ABSTRACT: Hedonic tests are routinely used to assess the acceptance of food products. However, these single tests
may not be the best approach for predicting long-term use. The objectives of this study were, first, to check whether a
difference from reference score is more sensitive to changes in hedonic scores, second, to assess whether the labeled
affective scale (LAM) is more sensitive to differences than the 9-point scale, and third, to assess the effect of repeated
exposure on the hedonic scores of neophilic and neophobic panelists for familiar and novel foods. Two groups of 41
panelists were tested with either the 9-point hedonic scale or LAM scale. Panelists received a food neophobia ques-
tionnaire and were subsequently classified to neophobic, neophilic, or neutral. Ten foods, including 5 novel and 5
familiar, were used. In each session, 5 to 6 foods were served twice/week for 4 wk. Serving frequency ranged between
1 and 8 times (1, 2, 4, 6, 8). Data analyses were performed 3 times, using either absolute acceptability scores or rela-
tive scores, that is, the difference between absolute scores and scores for either the reference (cracker, RELFAM) or a
novel food (pickled-ginger, RELNOV) served in every session. The 3 analyses (absolute, RELFAM, and RELNOV) gen-
erated similar results with respect to the number of significant differences between foods. There was no major drift
in acceptability scores with sessions. A significant food effect was obtained (P < 0.05) and a significant food × neo-
phobia (P < 0.05) was noted for 2 novel foods, pickled ginger, and lychee, whereby neophobic panelists were less
accepting of them. Both scales were equally sensitive with some advantages for LAM over the 9-point hedonic scale.
Keywords: acceptability, LAM scale, neophobia, reference foods, repeated exposure
Quality
hedonic assessments and the consequent cost savings, several eficial for increasing the acceptability of healthier foods. Tuorila-
problems have been noticed. These drawbacks range from the Olikainen (1987) reported a reduction in liking for whole milk and
serving size (Zandstra and others 1999), to the artificial testing an increase in liking for the skim milk version. Similarly, Mattes
environment (Meiselman and others 2000a) and the frequency of (1993) obtained a significant decrease in preference for 4 preferred
exposure to food items (De Graaf and others 2005). In these labo- high fat level foods after a long-term reduced fat diet.
ratory tests, the measurement is restricted to momentary impres- The development of food boredom or a monotony effect is
sions and the development of the appreciation of products over demonstrated by a decrease in hedonic ratings with repeated expo-
time is not considered. Thus, they could not be always considered sure (Stubenitsky and others 1999; Zandstra and others 2000b). The
as good predictors of consumers’ behavior toward foods in real- monotony effect has been shown to be dependent on several fac-
life situations. Accordingly, the acceptability of foods may change tors, including the type of food (Schutz and Pilgrim 1958), its initial
after repeated consumption (Kamen and Peryam 1961; Tuorila- pleasantness (Schutz and Pilgrim 1958; Siegel and Pilgrim 1958),
Ollikainen and others 1986; Vickers and Holton 1998; Hetherington typical frequency of intake (Hetherington and others 2000, 2002),
and others 2000; Zandstra and others 2000a, 2000b; Hetherington and the availability of choice (Kamen and Peryam 1961; Zandstra
and others 2002; Chung and Vickers 2007), though there are stud- and others 2000b; Raynor and others 2006), which generally in-
ies that have revealed consistent hedonic scores over a short time creases acceptability. Regarding the type of food, staple foods have
period of few days (Goldman 1994). Therefore, acceptance could been shown to be less susceptible to a monotony effect (Schutz and
be reinforced (Pliner 1982) or could decline with repeated exposure Pilgrim 1958).
(Hetherington and others 2000, 2002). Furthermore, initial hedo- In addition, the change in acceptability with repeated exposure
nic ratings were maintained for some foods while that of others could depend on the novelty of the food and the food neophobia
decreased in previous studies (Schutz and Pilgrim 1958; Siegel and level of panelists. The Food Neophobia Scale (FNS) was developed
Pilgrim 1958). by Pliner and Hobden (1992) and has been the tool of choice for
measuring food neophobia levels. Several studies have shown that
“mere exposure” to the taste of unfamiliar foods increased liking for
MS 20080536 Submitted 7/17/2008, Accepted 11/5/2008. Authors are with these foods (Birch and Marlin 1982; Pliner 1982). In Pliner’s (1982)
Nutrition and Food Science Dept., American Univ. of Beirut, Riad El Solh investigation, acceptability of novel tropical fruit juices increased
1107 2020, Beirut, Lebanon. Direct inquiries to author Olabi (E-mail: after 20 trials and the exposure effect was explained as the dissipa-
ammar.olabi@aub.edu.lb).
tion of food neophobia.
R
C 2009 Institute of Food Technologists Vol. 74, Nr. 2, 2009—JOURNAL OF FOOD SCIENCE S97
doi: 10.1111/j.1750-3841.2008.01034.x
Further reproduction without permission is prohibited
Use of reference foods in acceptability . . .
A variety of sensory tools have been used to assess the accept- differences, and third, that neophobic panelists would be less ac-
ability of foods, including the 9-point hedonic scale (Peryam and cepting of novel foods. The objectives of this study were, first, to
Girardot 1952; Peryam and Pilgrim 1957), the labeled affective mag- check whether a difference from reference score is more sensitive
nitude scale (Schutz and Cardello 2001; Cardello and Schutz 2004), to changes in hedonic scores, second, to assess whether the labeled
the FACT scale (Schutz 1964), and others. However, despite the affective scale (LAM) is more sensitive to differences than the 9-
simplicity and ease of use of the 9-point hedonic scale, several point scale, and third, to assess the effect of repeated exposure on
problems were reported, including the lack of equivalence between the hedonic scores of neophilic and neophobic panelists for famil-
intervals (Peryam and Pilgrim 1957; Schutz and Cardello 2001), ex- iar and novel foods.
cessive use of the neutral category as a “safe place” (Jones and oth-
ers 1955), avoidance of end anchors, and lack of a true zero and Materials and Methods
of independent judgments (Stevens and Galanter 1957). Accord-
ingly, studies have compared the sensitivity and reliability of this Preliminary tests
scale compared with the magnitude estimation scale and obtained Preliminary tests were carried out with the aim of selecting ade-
better sensitivity with the magnitude estimation (Mc Daniel and quate foods for the experimental sessions. A total of 30 to 35 pan-
Sawyer 1981) or similar levels of sensitivity (Moskowitz and Sidel elists participated in each of the 7 preliminary tests. The panelists
1971; Lavenka and Kamen 1994; Barylko-Pikielna and others 2004) were mostly graduate students and staff from the Nutrition and
and similar levels of sensitivity, reliability, and precision (Warren Food Science Dept. at the American Univ. of Beirut. During each
and others 1982; Pearce and others 1986). The labeled affective session, panelists rated their acceptability for 5 food samples from
magnitude scale was developed to evaluate liking/disliking (Schutz different food categories using a 9-point hedonic scale. They also
and Cardello 2001) based on the study of Green and others (1993, rated the level of familiarity of each food on a 9-point categorical
1996) who devised the “labeled magnitude” scale (LMS), modeled scale with “very unfamiliar” and “very familiar” as its outermost
after the “category-ratio” scale (Borg 1982). Jeon and others (2004) categories. The goal of these preliminary tests was to select a ref-
compared the variation among the mean errors for these 2 scales erence food, in addition to 5 familiar and 5 novel foods. Standard
and demonstrated an equal level of discrimination ability between deviations as well as means of acceptability ratings and familiar-
the two, while studies by Green and others (2006) and Schutz and ity scores were calculated for the preliminary sessions. Food items
Cardello (2001) showed that the 9-point hedonic scale was a more with mean familiarity ratings higher than 6 were considered famil-
conservative sensory tool in determining differences among sam- iar products, and lower than 5 were considered novel products. In
ples and the LAM scale was more sensitive. addition, the selection of familiar and novel food items was lim-
Reference samples are regularly and successfully used in qual- ited to foods requiring little or no preparation. The familiar foods
ity control studies or in descriptive analysis as benchmarks for were typical of the Lebanese diet and were selected for the experi-
comparison and to stabilize panelists’ performance and reliability mental sessions based on high familiarity ratings, while the oppo-
(Nielsen and others 2005), in addition to its use in certain threshold site held true for novel foods, and varying acceptability levels to
methods such as the method of constant stimuli. However, the use span the acceptability spectrum. Foods from the main food cat-
of reference foods in hedonic tests is still rare despite the existence egories (main dish, side dish, dessert, soup, and beverage) were
S: Sensory & Food
of few studies. These included the use of the self-adjusting scale assessed in the preliminary tests to imitate all the items and the
(Gay and Mead 1992; Mead and Gay 1995) and comparisons be- variety of a meal. One to 2 foods were selected from each food
Quality
tween the 9-point hedonic scale with reference compared with the category for the experimental sessions based on the criteria re-
typical hedonic scale (Bergara-Almeida and da Silva 2002). A desert lated to novelty, familiarity, and acceptability as described previ-
chocolate bar formulated for warm-weather military rations was ously. Food items were purchased locally from different grocery
used by Lawless (1994) in the Barter scale to predict the total ap- stores.
peal of foods or meal combinations. In addition, Olabi (2001) used
reference foods in a study about the optimization of a space diet Main experiment
and found that the use of reference foods could be helpful when Stimuli. In every session, several foods were assessed, including
a significant session effect is expected and with incomplete design a reference food, which was included for the purpose of providing a
studies by increasing the precision of statistical tests. reference point for calculating relative scores, which could be more
In the development of innovative, high quality, and appealing sensitive than absolute scores, especially in the event of a session
foods, sensory evaluation is key. Foods that do not attain high con- effect. Several criteria were followed in the selection of the reference
sumer acceptance are destined to fail. It is important to provide the food. It had to have a neutral taste quality to minimize carryover
industry with methods and tools that will reduce risks and uncer- effects and was chosen to have little or no preparation to prevent
tainty in decision making and would be of great economical value. the changes in quality as well as in the sensory properties during
Previous studies discussed previously have revealed changes in ac- preparation. Thus, commercial foods that are staples fitted these
ceptability in repeated acceptability tests. Moreover, these changes requirements. In addition, the product had to have relatively high
were known to be affected by many factors, including context or acceptability and familiarity ratings with a low standard deviation
R
session effects. The main goal of this study was to assess the stabil- for acceptability. The salty cracker (ETI , Bursa, Turkey) fulfilled
ity and differentiating power of acceptance testing with and with- these requirements and was selected as the reference food in the
out the use of a hedonic reference. Of further interest was the use of present study.
R
the LAM scale, which is expected to stabilize consumers’ frame of The selected familiar foods were Falafel (Barbar , Beirut,
reference in making acceptability judgments (Schutz and Cardello Lebanon), canned Baba ghanouj (mashed roasted eggplants mixed
R
2001). Therefore, the hypotheses were first, that the use of a hedo- with tahini, lemon juice, and oil-Al Wadi Alakhdar , Chtoura,
R
nic reference would improve the sensitivity of assessing acceptabil- Lebanon), asparagus soup (Maggi , Bursa, Turkey), and apricot
ity changes through the use of relative scores, which in turn will nectar (mashed dried apricot, Damascus, Syria, dissolved in water
“capture” any session-related changes, instead of absolute scores, in the proportion of 1 kg:1 L in the laboratory) were obtained from
second, that the LAM would be more sensitive to acceptability local grocery stores.
As for the novel foods, different ethnic processed foods such as elists filled in the beginning of every session a questionnaire, which
Chinese, Thai, and French were purchased locally. These included included questions about mood, which was rated on a 7-point cat-
R
lychee (Monoprix , France), which was used as a dessert; pick- egory smiley scale (Chen and others 1996; Morien and others 2008),
led ginger (Bangkok, Thailand); the known Chinese item was used with “very pleasant” and “very unpleasant” as end anchors, and
R
as a side dish. Potato soup (Le Potage , Sevres, France) and chili hunger that was rated on a 15-cm line scale. The panelists were
R
(Hormel , San Antonio, Tex., U.S.A.) represented the soup and classified based on their neophobia scores to 1 of 3 categories: neo-
main dish categories, respectively. Additionally, the beverage used phobic, neophilic, or neutral. The categories were based on a ± 1.0
was freshly squeezed sugar cane juice. All the familiar and novel standard deviation (positive for neophobic, negative for neophilic)
food items are listed in Table 1. difference from the mean of a large sample obtained in the Pliner
and Hobden’s (1992) study. A neutral category included those who
Participants had a score of < ±1 standard deviation from the mean. Pliner and
Eighty-two panelists participated in the study. The panelists Hobden (1992) obtained a mean of 34.51 and classified panelists
were 23 male and 59 female (mean age = 22, age range = 18 to to neophobic and neophilic based on a 1 standard deviation differ-
54, SD = 5.93), mainly graduate, undergraduate students, and staff ence (11.86) from the mean.
at the American Univ. of Beirut. They were recruited through an Furthermore, panelists were asked to taste the food samples in
advertisement in the university electronic bulletin and posters at the order provided and to rate their degree of liking for each food,
campus buildings. They were randomly and evenly assigned to 1 of the level of familiarity, and their attitude toward each food item.
the 2 groups (41 panelists/group), whereby each group of panelists The assessments of acceptability were performed using either a
conducted the taste sessions using either the 9-point hedonic or the 9-point hedonic scale, or on a 10-cm vertical LAM scale anchored
labeled affective magnitude (LAM, reference Schutz and Cardello by “greatest imaginable dislike” at 0 and “greatest imaginable like”
2001) scale. To be eligible for participation, panelists had not to at 10. The scale presented to the panelists had no numbers, only
have a dislike for a large number of foods and not to have dietary re- descriptors. Panelists were asked to indicate their acceptability rat-
strictions or food allergies. Table 2 summarizes the main character- ing on the LAM scale by marking a slash (/) on any location on the
istics of the panelists. Panelists were rewarded with money coupons scale. All respondents (both acceptability scales) also chose their fa-
for their participation. miliarity ratings on a 9-point category scale labeled with numbers
for each category and with only “not at all familiar” and “extremely
Procedure familiar” as end anchors, as well as their attitude toward the food
On each test day, panelists arrived at the laboratory at least 2 h af- products as measured by the FACT scale consisting of 9 categories
ter their last meal (either mid-morning or mid-afternoon). Upon ar- assigned to statements listed from the most positive attitude such
rival, respondents were seated in individual booths with white fluo- as category “1” or “I would eat this every opportunity I have” to the
rescent lighting located in the sensory evaluation laboratory of the most negative food attitude such as category “9” or “I would eat this
Dept. of Nutrition and Food Science. The environment was calm only if I were forced to” (Schutz 1964).
and undisruptive. Then panelists were supplied with the food tray, Twenty to 30 g of each sample were presented in individual
a pen, and a questionnaire. 50-mL portion plastic cups (produced locally) coded with 3-digit
Quality
7-point scale assessing the food neophobia level with levels ranging of the soups and falafel, which were prepared on the morning of
from 1 (disagree strongly) to 7 (agree strongly). In addition, the pan- the session day. The soups and chili were heated in a microwave
(Daewoo, Seoul, South Korea, KOR-6C2B- 1200 Watts) for 30 and 60
s, respectively. Food stimuli were presented on a tray with a nap-
Table 1 --- Food items for taste sessions.
kin, plastic forks and spoons, and spring water for rinsing between
Food Food Times Session samples. It is noteworthy to mention that the order of the sam-
category item/type served served ples was counterbalanced in each session based on the design for
Main dish Falafela (F) 4 2, 4, 6, 8 5 or 6 samples as described by Macfie and Bratchell (1989), which
Chilib (N) 4 1, 3, 5, 7 balances the carryover effects. The counterbalancing was adopted
Appetizer Baba ghanoujc (F) 6 1, 3, 4, 5, 6, 7
for several reasons, first to prevent any carryover and order effects,
Pickled gingerd (N) 8 1 to 8
Soup Asparagus soupe (F) 2 2, 5 second because this study was conducted in a laboratory environ-
Potato soupf (N) 1 8 ment, with small portions served for every food and accordingly did
Beverage Apricot nectar (F) 1 7 not mimic a home or restaurant environment, and third because
Sugar cane juice (N) 2 1, 4 with the presence of the aforementioned conditions, it is not pos-
Dessert Lycheeg (N) 6 1, 2, 3, 4, 6, 8
Reference food Salty crackersh (F) 8 1 to 8 sible to call the group of foods served in every session a meal but
rather a meal simulation. There are pros and cons for having a “real
F = familiar; N = novel. Brands: a Barbar, Beirut, Lebanon; b Hormel, San
Anotonio, Tex., U.S.A.; c Al Wadi Alakhdar, Chtoura, Lebanon; d Bangkok, life” compared with a laboratory type of scenario and obviously the
Thailand; e Maggi, Nestle, Bursa, Turkey. f Le Potage, Sevres, France.g Monoprix, choice in this study was for the latter, which provides better control
France. h ETI, Bursa, Turkey.
at the expense of having a more representative type of experience.
Table 2 --- Characteristics of panelists. Additionally, new 3-digit random numbers were used to code the
food items between sessions even for the repeated samples thus,
Gender Neophobia level
reducing number bias.
Female Male Neophobic Neophilic Intermediate
9-point 26 15 3 9 29 Experimental design
hedonic scale Eighty-two panelists were evenly and randomly divided be-
LAM scale 33 8 4 9 28 tween 2 groups, one using the 9-point hedonic scale and the
Total 59 23 7 18 57
other using the LAM scale. Ten foods, including 5 novel and
5 familiar, were served to all the panelists. The selection of foods ing frequency (TSERV: 1, 2, 4, 6, and 8 times), neophobia category
as familiar or novel was based on the familiarity scores of the (neophobic, neophilic, and neutral), food (1 to 10; nested within
preliminary sessions. The serving frequency ranged between 1 and novelty and TSERV), and panelist (nested within scale and neo-
8 times (1, 2, 4, 6, 8), whereby for each serving frequency there is phobia category). In addition, mood and hunger were added as
both a familiar and a novel food. The serving frequency was var- covariates to the model. An analysis of variance was performed
ied for the foods to assess the effect of different serving frequencies using PROC GLM of the SAS statistical software (SAS version 8.02,
on the changes in acceptability scores and to increase the number 1999-2001, SAS Inst. Inc., Cary, N.C., U.S.A.). In the statistical
of tested foods. Five to 6 foods were served to the panelists twice a model, the response variable was the acceptability of the foods.
week at lunchtime and included the typical items of a meal (main Panelist, food, and session were random effects, while scale type,
dish, appetizer, and so on) for a period of 4 wk. The foods were ran- food novelty, serving frequency, neophobia category, mood, and
domized over the sessions in such a way that the reference food hunger were fixed effects in the model. Each main effect was tested
(cracker) and 1 novel food (pickled ginger) were repeated every ses- as well as all 2-way interactions and selected 3-way interactions.
sion. It is important to note that pickled ginger was not meant to be In addition, the analysis was repeated twice using relative scores,
another reference food especially that it did not have a low standard which represented the difference between the absolute scores and
deviation for acceptability in the preliminary tests, had low aver- the reference scores from either the regular reference (cracker) or
age acceptability and low familiarity ratings, all opposite qualities the novel food (pickled ginger) that was served in every session.
to what is needed in a reference food. However, it was important The relative values were calculated as a difference score for each
for the sake of balancing the experimental design to include a novel individual panelist in each individual session. Mean separation was
food that was served in every session, similar to the reference food performed using Tukey’s honestly significant difference (HSD). Sig-
(cracker). Hence it was considered useful to do the analyses using nificance was preestablished at α < 0.05.
the relative scores with both the cracker and pickled ginger to com- Separate analyses of variance were conducted with the absolute
pare and contrast the results of these 2 analyses with each other and scores for the 9-point hedonic scale or the LAM original data to ob-
with that of the absolute scores. The detailed experimental design tain the results that are summarized in Table 5, in addition to calcu-
of the study is in Figure 1 and summarized in Table 3. lating the normality of data and the frequency of neutral responses.
Naturally, the analysis of variance model did not include any scale
Statistical analyses effect or any of its interactions. Mean separation was performed us-
Two types of analyses of variance were conducted. The 1st one ing Tukey’s HSD. Significance was preestablished at α < 0.05.
included the data for the 2 scales and is summarized in Table 4, 5, Moreover, the normality of the data per food and per scale was
and 6. For this analysis, the effect of the different predictor vari- checked using the Anderson–Darling (AD) values. AD values reflect
ables on the acceptability of the foods was assessed. The main the goodness of fit to a standard normal distribution (lower values
factors were: scale type (9-point compared with LAM), food nov- indicate a better fit), the LAM scale was closer to normal distribu-
elty (familiar compared with novel), session (1 through 8), serv- tion than the 9-point hedonic scale.
S: Sensory & Food
F1 F2 F3 F4 F5 F1 F2 F3 F4 F5 F1 F2 F3 F4 F5 F1 F2 F3 F4 F5 Food
(Familiar: 1-5; Novel: 6-10)
Results and Discussion of variance included in Table 5) are summarized in the 2nd column
of Table 4. The familiar reference food used (cracker) had the high-
Comparison of different models: absolute est acceptability score among the foods (8.1). On the other hand,
compared with relative scores the novel food (pickled ginger) that was served in every session re-
The strongest effect was the differences among the foods. The ceived the lowest acceptability rating (3.0). Differences (P < 0.05)
acceptability scores (absolute scores) for foods (from the analysis between the top 3 foods were found. These foods were cracker,
falafel, and lychee, with ratings of 8.1, 7.6, and 7.0, respectively,
Table 4 --- Acceptability means for 10 food items using and differences were also found between the bottom three, namely,
absolute scores and relative to both familiar (RELFAM) Baba ghanouj, sugar cane juice, and pickled ginger, with 5.3, 3.7,
and novel food (RELNOV). and 3.0, respectively. The other foods (apricot nectar, chili, aspara-
Food name Absolute values RELFAM RELNOV gus and potato soups) had scores ranging between 5.8 and 7.0 and
Pickled ginger (N) 3.0 a
5.2 a
−0.0a in general did not show differences. The average acceptability of fa-
Sugar cane juice (N) 3.7b 4.4b −0.8b miliar foods tended to be higher than that of novel foods, though
Baba ghanouj (F) 5.3c 2.9c −2.3c this was not shown to be significant in the analysis of variance
Apricot nectar (F) 5.8cd 2.3cd −2.9cd (Table 5).
Chili (N) 6.8de 1.4de −3.8de
Analysis of variance results for absolute scores. The analysis
Asparagus soup (F) 6.9ef 1.2def −4.0def
Potato soup (N) 7.0def 1.2def −4.0def of variance results for the absolute scores are summarized in the
Lychee (N) 7.0e 1.1ef −4.0ef 3rd column of Table 5. Differences were obtained for food (nested
Falefel (F) 7.6f 0.6f −4.6f within TSERV and novel; Table 4, P < 0.001). The panelist × TSERV
Crackers (F) 8.1g 0.0g −5.2g and panelist × novel interactions (P < 0.001) are an indication that
Means in a column followed by the same letter are not significantly different the serving frequency and novelty had a different effect on the gen-
(P > 0.05).
RELFAM: Relative to familiar reference food. eral acceptability of foods for different panelists. This outcome is
RELNOV: Relative to novel food item repeated in every session. to be expected because panelists always differ and hence the need
Table 5 --- Significance of predictor variables (F-values and P-values) in 3 different models (absolute scores, RELFAM,
and RELNOV) for main effects and their interactions.
Absolute scores RELFAMb RELNOVc FACTd
Variablesa df Error term F -values
Panelist (scale-neophobia) 76 Error 1 1.00 1.80∗ ∗ ∗ 4.73∗ ∗ ∗ 4.14∗ ∗ ∗
Session 7 Error 2 0.11 0.32 0.44 0.83
Scale 1 Error 3 0.05 1.00 0.26 1.39
Times served (TSERV) 4 Error 4 0.07 0.07 0.07 0.11
Novel (scale-neophobia) 1 Error 5 0.48 0.44 0.47 0.02
Neophobia level 2 Error 6 0.86 1.28 1.19 0.90
Hunger 1 MSE2 1.58 0.00 2.11 0.32
Quality
Panelist × TSERV 304 MSE 5.61∗ ∗ ∗ 5.32∗ ∗ ∗ 4.97∗ ∗ ∗ 3.55∗ ∗ ∗
Panelist × novel 76 MSE 13.07∗ ∗ ∗ 12.33∗ ∗ ∗ 11.50∗ ∗ ∗ 7.53∗ ∗ ∗
Scale × food 4 MSE 4.76∗ ∗ ∗ 4.49∗ ∗ 4.19∗ ∗ 146.15∗ ∗ ∗
Food × neophobia 8 MSE 2.52∗ ∗ 2.39∗ 2.12∗ 1.37
Hunger × food 4 MSE 6.06∗ ∗ ∗ 5.62∗ ∗ ∗ 4.80∗ ∗ ∗ 1.46
Mood × food 4 MSE 3.05∗ 2.85∗ 3.04∗ 0.80
Mood × neophobia 2 MSE 1.77 3.76∗ 0.04 0.47
Session × neophobia 14 Error 8 2.31 3.47 1.82 5.13
Novel × neophobia 2 Error 9 0.80 0.80 0.87 1.37
Scale × novel 1 Error 10 0.09 0.09 0.07 0.58
Scale × neophobia 2 Error 11 0.85 0.18 2.69 1.80
Session × novel 7 Error 12 1.93 2.19 2.58 0.86
TSERV × neophobia 8 Error 13 0.71 0.71 0.74 0.46
Scale × session 7 Error 14 1.04 1.90 0.95 0.55
Scale × TSERV 4 Error 15 0.29 0.28 0.30 0.23
Hunger × panelist 76 MSE 0.66 1.16 1.18 0.88
Hunger × scale 1 MSE 0.42 0.70 1.09 0.07
Mood × scale 1 MSE 1.52 0.53 1.80 3.13
Hunger × session 7 MSE 0.45 0.40 0.52 0.40
Mood × session 7 MSE 0.19 0.95 0.22 0.85
Hunger × TSERV 4 MSE 0.12 0.13 0.48 0.41
Mood × TSERV 4 MSE 0.20 0.24 0.10 0.59
Hunger × novel 1 MSE 1.79 1.58 0.34 0.43
Mood × novel 1 MSE 0.02 0.00 0.25 2.46
Hunger × neophobia 2 MSE 0.56 0.03 0.08 0.06
Hunger × mood 1 MSE 1.40 0.01 0.68 0.34
Scale × session × neophobia 14 MSE 0.61 0.72 1.43 0.69
Scale × session × novel 7 MSE 0.20 0.19 0.18 1.00
Session × novel × neophobia 14 MSE 0.34 0.34 0.31 0.34
Scale × TSERV × neophobia 8 Error 16 2.31∗ 5.32∗ ∗ ∗ 4.97∗ ∗ ∗ 1.37
Scale × novel × neophobia 2 Error 17 1.31 1.32 0.37 0.34
a
Response variable is acceptability.
b
RELFAM: Relative scores to cracker (familiar reference food), c RELNOV: Relative scores to the pickled ginger (novel food item repeated in every session), d FACT:
Food action rating scale, ∗ P < 0.05, ∗ ∗ P < 0.01,∗ ∗ ∗ P < 0.001.
to have a panelist effect. In addition, scale × food, hunger × food ceptions were panelist effect (P > 0.05) with the absolute ac-
(P < 0.001), and mood × food (P < 0.05) interactions reveal the ceptability scores and mood × neophobia level, which was sig-
different effects of scale, mood, and hunger on the acceptability nificant (P < 0.05) for RELFAM but not for the absolute values
of different foods, again showing the benefit of including mood or RELNOV. Session effect was also not significant (P > 0.05) in
and hunger as covariates in the model. Unfortunately, mood and all models, confirming the absence of a trend in the acceptabil-
hunger were continuous variables and accordingly the means of ity scores with time. The same level of significance (P-value) was
the different levels of these interactions could not be compared. found with the different models for most main factors and the 2-
The food × neophobia level interaction (P < 0.01) demonstrated or 3-way interactions; thus differences were noted for food (nested
that the acceptability of certain foods was different for different for time served and novelty), panelist × TSERV (nested for scale
neophobia groups. This is further discussed in the “Effect of food type and neophobia level), panelist × novel (nested for scale type
neophobia” section below. As for the scale × TSERV × neo in- and neophobia level), hunger × food and scale × neo × TSERV
teraction, naturally there were significant differences (P < 0.001) (P < 0.001), and mood × food interaction (P < 0.05). However,
between the levels of the different foods, as demonstrated by different levels of significance (P-value) were noted for food ×
differences for varying levels of TSERV (for example, the accept- neophobia, scale × food, and mood × neophobia. Scale × food
ability of TSERV = 4 for the 9-point scale and neophilic panelists was more significant with the absolute values model (P < 0.001)
was significantly different (P < 0.001) from the acceptability of than the RELFAM and RELNOV models (P < 0.01). The same ap-
TSERV = 8 for the LAM scale and neophobic panelists). However, plied to food × neophobia, though with different significance levels
when comparisons were made between the 3 neophobia groups for (0.001 compared with 0.05, respectively). As for mood × neopho-
the same scale and the same TSERV, differences (P < 0.05) were ob- bia interaction, it was significant (P < 0.05) only with the RELFAM
tained for some or all of the neophobia groups for the 9-point scale model. Given the results above, it could be noted that in general,
and TSERV = 6, and for the LAM scale and TSERV = 8, respectively. the relative sores models had a similar sensitivity to differences in
The least squares means were neophilic = 6.9a , intermediate = the investigated variables and interactions as the typical absolute
6.1ab , and neophobic = 5.2c ; neophilic = 6.6a , intermediate = 5.4b , scores model. All of the above-mentioned interactions have been
and neophobic = 4.3c for the 9-point scale and TSERV = 6 and for described in the previous section.
the LAM scale and TSERV = 8, respectively. Variables analyzed using the FACT scores are summarized in
Relative compared with absolute scores. The results of the the 6th column of Table 5. Similar results to the absolute scores
analyses of variance for the relative scores (cracker: RELFAM and were obtained for panelist × novel and panelist × TSERV and
pickled ginger: RELNOV) are summarized in Table 5. In general, scale × food. On the other hand, other variables were shown to
the relative scores models had similar sensitivity to the effects be significant with the absolute scores but not with the FACT.
and interactions of the variables as the absolute scores. The ex- This applied to food × neo, hunger × food, mood × food, and
scale × TSERV × neo.
The acceptability scores (absolute scores and relative scores) for
Table 6 --- Least squares mean absolute acceptability
foods are summarized in Table 4. The 10 different foods could be
scores for food items using the labeled affective mag-
nitude scale (LAM) and 9-point hedonic scale. arbitrarily classified to 3 groups: highly liked (cracker, falafel, ly-
S: Sensory & Food
Table 7 --- Mean acceptability ratings and standard deviations for foods through taste sessions.
Session
Food 1 2 3 4 5 6 7 8 SD
a
Cracker 8 (1.1) 8 (1.2) 8.1 (1.2) 8.2 (1.2) 8.2 (1.3) 8.1 (1.3) 8.2 (1.3) 8.3 (1.2) 0.1
Pickled ginger 3 (2.9) 3 (2.8) 3.2 (3.2) 3.1 (3.2) 3.3 (3.1) 3.2 (3.2) 3.2 (3.1) 3.2 (3.2) 0.1
Baba ghanouj 5.7 (2.5) 5.3 (2.5) 5.2 (2.7) 5.2 (2.6) 5.3 (2.5) 5.4 (2.6) 0.2
Lychee 7 (2.2) 6.9 (2.3) 7.2 (2.1) 7.1 (2.3) 7.2 (2.3) 7.3 (2.2) 0.1
Falafel 7.8 (1.4) 7.8 (1.5) 7.8 (1.5) 7.9 (1.7) 0.1
Chili 7.1 (1.8) 6.9 (2.0) 7.1 (2.0) 7.2 (2.0) 0.1
Asparagus soup 6.7 (1.9) 6.8 (2.1) 0.1
Sugar cane juice 3.8 (2.2) 3.7 (2.5) 0.1
Apricot nectar 5.9 (2.3)
Potato soup 6.9 (1.9)
SD = standard deviation of mean ratings for each food in all sessions.
a
Standard deviation of mean ratings for individual food in single session.
RELNOV scores and resulted in inflating the standard deviations noted for the other foods. Cracker was different from falafel and ly-
for all the foods with the RELNOV model, thereby resulting in less chee with the LAM, while no differences were obtained with the 9-
differences between foods. This solidified the original hypothesis point hedonic scale when similar comparisons were made between
of the authors that a hedonic reference should have low variability, the top 3 foods, namely, cracker, falafel, and potato soup. This di-
high familiarity, and high acceptability. vergence implies that the LAM scale could be more sensitive to dif-
After the identification of significant predictors (Table 5), in- ferences in highly liked items, an outcome that was obtained by
cluding 2- and 3-way interactions (P < 0.05), the number of sig- Schutz and Cardello (2001). They found that both scales had equiv-
nificant differences was counted for each significant variable or alent levels of sensitivity except for high level of effect and related
interaction for each model separately to assess the sensitivity of this outcome to the presence of the end-point “greatest imaginable
each model in detecting the maximum possible number of differ- liking/disliking” in the LAM scale that enables more extreme rat-
ences. Of course, this comparison remains one of qualitative na- ings than “like/dislike extremely” in the 9-point hedonic scale.
ture given that these variables are already significant and some- However, it is important to note that in the Schutz and Cardello’s
times have the same level of significance (P-value). This type of (2001) study the same panelists used both scales in separate ses-
comparison has been used in previous studies (Stoer and Lawless sions, while in our case, there were 2 separate groups, each group
1993; Schutz and Cardello 2001). The predictor variables that were evaluating foods using a specific scale (either the 9-point hedonic
compared in this study were food, scale × food, food × neo, and scale or the LAM scale). Using the same panelists for both scales
scale × TSERV × neo. The number of significant differences was might have improved the sensitivity of “between scales” compar-
relatively similar for food (34, 32, and 32), scale × food (116, 116, isons in a similar manner to the difference between a paired t-
and 105), and food × neo (228, 221, and 241) with the absolute, test compared with a 2-sample t-test due to less variability for the
RELFAM, and RELNOV models, respectively. The exception was same panelists than for different panelists. However, this was not
scale × TSERV × neo (109, 105, and 191) where more differences the case for this study because having the same panelists assess
were obtained with RELNOV compared to the 2 other models. Con- both scales would have introduced the training effect factor and
sequently, the 3 models had similar levels of sensitivity because no made an already complex design more intricate. In addition, the
major differences were observed by using either the absolute rat- order for moderately liked foods was different between the 9-point
ings or the relative ratings. However, it should be noted that these hedonic and LAM scale groups. For the 9-point scale the ranking
numbers were very similar in absolute and RELFAM, while RELNOV for acceptability scores in decreasing order was: lychee, chili and
had slightly different numbers. asparagus soup, and apricot nectar, while for LAM it was aspara-
gus soup, potato soup, chili, and apricot nectar. It is important to
mention that no significant differences existed between all of these
Comparison of acceptability scales foods for both scales. Mean acceptability scores suggested that al-
(9-point and LAM) though the 2 panelist groups used 2 different scales to evaluate ac-
Sensitivity to significant differences among food items. ceptability, the results showed similar tendencies for acceptance for
There was a scale × food interaction (P < 0.05) for all mod- both groups. The rank of the different foods in terms of their ac-
els as noted previously. The least squares mean absolute accept- ceptability was the same for both scales except for the moderately
Quality
the same extent (that is, having the same superscripts) with both optimal liking because it was more sensitive to differences among
scales for the lower acceptability foods (pickled ginger, sugar cane well-liked foods and had relatively similar results when compared
juice, Baba ghanouj, and apricot nectar). However, a difference was to the 9-point hedonic scale for the other foods.
Pickled ginger
5
Pickled ginger
Lychee (Phil)
4
Lychee (Phob)
Chili
3
Chili
1
1 2 3 4 5 6 7 8
not the case with the reference food (cracker), with 8.66 compared mere exposure effect. A comparison between this study and previ-
with 24.9 for LAM and 9-point hedonic scale, respectively. Both
Quality
Quality
ception of pickled ginger, and to a certain extent chili and lychee References
(Figure 2), 3 novel foods, which is a phenomenon noted previously. Barylko-Pikielna N, Matuszewska I, Jeruszka M, Kozlowska K, Brzozowska A,
Roszkowski W. 2004. Discriminability and appropriateness of category scaling ver-
However, it is important to mention that there was no novel × neo- sus ranking methods to study sensory preferences in elderly. Food Qual Pref 15:167–
phobia interaction effect. In addition, although repeated expo- 75.
Bergara-Almeida S, da Silva M. 2002. Hedonic scale with reference: performance in
sure tends to have a positive impact on neophobia, neophobics obtaining predictive models. Food Qual Pref 13:57–64.
in the present intervention were found to have similar acceptabil- Birch LL, Marlin DW. 1982. I don’t like it; I never tried it: effects of exposure on two-
year-old children’s food preferences. Appetite 3:353–60.
ity ratings for foods during the 4-wk study with a slight increase Borg, GAV. 1982. Psychophysical bases of perceived exertion. Med Sci Sport Exer
and/or decrease through time. This outcome is not in agreement 14:377–81.
Cardello AV, Schutz HG. 2004. Numerical scale-point locations for constructing the
with other studies showing a positive effect of varied and multi- LAM (labeled affective magnitude) scale. J Sens Stud 19:341–6.
ple food exposures on neophobia in human infants (Gerrish and Cardello AV, Schutz HG, Lesher LL, Merill E. 2005. Development and testing of a la-
beled magnitude scale of perceived satiety. Appetite 44:1–13.
Mennella 2001), children (Loewen and Pliner 1999; Wardle and Chen AW, Resurreccion AVA, Paguio LP. 1996. Age appropriate hedonic scales to mea-
others 2003), and adults (Pliner and others 1993). On the other sure food preferences of young children. J Sens Stud 11:141–63.
Chung S, Vickers Z. 2007. Influence of sweetness on the sensory-specific satiety and
hand, the absence of a major effect of multiple food exposure long-term acceptability of tea. Food Qual Pref 18:256–64.
on food neophobia levels in a weight reduction program (WRP) Crandall CS. 1984. The liking of foods as a result of exposure: eating doughnuts in
of adolescents, which was maintained for a period of 9 mo (Ri- Alaska. J Soc Psychol 125(2):187–94.
De Graaf C, Kramer FM, Meiselman HL, Lesher LL, Baker-Fulco C, Hirsch ES, Warber
gal and others 2006), was in line with the findings of the present J. 2005. Food acceptability in field studies with US army men and women: relation-
experiment. ship with food intake ad food choice after repeated exposures. Appetite 44:23–31.
Gay C, Mead R. 1992. A statistical appraisal of the problem of sensory measurement. J
Given the results obtained in this study, it is recommended to Sens Stud 7:205–28.
have a higher number of exposures and a longer duration for sim- Gerrish CJ, Mennella JA. 2001. Flavor variety enhance food acceptance in formula-fed
infants. Am J Clin Nutr 73:1080–5.
ilar studies. Both of these changes could in turn induce significant Goldman A. 1994. Predicting product performance in the marketplace by immediate-
changes in acceptability with repeated exposure, an element that and extended-use sensory testing. Food Technol 48(10):103–6.
Green BG, Shaffer GS, Gilmore MM. 1993. Derivation and evaluation of a seman-
was relatively missing in this study. Another possible change would tic scale of oral sensation magnitude with apparent ratio properties. Chem Senses
be having the same panelists assess the foods with both scales to 18:683–702.
Green BG, Dalton P, Cowart B, Shaffer G, Rankin K, Higgins J. 1996. Evaluating the “la-
enable a more sensitive “between scales” comparison. In addition, beled magnitude scale” for measuring sensations of taste and smell. Chem Senses
repeating this study in either a better-controlled environment, such 21:323–35.
Green JL, Bratka KJ, Drake MA, Sanders TH. 2006. Effectiveness of category and line
as the army or boarding schools, or in a less controlled environment scales to characterize consumer perception of fruity fermented flavor in peanuts. J
such as an in-home use test would be advantageous. Sens Stud 21:146–54.
Henriques AS, King SC, Meiselman, HL. 2008. Consumer segmentation based on food Raudenbush B, Frank RA. 1999. Assessing food neophobia: the role of stimulus famil-
neophobia and its application to product development. Food Qual Pref 20:83–91. iarity. Appetite 32:261–71.
Hetherington MM, Bell A, Rolls BJ. 2000. Effects of repeat consumption on pleasant- Raynor HA, Niemeier HM, Wing RR. 2006. Effect of limiting snack food variety on
ness, preference and intake. Br Food J 102(7):507–21. long-term sensory-specific satiety and monotony during obesity treatment. Eating
Hetherington MM, Pirie LM, Nabb S. 2002. Stimulus satiation: effects of repeated ex- Behav 7:1–14.
posure to foods on pleasantness and intake. Appetite 38:19–28. Rigal N, Frelut M, Monneuse M, Hladik C, Simmen B, Pasquet P. 2006. Food neopho-
Jeon SY, O’Mahony M, Kim KO. 2004. A comparison of category and line scales under bia in the context of a varied diet induced by a weight reduction program in mas-
various experimental protocols. J Sens Stud 19:49–66. sively obese adolescents. Appetite 46:207–14.
Jones LV, Peryam DR, Thurstone LL. 1955. Development of a scale for measuring sol- Schutz HG, Cardello AV. 2001. A labeled affective magnitude scale for assessing food
diers’ food preferences. Food Res 20:512–20. liking/disliking. J Sens Stud 16:117–59.
Kamen JM, Peryam DR. 1961. Acceptability of repetitive diets. Food Technol 26:173–7. Schutz HG, Pilgrim FJ. 1958. A field study of food monotony. Psychol Reports 4:559–
Lavenka N, Kamen J. 1994. Magnitude estimation of food acceptance. J Food Sci 65.
59(6):1322–4. Schutz HG. 1964. A food action rating scale for measuring food acceptance. Paper
Lawless HT. 1994. Contextual and measurement aspects of acceptability. Final Report presented at the 24th Annual Meeting of the Institute of Food Technologists, May
nr TCN 94178, U.S. Army Research Office. 24-28, 1964, Washington, D.C.
Loewen R, Pliner P. 1999. Effects of prior exposure to palatable and unpalatable Siegel PS, Pilgrim FJ. 1958. The effect of monotony on the acceptance of food. Am J
novel foods on children’s willingness to try novel foods in stimulating and non- Psychol 71:756–9.
stimulating situations. Appetite 32:351–66. Stein LJ, Nagai H, Nakagawa M, Beauchamp GK. 2003. Effects of repeated exposure
Macfie HJ, Bratchell N. 1989. Designs to balance the effect of order of presentation and health-related information on hedonic evaluation and acceptance of a bitter
and first order carry over effects in hall tests. J Sens Stud 4:129–48. beverage. Appetite 40:119–29.
Mattes RD. 1993. Fat preference and adherence to a reduced-fat diet. Am J Clin Nutr Stevens SS, Galanter EH. 1957. Ratio scales and category scales for a dozen of percep-
57:373–81. tual continua. J Exp Psychol 54(6):377–411.
Mattes RD. 1994. Influences on acceptance of bitter foods and beverages. Physiol Be- Stoer NL, Lawless HT. 1993. Comparison of single product scaling and relative-
hav 56:1229–36. to-reference scaling in sensory evaluation of dairy products. J Sens Stud 8:257–
Mc Daniel MR, Sawyer FM. 1981. Preference testing of whiskey sour formulation: 70.
magnitude estimation versus the 9-point hedonic scale. J Food Sci 46:182–5. Stubenitsky K, Aaron JI, Catt SL, Mela DJ. 1999. Effect of information and extended
Mead R, Gay C. 1995. Sequential design of sensory trials. Food Qual Pref 6:271–80. use on the acceptance of reduced-fat products. Food Qual Pref 10:367–76.
Meiselman HL, Johnson JL, Reeve W, Crouch JE. 2000a. Demonstrations of the influ- Tuorila H, Meiselman HL, Bell R, Cardello AV, Johnson W. 1994. Role of sensory and
ence of the eating environment on food acceptance. Appetite 35:231–37. cognitive information in the enhancement of certainty and liking for novel and fa-
Meiselman HL, de Graaf C, Lesher L. 2000b. The effects of variety and monotony on miliar foods. Appetite 23:231–46.
food acceptance and intake at a midday meal. Physiol Behav 70:119–25. Tuorila H, Lahteenmaki L, Pohjalainen L, Lotti I. 2001. Food neophobia among the
Morien A, Garrison D, Keeney Smith N. 2008. Range of motion improves after massage Finns and related responses to familiar and unfamiliar foods. Food Qual Pref 12:29–
in children with burns: a pilot study. J Bodywork Movement Therap 12:67–71. 37.
Moskowitz HR, Sidel JL. 1971. Magnitude and hedonic scales of food acceptability. J Tuorila-Olikainen H. 1987. Selection of milk with varying fat contents and related to
Food Sci 36:677–80. overall liking, attitudes, norms and intentions. Appetite 8:1–14.
Nielsen D, Hyldig G, Sorensen R. 2005. An effective way to minimize drifting and mon- Tuorila-Ollikainen H, Lahteenmaki L, Salovarra H. 1986. Attitudes, norms, intentions
itor the performance of a sensory panel during long-term projects—a case study and hedonic responses in the selection of low salt bread in a longitudinal choice
from a project on Herring quality. J Sens Stud 20:35–47. experiment. Appetite 7:127–39.
Olabi A. 2001. The optimization of a bioregenerative life support space diet [PhD the- Vickers Z, Holton E. 1998. A comparison of taste test ratings, repeated consumption,
sis]. Ithaca, N.Y.: Cornell Univ. and post-consumption ratings of different strengths of iced tea. J Sens Stud 13:199–
Peryam DR, Girardot NF. 1952. Advanced taste-test method. Food Engr 24:58–61. 212.
Peryam DR, Pilgrim FJ. 1957. Hedonic scale method of measuring food preferences. Wardle J, Herrera ML, Cooke L, Gibson EL. 2003. Modifying children’s food prefer-
Food Technol 11:9–14. ences: the effects of exposure and reward on acceptance of an unfamiliar vegetable.
Pearce JH, Korth B, Warren CB. 1986. Evaluation of three scaling methods for hedo- Eur J Clin Nutr 57:341–8.
nics. J Sens Stud 1(1):27–46. Warren C, Pearce J, Korth B. 1982. Magnitude estimation and category scaling. ASTM
Pliner P. 1982. The effect of mere exposure on liking for edible substances. Appetite Standard News 10:15–6.
3:283–90. Zandstra EH, de Graaf C, van Trijp HCM, van Staveren WA. 1999. Laboratory hedonic
Pliner P, Hobden K. 1992. Development of a scale to measure the trait of food neo- ratings as predictors of consumption. Food Qual Pref 10:411–8.
phobia in humans. Appetite 19:105–20. Zandstra EH, de Graaf C. Mela DJ, van Staveren WA. 2000a. Short- and long-term ef-
S: Sensory & Food
Pliner P, Pelchat M, Grabski M. 1993. Reduction of neophobia in humans by exposure fects of changes in pleasantness on food intake. Appetite 34:253–60.
to novel foods. Appetite 20:111–23. Zandstra EH, de Graaf C, van Trijp HCM. 2000b. Effects of variety and repeated in-
Porcherot C, Issanchou S. 1998. Dynamics of liking for flavored crackers: test of pre- home consumption on product acceptance. Appetite 35:113–9.
Quality