Consumers under the contextcondition felt it was easier to rate their liking/dislike of the applejuices sample (�x ¼ 6:7, SE = 0.2) than did consumers in the controlsetting (�x ¼ 5:7, SE = 0.3), (t140 = 2.8, p 6 0.01). Similarly, consum-ers in the context condition felt that the liking information theyprovided was more accurate (�x ¼ 7:0, SE = 0.2) than consumers inthe control setting (�x ¼ 5:7, SE = 0.1), (t140, p 6 0.05).

4. Discussion

By not accounting for context in the laboratory, consumer hedo-nic ratings may not be accurately measured. In this study, a context

was evoked using a written scenario in a laboratory setting. Hedo-nic ratings elicited when using an evoked context were comparedto those elicited in a control condition. Findings suggest that theinterpretation of how samples differed in their hedonic apprecia-tion would be different when comparing between the two condi-tions. This was demonstrated by differences in aggregate hedonicratings, and greater sample discrimination obtained when elicitedusing an evoked context compared to a control condition. By evok-ing a context, a greater understanding of how samples differed intheir hedonic appreciation may be achieved, while enabling con-sumers to be confident in their evaluations. This discussion will fo-cus on how the context was evoked and its impact on hedonicratings, and give suggestion for further research.

4.1. Influence of contextual information on hedonic ratings

The evoked context may have allowed consumers to have asense of being present in the occasion when a refreshing drinkwas desired. Evoking the context may have triggered an increasein consumer concentration or attention. As indicated by responsesobtained from the questionnaire, consumers in the evoked contextfound product evaluation easier and felt that their responses weremore accurate than individuals in the control condition. These re-sults tentatively suggest that evaluation of samples using anevoked context changed the consumer’s state of mind in someway. Consumers may have been relaxed when reading of the sce-nario, and subsequently felt more confident in the product re-sponses that they provided. The written scenario may also havefocused the consumer’s attention.

Consumers in the control condition may have felt that their ‘job’was to look for differences in the product, and therefore took ananalytical approach. The open-ended question asking consumerswhat they thought the study was about, may provide insights asinto the consumer’s mindset. In the present study consumers inthe control condition more frequently stated that the current studycarried out was investigating specific sensory qualities of applejuice (e.g., sweeteners, sourness), than those in the context condi-tion. Consumers in the control condition may have developed animpression that the study was investigating specific sensory char-acter, even though no reference to specific sensory attributes wasmade during the study. More research is needed to understandwhether or not the consumer’s mindset in the evoked context wereaffected or whether differences in hedonic ratings were a result offocusing consumers’ attention.

The authenticity test paradigm evokes an affective mindset inconsumers by modifying the instructions used in the study. Studiesby Frandsen and colleagues found better discrimination of milkswith subtle flavor differences using an authenticity test rather thananalysis of samples using a trained descriptive panel (Frandsen,Dijksterhuis, Brockhoff, Nielsen, & Martens, 2003) or using a dis-crimination test (Frandsen, Dijksterhuis, Brockhoff, Nielsen, & Mar-tens, 2007). Authors suggested that this improved discriminationwas due to an affective response rather than an analytical re-sponse. As with the present study, these studies demonstratedhow context can be evoked and that in doing so, a consumer’s re-sponse can be modified.

Differences in product hedonic ratings are a response to under-lying sensory characters. In the present study, samples containingcitric acid were less liked than those without citric acid when eval-uated under the context condition. It is unknown whether specificsensory qualities such as ‘sour’ or ‘flavor intensity’ would havebeen more/less perceived in the evoked context condition com-pared to the controlled condition as a result of citric acid or straw-berry essence addition. Additionally, it would be important todetermine whether the ability to discriminate samples in termsof sensory characteristics and the direction of their discrimination

K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416 415

was influenced by the context. A study by Pound, Duizer, andMcDowell (2000) found that consumers were generally more crit-ical of sensory attributes in chocolate bars when tested in a labora-tory setting compared to other test locations (central location,home-use, teaching lab), while liking was not substantiallyimpacted. In this case, consumers may have had a more ‘analytical’approach to product evaluations when less natural consumptioncontexts were used. Further research is needed to understand theimpact of an evoked context on perception of underlying sensoryqualities in relation to hedonic ratings.

4.2. Considerations in developing and applying a written scenario toevoke context for use in consumer testing

The present study used a written scenario in a controlled labo-ratory setting to evoke a context and this resulted in differences ofproduct hedonic ratings when compared to the control condition.The written scenario allowed consumers to create and personalizethe context being evoked. Having consumers provide a written re-sponse of the occasion that they were imagining may have in-creased the level of consumer involvement and subsequently theavailability of different aspects of the context being evoked (Qiu& Yeung, 2008; Schwarz & Clore, 1983). Although no direct mea-sure was taken, consumer involvement was evident through thedescriptions provided by the consumers (see Section 3.1). Furtherunderstanding is needed on the impact of having consumers pro-vide a written response on context effectiveness.

Previous research using physical means to evoke context hasbeen performed. However no difference in hedonic ratings was re-ported when compared to the control condition (Petit & Sieffer-mann, 2007). In this study visual, olfactory and auditory cueswere used to create a warm context for the consumption of icedcoffee. The study may not have effectively evoked a realistic con-text due to the experimental setting and the novelty of producttype to the test population. Further research is needed to explorethe potential application of physical means to evoke a context,and to compare it to the use of a written scenario. The use of thewritten scenario in the present study effectively allowed consum-ers the freedom to specify what they defined as a ‘refreshing occa-sion.’ It would be interesting to know whether similar resultswould be obtained if consumers were directed to imagine a speci-fied occasion when a refreshing drink might be appropriate.

It is important to make clear that the results presented in thisstudy are obtained when a particular context is evoked for a partic-ular product set evaluated. It is uncertain whether similar differ-ences in hedonic ratings would be achieved if a context ofsimilar meaning such as ‘thirst-quenching’ had been evoked. It isalso unclear whether differences in results would be observed ifa very different context had been created. If a different contextwas evoked, it is possible that differences in hedonic ratings mayhave been obtained to those seen in the present study. During pre-liminary work for this study, consumers (n = 19) were asked ‘whatdoes refreshing mean to you?’ Eight consumer responses referredto some form of thirst quenching such as ‘quench thirst,’ ‘resolvemy thirstiness,’ or ‘relieve thirst.’ Hence it would seem reasonableto suggest that if a similar context such as ‘thirst-quenching’ wascreated, comparable results as shown in this study would beachieved. However, it is likely that a different occasions may havebeen imagined. Alternatively, the effects of context on productsthat vary to a relatively large extent in sensory properties, eitherwithin product type or across different product, could be studied.This may involve different product types being presented to ob-serve the impact of an evoked context. Further research is neededto fully understand whether and how the application of differentcontext impact consumer hedonic ratings.

More research is needed to provide conclusive recommendationof evoking a context in a controlled setting for use in consumertesting. Results from this study suggest that improved discrimina-tion for samples hedonic ratings may be achieved using an evokedcontext when sample differences are subtle. Furthermore consum-ers were found to be more confident with their results when a con-text was used than in the control condition. In order to understanddifferences in individual consumers sensitivity to evoked contexts,a study applying a within subjects design needs to be performed.The between groups design used in study does not permit observa-tion of how individual consumer preference patterns wereaffected.

For the purposes of investigating the effects of context andproduct differences, this study used effects that were subtle andproportionate. Through pilot test, the samples were created to besimilar in hedonic rating, while varying in sensory character. Thecontext was created to match the product type. Regardless ofwhether these effects of context and product differences are pro-portionate or not, the interaction between the two effects can beestimated from a statistical point of view. This is demonstratedin study by Boutrolle et al. (2007) who measured hedonic ratingsin home and a central location for three different types of products.A significant interaction between location and the samples was ob-served for one of the three product types when monadic sequentialdata were analyzed.

The present study was not intended to demonstrate that theseresults are indicative of what would happen in the ‘real’ consump-tion condition. No simulation will replace a products ‘real’ contextfor consumption. However in this study, the evoked context was anapproximation of a consumption context and did demonstrate sim-ilarities with results in literature that compared control and natu-ral consumption conditions. Hence it is important to understandthe influence of an evoked context on consumer hedonic ratingsto better measure product appreciation.

5. Conclusion

A written scenario that evoked a context within a controlledsetting affected mean hedonic ratings and sample discrimination.Further research is needed to validate the use of a written scenarioto evoke a context in a laboratory setting. It is also important tounderstand the effect of context wording on hedonic ratings andto determine whether these ratings are affected similarly acrossproducts that vary more widely in sensory characteristics, or acrossdifferent product types. Sensory practitioners need to be aware ofthe potential use of evoking a context in a controlled setting foreliciting hedonic ratings, and to understand that context may affecttheir sensory-based product decisions.

Acknowledgements

We would like to thank Christina Bava at Plant and Food Re-search, New Zealand, for advice and guidance with the qualitativeanalysis. Author K.A. Hein would like to acknowledge support fromRose Marie Pangborn Sensory Science Scholarship.

References

Belk, R. W. (1975). Situational variables and consumer behavior. Journal of ConsumerResearch, 2, 157–164.

Bell, R., & Meiselman, H. L. (1994). Effects of adding an Italian theme to a restauranton the perceived ethnicity, acceptability and selection of foods. Appetite, 22,11–24.

Boutrolle, I., Arranz, D., Rogeaux, M., & Delarue, J. (2005). Comparing centrallocation test and home use test results: Application of a new criterion. FoodQuality and Preference, 16, 704–713.

416 K.A. Hein et al. / Food Quality and Preference 21 (2010) 410–416

Boutrolle, I., Delarue, J., Arranz, D., Rogeaux, M., & Köster, E. (2007). Central locationtest vs. home use test: Contrasting results depending on product type. FoodQuality and Preference, 18, 490–499.

Frandsen, L. W., Dijksterhuis, G. B., Brockhoff, P., Nielsen, J. H., & Martens, M. (2003).Subtle differences in milk: Comparison of an analytical and an affective test.Food Quality and Preference, 14, 515–526.

Frandsen, L. W., Dijksterhuis, G. B., Brockhoff, P., Nielsen, J. H., & Martens, M. (2007).Feelings as a basis for discrimination: Comparison of a modified authenticitytest with the same-different test for slightly different types of milk. Food Qualityand Preference, 18, 97–105.

Hansen, T. (2005). Understanding consumer perception of food quality: The cases ofshrimps and cheese. British Food Journal, 107, 500–525.

Hersleth, M., Mevik, B. H., Naes, T., & Guinard, J. X. (2003). Effect of contextualfactors on liking for wine – Use of robust design methodology. Food Quality andPreference, 14, 615–622.

Jaeger, S. R., & Meiselman, H. L. (2004). Perceptions of meal convenience: The case ofat-home evening meals. Appetite, 42, 317–325.

Köster, E. P. (2003). The psychology of food choice: Some often encounteredfallacies. Food Quality and Preference, 14, 359–373.

Köster, E. P. (2009). Diversity in the determinants of food choice: A psychologicalperspective. Food Quality and Preference, 20, 70–82.

Kozlowska, K., Jeruszka, M., Matuszewska, I., Roszkowski, W., Barylko-Pikielna, N., &Brzozowska, A. (2003). Hedonic tests in different locations as predictors ofapple juice consumption at home in elderly and young subjects. Food Qualityand Preference, 14, 653–661.

Labbe, D., Gilbert, F., Antille, N., & Martin, N. (2009). Sensory determinants ofrefreshing. Food Quality and Preference, 20, 100–109.

Lawless, H. T., & Heymann, H. (1998). Acceptance and preference testing. InSensory evaluation of food, principles and practices (pp. 430–479). Gaithersburg:Aspen.

Lyman, B. (1989). Food meanings. In A psychology of food (pp. 125–138). New York:Van Nostrand Reinhold.

MacFie, H. J. H., Bratchell, N., Greenhoff, K., & Vallis, L. (1989). Designs to balance theeffect of order of presentation and first-order carry-over effects in hall tests.Journal of Sensory Studies, 4, 129–148.

McEwan, J. A. (1997). A comparative study of three product acceptability trials. FoodQuality and Preference, 8, 183–190.

McEwan, J. A., & Colwill, J. S. (1995). The sensory assessment of the thirst-quenchingcharacteristics of drinks. Food Quality and Preference, 7, 101–111.

Meilgaard, M., Civille, G. V., & Carr, B. T. (1999). Affective tests: Consumer tests and in-house panel acceptance tests. Boca Raton: CRC Press. pp. 231–254.

Peryman, D. R., & Pilgrim, F. J. (1957). Hedonic scale method of measuring foodpreferences. Food Technology, 11, 9–14.

Petit, C., & Sieffermann, J. M. (2007). Testing consumer preference for iced-coffee:Does the drinking environment have any influence? Food Quality and Preference,18, 161–172.

Pound, C., Duizer, L., & McDowell, K. (2000). Improved consumer productdevelopment. Part one. Is a laboratory necessary to assess consumer opinion?British Food Journal, 102, 810–820.

Qiu, C., & Yeung, C. W. M. (2008). Mood and comparative judgement: Does moodinfluence everything and finally nothing? Journal of Consumer Research, 34,658–669.

Runkel, P. J., & McGrath, J. E. (1972). Experimental simulations. In Research onhuman behavior: A systematic guide to method (pp. 96–103). New York: Holt,Rinehart and Winston, Inc..

Schwarz, N., & Clore, G. (1983). Mood, misattribution, and judgments of well-being:Informative and directive functions of affective states. Journal of Personality andSocial Psychology, 45, 513–523.

Stone, H., & Sidel, J. L. (2004). Affective testing. In Sensory evaluation practices(pp. 247–277). San Diego: Elsevier.