- Email: [email protected]
Sensory profile and Beijing youth preference of seven cheese varieties
Food Quality and Preference 22 (2011) 101–109
Contents lists available at ScienceDirect
Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual
Sensory profile and Beijing youth preference of seven cheese varieties X.Y. Zhang a,b, H.Y. Guo a,b, L. Zhao a,b, W.F. Sun c, S.S. Zeng d, X.M. Lu a,b, X. Cao a,b, F.Z. Ren a,b,⇑ a
College of Food Science & Nutritional Engineering, China Agricultural University, Beijing 100083, China Key Laboratory of Functional Dairy, Ministry of Education, Beijing 100083, China c College of Food Science, Xinjiang Agricultural University, Urumuchi 830052, China d Department of Agriculture & Natural Resources, Langston University, P.O. Box 1730, Langston, OK 73050, United States b
a r t i c l e
i n f o
Article history: Received 3 December 2008 Received in revised form 21 April 2010 Accepted 9 August 2010 Available online 13 August 2010 Keywords: Cheese flavor Descriptive sensory analysis External preference mapping Cluster analysis China
a b s t r a c t The sensory characteristics that determined consumer preference of seven imported cheeses were investigated. Descriptive sensory analysis was performed by seven trained panelists who used 24 descriptors to quantitatively describe the sensory characteristics of these cheeses. Principal component analysis of descriptive data found significant differences between the cheeses (P < 0.05) on the first three principal components which accounted for 44%, 29% and 11% of the experimental variance, respectively. In parallel, the preference of 217 Beijing youth on these cheeses was expressed on a nine-point hedonic scale. Cluster analysis of the preference data identified five consumer clusters with different preferences and demonstrated that there existed a potential market for each of these cheeses. External preference mapping indicated a relative homogenous preference of Beijing youth for cheese flavor. ‘‘Soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures appeared to be drivers of liking for most Chinese youth, whereas ‘‘bitter”, ‘‘salty”, ‘‘umami” and ‘‘free fatty acid” flavors and ‘‘firmness” texture were drivers of disliking. In addition, ‘‘nutty” as a typical aged flavor in cheese was desirable to some consumer cluster and could be cultivated among Chinese. Ó 2010 Published by Elsevier Ltd.
1. Introduction Over the last two decades, the demand for dairy products in China has increased drastically (Yang, 2007). With the improvement of people’s living standards, the structure of dairy products has also changed. Chinese have mainly consumed fresh milk and yogurt. Nowadays, more attention has been paid to non-traditional dairy products such as probiotic beverage and cheese. Cheese is a relatively new type of food in the traditional Chinese diet. In 2006 the cheese consumption in China was 20,000 tons, of which 90% were imported (Gee & Townson, 2007). Although the cheese import has increased markedly in recent years (1320 tons annually 2000–2006) (Dairy Association of China, 2007), the Chinese cheese market is still at its initial stage. The current Chinese cheese consumption is 200 g per capita annually according to Dairy Association of China (2007). In comparison, the annual per capita consumption is 16 kg in the United States and 24 kg in France (International Dairy Federation, 2007). Diet cultures and preference of ethnic Chinese are significantly different from those of Western people. To date, there is no data available regarding the preference of Chinese consumers on ⇑ Corresponding author. Address: College of Food Science & Nutritional Engineering, China Agricultural University, P.O. Box 303, No. 17 Qinghua East Road, Haidian, Beijing 100083, China. Tel./fax: +86 10 62736344. E-mail address: [email protected] (F.Z. Ren). 0950-3293/$ - see front matter Ó 2010 Published by Elsevier Ltd. doi:10.1016/j.foodqual.2010.08.007
cheese. Due to the great variability of cheese varieties and flavors, an extensive and systematic study would be important to the development of a cheese industry in China. Sensory attributes (flavor and texture) are critical to the identity of cheese and consumer acceptance (Van Hekken et al., 2006). Therefore, the first step would be to identify the favorable flavor and texture characteristics that drive Chinese consumer preference for cheese. Sensory evaluation is one means of measuring and quantifying the relationship between the sensory characteristics of food and its consumer preferences. Techniques such as descriptive sensory analysis, cluster analysis and external preference mapping have been applied to cheese study. Descriptive sensory analysis is a research tool to characterize the aromas and flavors in cheese (Drake, McIngvale, Gerard, Cadwallader, & Civile, 2001). This procedure requires a trained panel to evaluate cheeses and a sensory lexicon to describe their sensory attributes. Cluster analysis techniques can be applied to the analysis of consumer preference data to identify clusters of consumers with similar preference patterns within the sampled population (Pagliarini, Monteleone, & Wakeling, 1997). External preference mapping relates the descriptive sensory data with the consumer preference data and the resulting model helps identify the sensory characteristics required for consumer acceptance or preference (McEwan, 1996; McEwan, Earthy, & Ducher, 1998; Schlich, 1995). External preference mapping has been widely applied to determine consumer preference for cheese sensory characteristics (Caspia, Coggins, Schilling, Yoon, & White,
102
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
2006; Murray & Delahunty, 2000; Young, Drake, Lopetcharat, & McDaniel, 2004). In this study, Beijing youth’s preference for seven imported cheeses was assessed using the preference mapping technique. We chose the youth as the target group since they are most open to new ideas including food. The objectives of this study were: (1) to evaluate the sensory profiles of seven imported cheeses; (2) to define Beijing youth preferences for these cheeses; (3) to characterize the clusters according to socioeconomic data; and (4) to establish the relationships between sensory attributes and Chinese consumer preferences. 2. Materials and methods 2.1. Cheese samples Seven cheese varieties (six natural and one processed, appropriate 10 kg of each) were purchased from local wholesale or retail shops in Beijing, China. These cheeses were selected to represent the majority of cheese types currently available on the Beijing market. The cheese origin and the main characteristics are summarized in Table 1. All samples were stored under refrigeration conditions and placed at room temperature for 1 h before evaluation. Both sensory analysis by the trained panelists and consumer evaluation by the youth were conducted within 1 month after the cheeses were purchased. 2.2. Descriptive sensory analysis A portion of all cheese purchased were repacked and shipped to the US. Descriptive sensory assessment was carried out in Kraft Foods Limit (Glenview, IL, USA). Cheese flavors were evaluated using a 0- to 15-point SpectrumTM universal intensity scale while cheese texture attributes were scaled using a 0- to 15-point product specific intensity scale (Meilgaard, Civile, & Carr, 2007). A trained panel (n = 7) (6 females/1 male) with over 150 h of experience each with descriptive analysis of cheese sensory participated in the assessment. During a preliminary session the objectives of the experiment were explained in details. Then the panelists attended six 1.5 h training sessions during a one-week period prior to sample evaluation sessions. During this phase, panelists reviewed all samples, developed 24 descriptors for flavor and texture, and defined the intense of the descriptors (Table 2). Many descriptors in the vocabulary are found in literatures (Drake et al., 2001; Heisserer & Chambers, 1993; Meilgaard et al., 2007; Yates & Drake, 2007). However, some new descriptors specific to these seven cheeses were defined. Each descriptor was extensively described and explained to avoid any uncertainty about the relevant meaning. For each attribute, panelists also chose and identified references, i.e., ‘‘any chemical, spice or product which characterizes an attribute or attribute intensity”. Once the panel agreed on descriptors and the intensity, evaluation sessions were con-
Table 1 Cheese samples under investigations. Code
Cheesea
Cheese type
Origin
A B C D E F G
Brick Cottage Cream Havarti Cheddar Mozzarella Processed Cheddar
Semi-hard, brick-shaped cheese Low fat soft cheese Soft cheese Creamy semi-hard cheese with eyes Hard cheese labeled by mild flavor Semi-hard cheese Processed cheese
USA USA USA Denmark USA USA Australia
a Processed Cheddar cheese was a processed cheese, while all other six cheeses were natural cheeses.
ducted individually on paper ballots in separate booths dedicated to sensory analysis and free from external aromas, noise and distractions. On the day of assessment the outer layer (5 mm) of each cheese was discarded. A portion (5 g) of Cream and Cottage cheese was placed into 58 ml soufflé cups with lids, while the other five varieties of cheese were cut into 5 g cubes and placed into 58 ml soufflé cups. Each sample was coded with a randomly selected 3 digit number and the order of presentation was balanced to account for first order and carry-over effects (MacFie, Bratchell, Greenhoff, & Vallis, 1989). Flavor and texture evaluations were in separate sessions. Basic taste solutions and previously defined flavor and texture references were available during all flavor and texture evaluations. Panelists evaluated duplicates of each sample in eight evaluation sessions (three or four samples per session). They were instructed to expectorate samples after evaluation. Water (Evian, Danone, France), unsalted crackers (Unsalted Tops Premium Saltine Crackers, Nabisco, Parsippany, NJ, USA) and expectorant cups were provided for rinsing their palates between samples. 2.3. Consumer assessment Consumer test was performed by 217 youth who attended middle school, high school, or university) in Beijing, aged from 12 to 25. They received a consent form which included a clear description of this study. Only those who returned a consent form participated in the study. This consumer test was carried out at the sensory laboratory of China Agricultural University, Beijing. The consumers arrived in groups of 20. The seven cheeses were evaluated across 2 days for each group. A constant control cheese was presented with other three cheeses each day, to reduce the testing bias. The control cheese was selected from the descriptive analysis results based on its moderate sensory intensities. The outer layer of each cheese was removed. Samples were cut into 5 g cubes, numbered with 3digit codes, and served at room temperature. Consumers were allowed to swallow or expectorate the samples as they desired. Retasting was also allowed and each consumer proceeded at his or her own pace. Cheeses were presented individually in a randomized balanced order following presentation of the constant control cheese. The consumers were given two sets of questionnaires. The first questionnaire asked the consumer to score the cheese samples for the overall liking on a nine-point hedonic scale, where 1 = dislike extremely, and 9 = like extremely (Peryam & Pilgrim, 1957). The other questionnaire was designed to collect demographic information. 2.4. Statistical analysis Descriptive and consumer data were analyzed separately and then compositively. Descriptive data were analyzed with SPSS 16.0 for Windows (SPSS Inc, Chicago, IL, USA) using One-Way ANOVA and the post hoc tests were performed using Duncan’s Multiple Range test. Descriptive data were standardized and then analyzed by Principal Component Analysis (PCA) using Unscrambler (v9.7 CAMO, AS, Norway). A hierarchical clustering using Ward’s method was performed with inspection of the dissimilarity plot and dendrogram, which showed that a five-cluster solution would be optimal. After having identified the groups, the consumers were segmented by the kmeans clustering method. This method divides consumers into K clusters so that the within-cluster sum of squares is minimized. Chi-squared tests and Fisher’s exact tests were conducted to compare consumer demographic data from each segment. These analyses were carried out using SPSS (SPSS 16.0, Chicago, IL, USA). One-Way ANOVA using Duncan’s Multiple Range test was also applied to the consumer segments’ preference scores to
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
103
Table 2 Descriptive vocabulary used to characterize cheese flavor. Descriptors
Definition with reference
Reference and intensity Skim milk heated to 85 °C for 30 min Diacetyl, 20 ppm (prepared in 95% ethanol) Dillon’s whole milk Dannone’s plain yoghourt Land O Lakes unsalted butter Butyric acid, 20 ppm (prepared in 95% ethanol) Kretschmer wheat germ Eatable plastic package film Foods left in refrigerator for more than one week Raw button mushroom Toasted sliced break
Umamia Bittera
Aromatics associated with cooked milk Aromatics associated with diacetyl Aromatics associated with milk fat Sour aromatics associated with fermented dairy products Aromatics associated with natural, fresh, slightly salted butter Aromatics associated with short chain fatty acid Sweet roasted aromatic associated with various nuts Aromas associated with plastic wrap packaging material Stale aromatic characteristic of refrigerator with old food left in it Aromas associated with freshly turned soil Aromatics associated with a grain which has been sufficiently heated to caramelize some of the starches and sugars Chemical feeling factor elicited by certain peptides and nucleotides Basic taste sensation generated by caffeine
Saltya
Basic taste sensation generated by salts
Soura
Basic taste sensation generated by acids
a
Milky Diacetyla Milk fata Soured milke Butteryb Free fatty acida Nuttya Plastica Stalee Earthye Toastedc
Sweeta Astringent Slimy
Basic taste sensation generated by sugars a
e
Moistd Firmnessd
Stickyd
Dry, puckery mouthfeel generated by certain peptides or nucleotides Of the nature of slime, soft, glutinous or viscous substance, soft, moist and sticky The perceived moisture content of the cheese. Ranging from dry to moist. The amount of force required to completely bite through the sample/The extent to resistance offered by the cheese, probably judged in the first half of chewing using the front teeth The degree to which the chewed sample sticks to the surfaces of the mouth and teeth
Smoothd
The smoothness of the cheese against the palate as it breaks up during mastication
Grainyd
The extent to which granular structures are formed as the sample breaks down Evaluate how much the sample has broken down during mastication/The amount of breakdown that occurs in the sample as a result of mastication
Rate of breakdownd
1% MSG solution = 7.0 0.01% caffeine solution = 0.5 0.05% caffeine solution = 2 0.35% NaCl solution = 5.0 0.5% NaCl solution = 7.5 0.8% NaCl solution = 9.0 0.015% Citric acid solution = 1.5 0.025% Citric acid solution = 2.5 0.035% Citric acid solution = 3.5 2% sucrose solution = 1 5% sucrose solution = 2.5 0.05% alum solution = 1 0.15% alum solution = 2.5 Processed cheese = 1 Parmesan = 1 Processed cheese = 5 Processed cheese = 2 Sharp Cheddar = 9 Parmesan = 14 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 1 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 14 Sharp Cheddar = 1 Parmesan = 5 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 14
Processed cheese: Velveeta brand 907-g loaf (Kraft Foods North America; Glenview, IL). Sharp Cheddar: Kraft brand 227-g block (Kraft Foods, Glenview, IL). Parmesan: Kraft brand 113-g triangle (Kraft Foods, Glenview, IL). a Attribute found in Drake et al. (2001). b Attribute found in Heisserer and Chambers (1993). c Attribute found in Meilgaard et al. (2007). d Attribute found inYates and Drake (2007). e Not found in the literatures above.
establish whether they had significantly different preference for the cheeses. External preference mapping was conducted using Partial Least Squares regression (PLS) (Martens & Martens, 1986). The sensory data were used as the explanatory variables (X-matrix) and the average preference data within each cluster as the response data (Y-matrix). Unscrambler (v9.7 CAMO, AS, Norway) was used to carry out PLS and to determine the relationships between the two independent data sets. 3. Results and discussions 3.1. Descriptive sensory analysis The panels mean scores and ANOVA from the descriptive sensory data are shown in Table 3. Significant differences (P < 0.05) were found in 19 of the 24 sensory attributes. These included the flavor attributes ‘‘milky”, ‘‘milk fat”, ‘‘soured milk”, ‘‘free fatty acid”,
‘‘nutty”, ‘‘plastic”, ‘‘stale”, ‘‘earthy”, ‘‘toasted”, ‘‘umami”, ‘‘bitter”, ‘‘salty” and ‘‘sour”, and the texture attributes ‘‘slimy’, ‘‘moist”, ‘‘firmness”, ‘‘smooth”, ‘‘grainy’ and ‘‘rate of breakdown”. However, ANOVA failed to show how these descriptors interacted with the overall sensory characters and the extent of difference between the cheeses. PCA was used to analyze the descriptive sensory data. This analysis significantly discriminated between the cheeses (P < 0.05) on the first three PCs which accounted for 44%, 29% and 11% of the experimental variance, respectively (Figs. 1 and 2). As seen from Fig. 1, Brick cheese was characterized by ‘‘plastic”, ‘‘stale” and ‘‘earthy” flavors. Brick cheese was negatively correlated with Cream cheese indicating that they were quite different in sensory character. Cream cheese was ‘‘slimy” in texture and ‘‘milky” and ‘‘milk fat” in flavor which were defined as young/fresh cheese flavor (Van Hekken et al., 2006). Cottage cheese was in the quarter between positive PC1 and negative PC2 characterized by ‘‘soured milk” and ‘‘grainy”. Processed Cheddar cheese was positioned along the negative PC1 and was described as having a ‘‘salty”, ‘‘nutty” and
104
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Table 3 Descriptive attribute intensity means for the seven cheeses varieties. Attributes
Significant
Brick
Cottage
Cream
Havarti
Cheddar
Mozzarella
Process
Duncan’s Critical value
Milky Diacetyl Milk fat Soured milk Buttery Free fatty acid Nutty Plastic Stale Earthy Toasted Umami Bitter Salty Sour Sweet Astringent Slimy Moist Firmness Stick Smooth Grainy Rate of breakdown
**
2.5 1.04 0.93 0.14 2.25 2.54 0.21 1.39 0.75 1.25 1.07 1.11 0.39 7.57 1.32 0.14 0.57 0 4.71 5.46 1.64 6.11 0.5 7.61
3.64 0.57 1.82 2 2.86 1.89 0 0.36 0.36 1 0.46 0.18 0 6.14 1.64 0.18 0.68 0.36 9.5 3.71 0.86 1.71 3.64 9.93
4 0.43 3.07 0.5 3.5 2.36 0 0.07 0.11 0.61 1.14 0.39 0 6.71 1.54 0.21 0.32 1.61 7.86 3.54 2.36 11.21 0.04 9.5
3.18 0.36 2.46 0.36 3.36 3.5 0.39 0.43 0.21 1.14 1 1.54 0.07 7.36 1.82 0.14 0.5 0.54 7.29 4.82 1.75 8.29 0 8.21
3.39 0.21 2.79 0.46 3.36 4.11 0.39 0.36 0.18 0.64 1.71 1.79 0.21 7.64 1.54 0.14 0.57 0 4.79 6.57 1.93 6.86 0 7.39
3.18 0.54 2.07 0.25 2.89 2.46 0.54 0.86 0.21 0.96 0.96 1.04 0 7.14 1.32 0.14 0.46 0 5.21 5.64 1.54 6.96 0.07 8.07
2.64 1.21 1.89 0.21 3.21 1.96 0.79 1.07 0.5 1.04 2.04 1.96 0 7.93 1.39 0.14 0.5 0 3.75 8 2.04 5.21 0.46 6.93
6.9 2.1 3.0 29.3 1.9 15.6 3.0 2.2 2.3 2.3 6.9 12.5 13.6 3.0 2.4 0.4 1.3 32.3 28.1 17.2 2.2 26.9 29.2 5.9
ns ** **
ns ** * * * * ** ** ** * *
ns ns ** ** **
ns ** ** **
Flavor attributes were scored using a 0- to 15-point universal SpectrumTM intensity scale where 0 = absence and 15 = high intensity of an attribute. Texture attributes were scored using a 0- to 15-point product-specific scale where 0 = absence and 15 = high intensity of that attribute within that product category. ns = Not significant. * P < 0.05. ** P < 0.01.
‘‘umami” flavors and ‘‘firmness” texture. Cheddar and Havarti cheeses were similar in terms of their ‘‘free fatty acid” flavor, however, Cheddar cheese was more defined in this characteristic. These two descriptors ‘‘nutty” and ‘‘free fatty acid” were always detected in many aged/sharp cheeses (Drake et al., 2001; Heisserer & Chambers, 1993). This indicated the three cheeses (Processed Cheddar, Cheddar and Havarti) were characterized by higher intensities of aged cheese flavors than what was found in Cream and Cottage cheese. The flavor of Mozzarella cheese was positioned toward the center of the graph, showing that the sensory character of this cheese was not well explained by the first two PCs. As seen from Table 3, the intensity of all attributions in this cheese was neither too strong nor too mild. Mozzarella cheese was also chosen as the control cheese in the consumer assessment due to its mild sensory profile. The third PC of the analysis (Fig. 2) tended to distinguish cheeses on the basis of more subtle sensory characteristics. Cheddar and Havarti cheeses were also ‘‘sour” in flavor. 3.2. Consumer assessment Demographic information and consumption characteristics of 217 consumers are presented in Table 4. Of these consumers, 48% were male and 52% were female. Only 7% had been aboard. 36%, 29%, and 35% of the youth were between 12–15, 15–20, and 20–25 years old, respectively. Regarding the consumption frequency, it ranged mainly from once a week to once a month, which counted 65% of total. By contrast, a survey of 3538 American subjects about their frequency of cheese consumption indicated that 23% subjects consumed cheese nearly every day, 38% of them ate it several times a week, 15% subjects consumed it once a week, and 23% of them consumed it less than once a week (Dairy Management Inc., 1999). Thus, Chinese youth in this study consumed cheese infrequently as compared to the Americans. Till date, cheese still is not one of the favorite dairy products for most Chinese people.
Information regarding cheese-containing food was explored (data not showed). 201 youth (93%) had the experience of having hamburgers, 154 having pizza (71%) and 87 having sandwich (40%). Only 24 subjects (11%) had consumed cheese alone. It was probably that Chinese people were used to consuming cheese with other types of food as an ingredient and were not aware that cheese could be eaten alone. Moreover, eating cheese alone was thought to be not enticing. 16 subjects (7.4%) indicated that they had consumed other cheese-containing food such as Italian cuisine, cheesecake, Au gratin foods and bagel. Overall, hamburger and pizza were the most popular cheese-containing food among Chinese consumers. Now it is very common that people can easily find these cheese-containing foods at western fast-food chains (e.g., McDonald’s, KFC, Pizza Hut, Papa John’s, etc.). The mean scores of consumer preference for these cheeses showed that Cream and Cottage cheeses were the most preferred (mean scores of 7.0 and 6.5, respectively), while Havarti cheese was the least liked (mean score of 5.23). The differences in liking scale were more prominent between the clusters identified by the k-means clustering method. Cluster analysis identified five segments of consumers with different preferences with the consumer sample (Fig. 3). The demographic characteristics of these consumer segments are displayed in Table 4. Cluster 1 (40 consumers) would accept all the cheeses except Havarti, and appreciated Cottage, Cream and Processed Cheddar cheeses more. 73% consumers within this cluster claimed they had consumed cheese-containing food at least once a week. This might be due to their relatively more foreign experience (35%). Cluster 2, including the fewest consumers (22 consumers), scored all cheeses the lowest and disliked Cheddar cheese the most. This segment consisted mainly of consumers between 20 and 25 years old (73%). In addition, half of these consumers had never consumed cheese food before (45%). Presumably, consumers in this group might be greatly influenced by the Chinese traditional culture, and this cluster would not be easily pleased. Cluster 3 (40 consumers) liked Cream and Cottage cheeses
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
105
Fig. 1. Scores and loadings for PCA of the results of descriptive sensory analysis.
Fig. 2. Scores and loadings for PCA of the results of descriptive sensory analysis.
and showed less interest in Brick, Havarti, Cheddar and Mozzarella cheeses. Consumers in Cluster 4 (25 consumers) scored all cheeses lowly, with the exception of Cream cheese. Consumers in Cluster 5 (90 consumers) scored all cheeses highly and expressed more appreciation for Cottage cheese. This group had the highest proportion of the 12–15 years old (51%). This finding suggested that perhaps younger teenagers had a higher acceptance of new foods and
perceived that food products from foreign countries had better quality. Nicklaus, Boggio, Chabanet, and Issanchou (2004) also pointed that younger consumers were most open to the new type of food-cheese and really liked all the tested cheeses. Cluster analysis in our study allowed for a more detailed insight into consumer likes and illustrated the diversity of preference across the consumer testers. Pagliarini et al. (1997)
106
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Table 4 Number and percentage of consumers within each segment identified by HCA. Cluster
*
All
1
2
3
4
5
Sex Male Female
25(63%) 15(37%)
11(50%) 11(50%)
18(45%) 22(55%)
10(40%) 15(60%)
40(44%) 50(56%)
104(48%) 113(52%)
Age class* 12–15 15–20 20–25
15(37%) 9(23%) 16(40%)
4(18%) 2(9%) 16(73%)
5(13%) 18(45%) 17(42%)
8(32%) 11(44%) 6(24%)
46(51%) 23(26%) 21(23%)
78(36%) 63(29%) 76(35%)
Foreign experience* With Without
14(35%) 26(65%)
0(0%) 22(100%)
0(0%) 40(100%)
0(0%) 25(100%)
2(2%) 88(98%)
16(7%) 201(93%)
Cheese consumption* Never Once per year Once per month Once per week Several times a week Total
0(0%) 3(8%) 8(20%) 17(43%) 12(30%) 40
10(45%) 9(41%) 2(9%) 1(5%) 0(0%) 22
1(3%) 5(13%) 22(55%) 10(25%) 2(5%) 40
5(20%) 12(48%) 6(24%) 2(8%) 0(0%) 25
2(2%) 15(17%) 44(49%) 28(31%) 1(1%) 90
18(8%) 44(20%) 82(38%) 58(27%) 15(7%) 217
Indicates different between the 5 clusters (P < 0.05).
Fig. 3. Mean preference scores for cheeses within clusters. Preference was scored on a 9-point hedonic scale where 1 = dislike extremely and 9 = like extremely. (j Brick h Cottage Cream Havarti Cheddar Mozzarella Process).
identified six clusters for three different types of Mozzarella cheese and concluded that cluster analysis resulted in a deeper understanding of consumer liking for these cheeses. Additionally, these differences in preferences among the clusters in our study indicated that cheeses with specific flavor profiles could be mar-
keted to specific target consumers. Murray and Delahunty (2000) carried out a consumer questionnaire and came to similar conclusions when demographic and sociological variables were related to a final four cluster solution. It was also agreed that socioeconomic data had influenced consumer preference.
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
3.3. External preference mapping For a full interpretation of relationship between descriptive data and consumer evaluation, external preference mapping was conducted using the average preference scores of each cluster. The main position of preferences of the five clusters is distinct in Fig. 4, and the first two PC’s explained 60% of the liking variation and 69% of the sensory variation. As seen from the graph of validation variance of the Y-variance (Fig. 5), the first two PCs reduced the error variance and then the third PC went up. Therefore, we commented the data based on the first two PCs in our study. On the basis of overall liking of cheeses, most consumers were apt to accept milk-like and yoghurt-like sensory attributes, such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures. Meanwhile, most of them could not readily accept ‘‘bitter”, ‘‘salty”, ‘‘umami” and ‘‘free fatty acid” flavors and ‘‘firmness” texture in cheese. Cream and Cottage cheese, the two milk-like and yoghurt-like cheeses, received the highest scores of the seven cheeses in our study. Many Chinese cheese researchers indicated Cream and Cottage would be most acceptable for Chinese consumers, but there was no exact data in their literatures (Bo & Zhou, 2009; Ma, Gong, Wu, & Liu, 2006). In addition, they reported another popular cheese – Chinese Royal Cheese. Chinese Royal Cheese, also called Gua-nai, is an oriental-type dairy product coagulated with glutinous rice wine (Onyeneho, Partridge, Brunner, & Guan, 1987). It is characterized by sweet, fruity, and slightly alcoholic flavor and soft curd appearance (Jiang, Chen, Xue, & Chen, 2007). A second observation from Fig. 4 is that cluster 5 which was closest to the ‘‘all” cluster. Drivers of liking for this group were ‘‘sour” and ‘‘milky” flavors, and ‘‘slimy” texture. Disliking was driven by ‘‘stale”, ‘‘plastic” and ‘‘earthy” flavors. Consumers in this cluster gave high and similar scores for all the seven cheeses. It could be assumed that all the products evaluated within the present study would be appreciated by this group. Therefore, this group could be considered as the ‘‘positive consumers” which means liking all the cheeses. Cluster 2 is very different from all the other groups. It was located along the positive PC2 axis. The attributes
107
that significantly influenced dislike were ‘‘free fatty acid” and ‘‘milk fat” flavors, and ‘‘smooth” texture. There was no significant preference among consumers in this group, as these consumers scored all cheeses lowly. Taking into account the low preference for cheese, we tentatively considered Cluster 2 as ‘‘negative consumers”. For Cluster 1, ‘‘grainy” texture and ‘‘soured milk” flavor were found to be drivers of liking, whereas ‘‘free fatty acid” flavor was driver of disliking. Cluster 1 scored all cheeses highly, with the exception of Havarti and Cheddar cheeses. These two cheeses were characterized with ‘‘free fatty acid” which is considered as aged cheese flavor. Another typical aged cheese flavor is ‘‘nutty” which is rich in Processed Cheddar cheese. However, consumers in this cluster scored Processed Cheddar cheese highly. This indicated that consumers in cluster 1 did not dislike all the aged cheese flavors, as some of them liked ‘‘nutty” flavor but disliked ‘‘free fatty acid” flavor. Therefore, we considered consumers in Cluster 1 as ‘‘free fatty acid dislikers”. Consumers in this cluster consumed cheese much more frequently, which indicating that some cheese attributes such as ‘‘nutty” could be cultured among Chinese consumers. Cluster 3 appeared to be positively influenced by similar sensory characteristics as Cluster 1, liking ‘‘grainy” texture and ‘‘soured milk” flavor and disliking ‘‘free fatty acid” flavor. However, Cluster 3 discriminated much more between the cheeses, as disliking was also driven by ‘‘umami” and ‘‘salty” flavors and ‘‘firmness” texture. Consumers in this group showed a strong appreciation for soft cheese (Cream and Cottage cheese). Thus this cluster could be considered as ‘‘soft texture lovers”. Cream Cheese particularly appealed to the consumers in Cluster 4, which could be explained by its ‘‘smooth” texture and ‘‘milk fat” flavor. The consumers within this group could be considered as ‘‘Cream cheese lovers”. Previous research has addressed cheese flavor and consumer testing methods. Lawlor and Delahunty (2000) conducted preference mapping and cluster analysis with 10 specialty cheeses in Ireland. Their results indicated diverse flavor differences among cheese varieties and wide consumer preferences for these cheeses. Most recently, Liggett, Drake, and Delwiche (2008) investigated the flavor attributes that drove consumer preference for Swiss cheese.
Fig. 4. External preference map of combined consumer data with descriptive analysis results.
108
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Fig. 5. Residual validation variance of the Y-variance of External preference map.
Diacetyl, whey, milk fat, and umami were found to be drivers of liking, whereas cabbage, cooked, and vinegar were drivers of disliking. Our study indicated diverse flavor differences among the seven cheese varieties but a relative homogenous preference of Beijing youth for cheese flavor. Milk-like and yoghurt-like sensory attributes, such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures were the favorable characteristics of cheese for most Chinese consumers. 4. Conclusions Descriptive analysis and consumer evaluation showed distinctive flavor profiles and their relationship with acceptability of the seven cheese varieties. 19 of the 24 descriptors were found significantly different among these cheeses. As for cheese consumption behavior, Chinese youth in this study consumed cheese infrequently as compared to the Americans, and hamburger was the most frequently consumed cheese-containing food. The Chinese, once thought to dislike cheese of all kinds, have adjusted their lifestyles to a variety of cheese and products. The potential cheese market in China is enormous and may be larger than once thought possible. Cluster analysis identified five clusters of consumers and greatly enhanced the understanding of consumer assessment. Cheeses with specific sensory profiles could be marketed to specific target consumers. External preference mapping revealed correlations between descriptive and preference data and identified the sensory characteristics of cheeses preferred by the consumer clusters. Overall, most Chinese youth preferred cheeses with mild sensory attributes such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” texture. In addition, ‘‘nutty” as a typical flavor in aged cheese was desirable to some consumer cluster and could be cultivated among Chinese consumers. Acknowledgments This project was financially supported by the Mega-projects of Science Research for the 11th Five-Year Plan (Item No. 2006BAD04A06) of Ministry of Science and Technology of China and National Key Technologies R & D Program of China (Item No. 2006BAD04A13).The authors thank members of the sensory panel
in Kraft Foods Limit (Glenview, IL, USA) and the consumer test members in Beijing for their participation in this study. The authors acknowledge Limin Zheng and Luda Zhang for their assistance in statistical analyses.
References Bo, S., & Zhou, G. H. (2009). The current situation and countermeasures of Cheese in China. Academic Periodical of Farm Products Processing, 181, 73–75 (In Chinese). Caspia, E. L., Coggins, P. C., Schilling, M. W., Yoon, Y., & White, C. H. (2006). The relationship between consumer acceptability and descriptive sensory attributes in Cheddar cheese. Journal of Sensory Studies, 21, 112–127. Dairy Association of China (2007). DAC statistical analysis Data.. Dairy Management Inc. (1999). 1998 attitudes and usage trend study (pp. 180). Rosemont, IL: Dairy Management Inc. Drake, M. A., McIngvale, S., Gerard, P. D., Cadwallader, K. R., & Civile, G. V. (2001). Development of a descriptive language for cheddar cheese. Journal of Food Science, 66, 1422–1427. Gee, P., & Townson, P. (2007). Cheese if you please: Getting the Chinese market to consume cheese is a challenge in a lactose-intolerant population. Dairy Industries International, 72, 18–21. Heisserer, D. M., & Chambers, E. I. (1993). Determination of the sensory flavor attributes of aged natural cheese. Journal of Sensory Studies, 8, 121–132. International Dairy Federation (2007). Bulletin of the IDF No. 423/2007- The World Dairy Situation 2007- E-Form. Jiang, T., Chen, L. J., Xue, L., & Chen, L. S. (2007). Study on milk-clotting mechanism of rennet-like enzyme from glutinous rice wine: Proteolytic property and the cleavage site on k-casein. Journal of Dairy Science, 90, 3126–3133. Lawlor, J. B., & Delahunty, C. M. (2000). The sensory profile and consumer preference for ten speciality cheeses. International Journal of Dairy Technology, 53, 28–36. Liggett, R. E., Drake, M. A., & Delwiche, J. F. (2008). Impact of flavor attributes on consumer liking of Swiss cheese. Journal of Dairy Science, 91, 466–476. Ma, Z. D., Gong, X. Q., Wu, Z. X., & Liu, H. P. (2006). Development advantage of cottage cheese in China. Food Science and Technology, 12, 66–69 (In Chinese). MacFie, H. J., Bratchell, N., Greenhoff, K., & Vallis, I. V. (1989). Designs to balance the effect of order of presentation and first-order carry-over effects in hall tests. Journal of Sensory Studies, 4, 129–148. Martens, M., & Martens, H. (1986). Partial least squares regression. In J. R. Piggott (Ed.), Statistical procedures in food research (pp. 293–359). London/New York: Elsevier. McEwan, J. A. (1996). Preference mapping for product optimisation. In T. Naes & E. Risvik (Eds.), Multivariate analysis of data in sensory science (pp. 71–80). Amsterdam: Elsevier Science. McEwan, J. A., Earthy, P. J., & Ducher, C. (1998). Preference mapping: a review (pp. 16). No. 6, Project No. 29742, Campden and Choreleywood, UK: Food Research Association. Meilgaard, M., Civile, G. V., & Carr, T. (2007). The SpectrumTM descriptive analysis method. In 4th Sensory evaluation techniques (pp. 189–194). New York: CRC Press.
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109 Murray, J. M., & Delahunty, C. M. (2000). Consumer preference for Irish farmhouse and factory cheeses. Irish Journal of Agricultural and Food Research, 39, 433–449. Nicklaus, S., Boggio, V., Chabanet, C., & Issanchou, S. (2004). A prospective study of food preferences in childhood. Food Quality and Preference, 15, 805–818. Onyeneho, S. N., Partridge, J. A., Brunner, J. R., & Guan, J. (1987). Manufacture and characterization of Gua-nai: A new dairy food produced with an oriental-type culture. Journal of Dairy Science., 70, 2499–2503. Pagliarini, E., Monteleone, E., & Wakeling, I. (1997). Sensory profile description of Mozzarella cheese and its relationship with consumer preference. Journal of Sensory Studies, 12, 285–301. Peryam, D. R., & Pilgrim, F. J. (1957). Hedonic scale method of measuring food preferences. Food Technology, 11, 9–14.
109
Schlich, P. (1995). Preference mapping: Relation consumer preferences to sensory or instrumental measurements. In P. Etivant & P. Schlich (Eds.), Bioflavor (pp. 231–245). INRA Editions: Versailles, France. Van Hekken, D. L., Drake, M. A., Molina Corral, F. J., Guerrero Prieto, V. M., & Gardea, A. A. (2006). Mexican Chihuahua cheese: Sensory profiles of young cheese. Journal of Dairy Science, 89(10), 3729–3738. Yang, J. (2007). The development prospects for China dairy industry from the increase of dairy consumption perspective. China Dairy, 3, 11–13 (In Chinese). Yates, M. D., & Drake, M. A. (2007). Texture properties of Gouda cheese. Journal of Sensory Studies, 22(5), 493–506. Young, N. D., Drake, M., Lopetcharat, K., & McDaniel, M. R. (2004). Preference mapping of Cheddar cheese with varying maturity levels. Journal of Dairy Science, 87, 11–19.
Contents lists available at ScienceDirect
Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual
Sensory profile and Beijing youth preference of seven cheese varieties X.Y. Zhang a,b, H.Y. Guo a,b, L. Zhao a,b, W.F. Sun c, S.S. Zeng d, X.M. Lu a,b, X. Cao a,b, F.Z. Ren a,b,⇑ a
College of Food Science & Nutritional Engineering, China Agricultural University, Beijing 100083, China Key Laboratory of Functional Dairy, Ministry of Education, Beijing 100083, China c College of Food Science, Xinjiang Agricultural University, Urumuchi 830052, China d Department of Agriculture & Natural Resources, Langston University, P.O. Box 1730, Langston, OK 73050, United States b
a r t i c l e
i n f o
Article history: Received 3 December 2008 Received in revised form 21 April 2010 Accepted 9 August 2010 Available online 13 August 2010 Keywords: Cheese flavor Descriptive sensory analysis External preference mapping Cluster analysis China
a b s t r a c t The sensory characteristics that determined consumer preference of seven imported cheeses were investigated. Descriptive sensory analysis was performed by seven trained panelists who used 24 descriptors to quantitatively describe the sensory characteristics of these cheeses. Principal component analysis of descriptive data found significant differences between the cheeses (P < 0.05) on the first three principal components which accounted for 44%, 29% and 11% of the experimental variance, respectively. In parallel, the preference of 217 Beijing youth on these cheeses was expressed on a nine-point hedonic scale. Cluster analysis of the preference data identified five consumer clusters with different preferences and demonstrated that there existed a potential market for each of these cheeses. External preference mapping indicated a relative homogenous preference of Beijing youth for cheese flavor. ‘‘Soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures appeared to be drivers of liking for most Chinese youth, whereas ‘‘bitter”, ‘‘salty”, ‘‘umami” and ‘‘free fatty acid” flavors and ‘‘firmness” texture were drivers of disliking. In addition, ‘‘nutty” as a typical aged flavor in cheese was desirable to some consumer cluster and could be cultivated among Chinese. Ó 2010 Published by Elsevier Ltd.
1. Introduction Over the last two decades, the demand for dairy products in China has increased drastically (Yang, 2007). With the improvement of people’s living standards, the structure of dairy products has also changed. Chinese have mainly consumed fresh milk and yogurt. Nowadays, more attention has been paid to non-traditional dairy products such as probiotic beverage and cheese. Cheese is a relatively new type of food in the traditional Chinese diet. In 2006 the cheese consumption in China was 20,000 tons, of which 90% were imported (Gee & Townson, 2007). Although the cheese import has increased markedly in recent years (1320 tons annually 2000–2006) (Dairy Association of China, 2007), the Chinese cheese market is still at its initial stage. The current Chinese cheese consumption is 200 g per capita annually according to Dairy Association of China (2007). In comparison, the annual per capita consumption is 16 kg in the United States and 24 kg in France (International Dairy Federation, 2007). Diet cultures and preference of ethnic Chinese are significantly different from those of Western people. To date, there is no data available regarding the preference of Chinese consumers on ⇑ Corresponding author. Address: College of Food Science & Nutritional Engineering, China Agricultural University, P.O. Box 303, No. 17 Qinghua East Road, Haidian, Beijing 100083, China. Tel./fax: +86 10 62736344. E-mail address: [email protected] (F.Z. Ren). 0950-3293/$ - see front matter Ó 2010 Published by Elsevier Ltd. doi:10.1016/j.foodqual.2010.08.007
cheese. Due to the great variability of cheese varieties and flavors, an extensive and systematic study would be important to the development of a cheese industry in China. Sensory attributes (flavor and texture) are critical to the identity of cheese and consumer acceptance (Van Hekken et al., 2006). Therefore, the first step would be to identify the favorable flavor and texture characteristics that drive Chinese consumer preference for cheese. Sensory evaluation is one means of measuring and quantifying the relationship between the sensory characteristics of food and its consumer preferences. Techniques such as descriptive sensory analysis, cluster analysis and external preference mapping have been applied to cheese study. Descriptive sensory analysis is a research tool to characterize the aromas and flavors in cheese (Drake, McIngvale, Gerard, Cadwallader, & Civile, 2001). This procedure requires a trained panel to evaluate cheeses and a sensory lexicon to describe their sensory attributes. Cluster analysis techniques can be applied to the analysis of consumer preference data to identify clusters of consumers with similar preference patterns within the sampled population (Pagliarini, Monteleone, & Wakeling, 1997). External preference mapping relates the descriptive sensory data with the consumer preference data and the resulting model helps identify the sensory characteristics required for consumer acceptance or preference (McEwan, 1996; McEwan, Earthy, & Ducher, 1998; Schlich, 1995). External preference mapping has been widely applied to determine consumer preference for cheese sensory characteristics (Caspia, Coggins, Schilling, Yoon, & White,
102
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
2006; Murray & Delahunty, 2000; Young, Drake, Lopetcharat, & McDaniel, 2004). In this study, Beijing youth’s preference for seven imported cheeses was assessed using the preference mapping technique. We chose the youth as the target group since they are most open to new ideas including food. The objectives of this study were: (1) to evaluate the sensory profiles of seven imported cheeses; (2) to define Beijing youth preferences for these cheeses; (3) to characterize the clusters according to socioeconomic data; and (4) to establish the relationships between sensory attributes and Chinese consumer preferences. 2. Materials and methods 2.1. Cheese samples Seven cheese varieties (six natural and one processed, appropriate 10 kg of each) were purchased from local wholesale or retail shops in Beijing, China. These cheeses were selected to represent the majority of cheese types currently available on the Beijing market. The cheese origin and the main characteristics are summarized in Table 1. All samples were stored under refrigeration conditions and placed at room temperature for 1 h before evaluation. Both sensory analysis by the trained panelists and consumer evaluation by the youth were conducted within 1 month after the cheeses were purchased. 2.2. Descriptive sensory analysis A portion of all cheese purchased were repacked and shipped to the US. Descriptive sensory assessment was carried out in Kraft Foods Limit (Glenview, IL, USA). Cheese flavors were evaluated using a 0- to 15-point SpectrumTM universal intensity scale while cheese texture attributes were scaled using a 0- to 15-point product specific intensity scale (Meilgaard, Civile, & Carr, 2007). A trained panel (n = 7) (6 females/1 male) with over 150 h of experience each with descriptive analysis of cheese sensory participated in the assessment. During a preliminary session the objectives of the experiment were explained in details. Then the panelists attended six 1.5 h training sessions during a one-week period prior to sample evaluation sessions. During this phase, panelists reviewed all samples, developed 24 descriptors for flavor and texture, and defined the intense of the descriptors (Table 2). Many descriptors in the vocabulary are found in literatures (Drake et al., 2001; Heisserer & Chambers, 1993; Meilgaard et al., 2007; Yates & Drake, 2007). However, some new descriptors specific to these seven cheeses were defined. Each descriptor was extensively described and explained to avoid any uncertainty about the relevant meaning. For each attribute, panelists also chose and identified references, i.e., ‘‘any chemical, spice or product which characterizes an attribute or attribute intensity”. Once the panel agreed on descriptors and the intensity, evaluation sessions were con-
Table 1 Cheese samples under investigations. Code
Cheesea
Cheese type
Origin
A B C D E F G
Brick Cottage Cream Havarti Cheddar Mozzarella Processed Cheddar
Semi-hard, brick-shaped cheese Low fat soft cheese Soft cheese Creamy semi-hard cheese with eyes Hard cheese labeled by mild flavor Semi-hard cheese Processed cheese
USA USA USA Denmark USA USA Australia
a Processed Cheddar cheese was a processed cheese, while all other six cheeses were natural cheeses.
ducted individually on paper ballots in separate booths dedicated to sensory analysis and free from external aromas, noise and distractions. On the day of assessment the outer layer (5 mm) of each cheese was discarded. A portion (5 g) of Cream and Cottage cheese was placed into 58 ml soufflé cups with lids, while the other five varieties of cheese were cut into 5 g cubes and placed into 58 ml soufflé cups. Each sample was coded with a randomly selected 3 digit number and the order of presentation was balanced to account for first order and carry-over effects (MacFie, Bratchell, Greenhoff, & Vallis, 1989). Flavor and texture evaluations were in separate sessions. Basic taste solutions and previously defined flavor and texture references were available during all flavor and texture evaluations. Panelists evaluated duplicates of each sample in eight evaluation sessions (three or four samples per session). They were instructed to expectorate samples after evaluation. Water (Evian, Danone, France), unsalted crackers (Unsalted Tops Premium Saltine Crackers, Nabisco, Parsippany, NJ, USA) and expectorant cups were provided for rinsing their palates between samples. 2.3. Consumer assessment Consumer test was performed by 217 youth who attended middle school, high school, or university) in Beijing, aged from 12 to 25. They received a consent form which included a clear description of this study. Only those who returned a consent form participated in the study. This consumer test was carried out at the sensory laboratory of China Agricultural University, Beijing. The consumers arrived in groups of 20. The seven cheeses were evaluated across 2 days for each group. A constant control cheese was presented with other three cheeses each day, to reduce the testing bias. The control cheese was selected from the descriptive analysis results based on its moderate sensory intensities. The outer layer of each cheese was removed. Samples were cut into 5 g cubes, numbered with 3digit codes, and served at room temperature. Consumers were allowed to swallow or expectorate the samples as they desired. Retasting was also allowed and each consumer proceeded at his or her own pace. Cheeses were presented individually in a randomized balanced order following presentation of the constant control cheese. The consumers were given two sets of questionnaires. The first questionnaire asked the consumer to score the cheese samples for the overall liking on a nine-point hedonic scale, where 1 = dislike extremely, and 9 = like extremely (Peryam & Pilgrim, 1957). The other questionnaire was designed to collect demographic information. 2.4. Statistical analysis Descriptive and consumer data were analyzed separately and then compositively. Descriptive data were analyzed with SPSS 16.0 for Windows (SPSS Inc, Chicago, IL, USA) using One-Way ANOVA and the post hoc tests were performed using Duncan’s Multiple Range test. Descriptive data were standardized and then analyzed by Principal Component Analysis (PCA) using Unscrambler (v9.7 CAMO, AS, Norway). A hierarchical clustering using Ward’s method was performed with inspection of the dissimilarity plot and dendrogram, which showed that a five-cluster solution would be optimal. After having identified the groups, the consumers were segmented by the kmeans clustering method. This method divides consumers into K clusters so that the within-cluster sum of squares is minimized. Chi-squared tests and Fisher’s exact tests were conducted to compare consumer demographic data from each segment. These analyses were carried out using SPSS (SPSS 16.0, Chicago, IL, USA). One-Way ANOVA using Duncan’s Multiple Range test was also applied to the consumer segments’ preference scores to
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
103
Table 2 Descriptive vocabulary used to characterize cheese flavor. Descriptors
Definition with reference
Reference and intensity Skim milk heated to 85 °C for 30 min Diacetyl, 20 ppm (prepared in 95% ethanol) Dillon’s whole milk Dannone’s plain yoghourt Land O Lakes unsalted butter Butyric acid, 20 ppm (prepared in 95% ethanol) Kretschmer wheat germ Eatable plastic package film Foods left in refrigerator for more than one week Raw button mushroom Toasted sliced break
Umamia Bittera
Aromatics associated with cooked milk Aromatics associated with diacetyl Aromatics associated with milk fat Sour aromatics associated with fermented dairy products Aromatics associated with natural, fresh, slightly salted butter Aromatics associated with short chain fatty acid Sweet roasted aromatic associated with various nuts Aromas associated with plastic wrap packaging material Stale aromatic characteristic of refrigerator with old food left in it Aromas associated with freshly turned soil Aromatics associated with a grain which has been sufficiently heated to caramelize some of the starches and sugars Chemical feeling factor elicited by certain peptides and nucleotides Basic taste sensation generated by caffeine
Saltya
Basic taste sensation generated by salts
Soura
Basic taste sensation generated by acids
a
Milky Diacetyla Milk fata Soured milke Butteryb Free fatty acida Nuttya Plastica Stalee Earthye Toastedc
Sweeta Astringent Slimy
Basic taste sensation generated by sugars a
e
Moistd Firmnessd
Stickyd
Dry, puckery mouthfeel generated by certain peptides or nucleotides Of the nature of slime, soft, glutinous or viscous substance, soft, moist and sticky The perceived moisture content of the cheese. Ranging from dry to moist. The amount of force required to completely bite through the sample/The extent to resistance offered by the cheese, probably judged in the first half of chewing using the front teeth The degree to which the chewed sample sticks to the surfaces of the mouth and teeth
Smoothd
The smoothness of the cheese against the palate as it breaks up during mastication
Grainyd
The extent to which granular structures are formed as the sample breaks down Evaluate how much the sample has broken down during mastication/The amount of breakdown that occurs in the sample as a result of mastication
Rate of breakdownd
1% MSG solution = 7.0 0.01% caffeine solution = 0.5 0.05% caffeine solution = 2 0.35% NaCl solution = 5.0 0.5% NaCl solution = 7.5 0.8% NaCl solution = 9.0 0.015% Citric acid solution = 1.5 0.025% Citric acid solution = 2.5 0.035% Citric acid solution = 3.5 2% sucrose solution = 1 5% sucrose solution = 2.5 0.05% alum solution = 1 0.15% alum solution = 2.5 Processed cheese = 1 Parmesan = 1 Processed cheese = 5 Processed cheese = 2 Sharp Cheddar = 9 Parmesan = 14 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 1 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 14 Sharp Cheddar = 1 Parmesan = 5 Parmesan = 1 Sharp Cheddar = 10 Processed cheese = 14
Processed cheese: Velveeta brand 907-g loaf (Kraft Foods North America; Glenview, IL). Sharp Cheddar: Kraft brand 227-g block (Kraft Foods, Glenview, IL). Parmesan: Kraft brand 113-g triangle (Kraft Foods, Glenview, IL). a Attribute found in Drake et al. (2001). b Attribute found in Heisserer and Chambers (1993). c Attribute found in Meilgaard et al. (2007). d Attribute found inYates and Drake (2007). e Not found in the literatures above.
establish whether they had significantly different preference for the cheeses. External preference mapping was conducted using Partial Least Squares regression (PLS) (Martens & Martens, 1986). The sensory data were used as the explanatory variables (X-matrix) and the average preference data within each cluster as the response data (Y-matrix). Unscrambler (v9.7 CAMO, AS, Norway) was used to carry out PLS and to determine the relationships between the two independent data sets. 3. Results and discussions 3.1. Descriptive sensory analysis The panels mean scores and ANOVA from the descriptive sensory data are shown in Table 3. Significant differences (P < 0.05) were found in 19 of the 24 sensory attributes. These included the flavor attributes ‘‘milky”, ‘‘milk fat”, ‘‘soured milk”, ‘‘free fatty acid”,
‘‘nutty”, ‘‘plastic”, ‘‘stale”, ‘‘earthy”, ‘‘toasted”, ‘‘umami”, ‘‘bitter”, ‘‘salty” and ‘‘sour”, and the texture attributes ‘‘slimy’, ‘‘moist”, ‘‘firmness”, ‘‘smooth”, ‘‘grainy’ and ‘‘rate of breakdown”. However, ANOVA failed to show how these descriptors interacted with the overall sensory characters and the extent of difference between the cheeses. PCA was used to analyze the descriptive sensory data. This analysis significantly discriminated between the cheeses (P < 0.05) on the first three PCs which accounted for 44%, 29% and 11% of the experimental variance, respectively (Figs. 1 and 2). As seen from Fig. 1, Brick cheese was characterized by ‘‘plastic”, ‘‘stale” and ‘‘earthy” flavors. Brick cheese was negatively correlated with Cream cheese indicating that they were quite different in sensory character. Cream cheese was ‘‘slimy” in texture and ‘‘milky” and ‘‘milk fat” in flavor which were defined as young/fresh cheese flavor (Van Hekken et al., 2006). Cottage cheese was in the quarter between positive PC1 and negative PC2 characterized by ‘‘soured milk” and ‘‘grainy”. Processed Cheddar cheese was positioned along the negative PC1 and was described as having a ‘‘salty”, ‘‘nutty” and
104
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Table 3 Descriptive attribute intensity means for the seven cheeses varieties. Attributes
Significant
Brick
Cottage
Cream
Havarti
Cheddar
Mozzarella
Process
Duncan’s Critical value
Milky Diacetyl Milk fat Soured milk Buttery Free fatty acid Nutty Plastic Stale Earthy Toasted Umami Bitter Salty Sour Sweet Astringent Slimy Moist Firmness Stick Smooth Grainy Rate of breakdown
**
2.5 1.04 0.93 0.14 2.25 2.54 0.21 1.39 0.75 1.25 1.07 1.11 0.39 7.57 1.32 0.14 0.57 0 4.71 5.46 1.64 6.11 0.5 7.61
3.64 0.57 1.82 2 2.86 1.89 0 0.36 0.36 1 0.46 0.18 0 6.14 1.64 0.18 0.68 0.36 9.5 3.71 0.86 1.71 3.64 9.93
4 0.43 3.07 0.5 3.5 2.36 0 0.07 0.11 0.61 1.14 0.39 0 6.71 1.54 0.21 0.32 1.61 7.86 3.54 2.36 11.21 0.04 9.5
3.18 0.36 2.46 0.36 3.36 3.5 0.39 0.43 0.21 1.14 1 1.54 0.07 7.36 1.82 0.14 0.5 0.54 7.29 4.82 1.75 8.29 0 8.21
3.39 0.21 2.79 0.46 3.36 4.11 0.39 0.36 0.18 0.64 1.71 1.79 0.21 7.64 1.54 0.14 0.57 0 4.79 6.57 1.93 6.86 0 7.39
3.18 0.54 2.07 0.25 2.89 2.46 0.54 0.86 0.21 0.96 0.96 1.04 0 7.14 1.32 0.14 0.46 0 5.21 5.64 1.54 6.96 0.07 8.07
2.64 1.21 1.89 0.21 3.21 1.96 0.79 1.07 0.5 1.04 2.04 1.96 0 7.93 1.39 0.14 0.5 0 3.75 8 2.04 5.21 0.46 6.93
6.9 2.1 3.0 29.3 1.9 15.6 3.0 2.2 2.3 2.3 6.9 12.5 13.6 3.0 2.4 0.4 1.3 32.3 28.1 17.2 2.2 26.9 29.2 5.9
ns ** **
ns ** * * * * ** ** ** * *
ns ns ** ** **
ns ** ** **
Flavor attributes were scored using a 0- to 15-point universal SpectrumTM intensity scale where 0 = absence and 15 = high intensity of an attribute. Texture attributes were scored using a 0- to 15-point product-specific scale where 0 = absence and 15 = high intensity of that attribute within that product category. ns = Not significant. * P < 0.05. ** P < 0.01.
‘‘umami” flavors and ‘‘firmness” texture. Cheddar and Havarti cheeses were similar in terms of their ‘‘free fatty acid” flavor, however, Cheddar cheese was more defined in this characteristic. These two descriptors ‘‘nutty” and ‘‘free fatty acid” were always detected in many aged/sharp cheeses (Drake et al., 2001; Heisserer & Chambers, 1993). This indicated the three cheeses (Processed Cheddar, Cheddar and Havarti) were characterized by higher intensities of aged cheese flavors than what was found in Cream and Cottage cheese. The flavor of Mozzarella cheese was positioned toward the center of the graph, showing that the sensory character of this cheese was not well explained by the first two PCs. As seen from Table 3, the intensity of all attributions in this cheese was neither too strong nor too mild. Mozzarella cheese was also chosen as the control cheese in the consumer assessment due to its mild sensory profile. The third PC of the analysis (Fig. 2) tended to distinguish cheeses on the basis of more subtle sensory characteristics. Cheddar and Havarti cheeses were also ‘‘sour” in flavor. 3.2. Consumer assessment Demographic information and consumption characteristics of 217 consumers are presented in Table 4. Of these consumers, 48% were male and 52% were female. Only 7% had been aboard. 36%, 29%, and 35% of the youth were between 12–15, 15–20, and 20–25 years old, respectively. Regarding the consumption frequency, it ranged mainly from once a week to once a month, which counted 65% of total. By contrast, a survey of 3538 American subjects about their frequency of cheese consumption indicated that 23% subjects consumed cheese nearly every day, 38% of them ate it several times a week, 15% subjects consumed it once a week, and 23% of them consumed it less than once a week (Dairy Management Inc., 1999). Thus, Chinese youth in this study consumed cheese infrequently as compared to the Americans. Till date, cheese still is not one of the favorite dairy products for most Chinese people.
Information regarding cheese-containing food was explored (data not showed). 201 youth (93%) had the experience of having hamburgers, 154 having pizza (71%) and 87 having sandwich (40%). Only 24 subjects (11%) had consumed cheese alone. It was probably that Chinese people were used to consuming cheese with other types of food as an ingredient and were not aware that cheese could be eaten alone. Moreover, eating cheese alone was thought to be not enticing. 16 subjects (7.4%) indicated that they had consumed other cheese-containing food such as Italian cuisine, cheesecake, Au gratin foods and bagel. Overall, hamburger and pizza were the most popular cheese-containing food among Chinese consumers. Now it is very common that people can easily find these cheese-containing foods at western fast-food chains (e.g., McDonald’s, KFC, Pizza Hut, Papa John’s, etc.). The mean scores of consumer preference for these cheeses showed that Cream and Cottage cheeses were the most preferred (mean scores of 7.0 and 6.5, respectively), while Havarti cheese was the least liked (mean score of 5.23). The differences in liking scale were more prominent between the clusters identified by the k-means clustering method. Cluster analysis identified five segments of consumers with different preferences with the consumer sample (Fig. 3). The demographic characteristics of these consumer segments are displayed in Table 4. Cluster 1 (40 consumers) would accept all the cheeses except Havarti, and appreciated Cottage, Cream and Processed Cheddar cheeses more. 73% consumers within this cluster claimed they had consumed cheese-containing food at least once a week. This might be due to their relatively more foreign experience (35%). Cluster 2, including the fewest consumers (22 consumers), scored all cheeses the lowest and disliked Cheddar cheese the most. This segment consisted mainly of consumers between 20 and 25 years old (73%). In addition, half of these consumers had never consumed cheese food before (45%). Presumably, consumers in this group might be greatly influenced by the Chinese traditional culture, and this cluster would not be easily pleased. Cluster 3 (40 consumers) liked Cream and Cottage cheeses
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
105
Fig. 1. Scores and loadings for PCA of the results of descriptive sensory analysis.
Fig. 2. Scores and loadings for PCA of the results of descriptive sensory analysis.
and showed less interest in Brick, Havarti, Cheddar and Mozzarella cheeses. Consumers in Cluster 4 (25 consumers) scored all cheeses lowly, with the exception of Cream cheese. Consumers in Cluster 5 (90 consumers) scored all cheeses highly and expressed more appreciation for Cottage cheese. This group had the highest proportion of the 12–15 years old (51%). This finding suggested that perhaps younger teenagers had a higher acceptance of new foods and
perceived that food products from foreign countries had better quality. Nicklaus, Boggio, Chabanet, and Issanchou (2004) also pointed that younger consumers were most open to the new type of food-cheese and really liked all the tested cheeses. Cluster analysis in our study allowed for a more detailed insight into consumer likes and illustrated the diversity of preference across the consumer testers. Pagliarini et al. (1997)
106
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Table 4 Number and percentage of consumers within each segment identified by HCA. Cluster
*
All
1
2
3
4
5
Sex Male Female
25(63%) 15(37%)
11(50%) 11(50%)
18(45%) 22(55%)
10(40%) 15(60%)
40(44%) 50(56%)
104(48%) 113(52%)
Age class* 12–15 15–20 20–25
15(37%) 9(23%) 16(40%)
4(18%) 2(9%) 16(73%)
5(13%) 18(45%) 17(42%)
8(32%) 11(44%) 6(24%)
46(51%) 23(26%) 21(23%)
78(36%) 63(29%) 76(35%)
Foreign experience* With Without
14(35%) 26(65%)
0(0%) 22(100%)
0(0%) 40(100%)
0(0%) 25(100%)
2(2%) 88(98%)
16(7%) 201(93%)
Cheese consumption* Never Once per year Once per month Once per week Several times a week Total
0(0%) 3(8%) 8(20%) 17(43%) 12(30%) 40
10(45%) 9(41%) 2(9%) 1(5%) 0(0%) 22
1(3%) 5(13%) 22(55%) 10(25%) 2(5%) 40
5(20%) 12(48%) 6(24%) 2(8%) 0(0%) 25
2(2%) 15(17%) 44(49%) 28(31%) 1(1%) 90
18(8%) 44(20%) 82(38%) 58(27%) 15(7%) 217
Indicates different between the 5 clusters (P < 0.05).
Fig. 3. Mean preference scores for cheeses within clusters. Preference was scored on a 9-point hedonic scale where 1 = dislike extremely and 9 = like extremely. (j Brick h Cottage Cream Havarti Cheddar Mozzarella Process).
identified six clusters for three different types of Mozzarella cheese and concluded that cluster analysis resulted in a deeper understanding of consumer liking for these cheeses. Additionally, these differences in preferences among the clusters in our study indicated that cheeses with specific flavor profiles could be mar-
keted to specific target consumers. Murray and Delahunty (2000) carried out a consumer questionnaire and came to similar conclusions when demographic and sociological variables were related to a final four cluster solution. It was also agreed that socioeconomic data had influenced consumer preference.
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
3.3. External preference mapping For a full interpretation of relationship between descriptive data and consumer evaluation, external preference mapping was conducted using the average preference scores of each cluster. The main position of preferences of the five clusters is distinct in Fig. 4, and the first two PC’s explained 60% of the liking variation and 69% of the sensory variation. As seen from the graph of validation variance of the Y-variance (Fig. 5), the first two PCs reduced the error variance and then the third PC went up. Therefore, we commented the data based on the first two PCs in our study. On the basis of overall liking of cheeses, most consumers were apt to accept milk-like and yoghurt-like sensory attributes, such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures. Meanwhile, most of them could not readily accept ‘‘bitter”, ‘‘salty”, ‘‘umami” and ‘‘free fatty acid” flavors and ‘‘firmness” texture in cheese. Cream and Cottage cheese, the two milk-like and yoghurt-like cheeses, received the highest scores of the seven cheeses in our study. Many Chinese cheese researchers indicated Cream and Cottage would be most acceptable for Chinese consumers, but there was no exact data in their literatures (Bo & Zhou, 2009; Ma, Gong, Wu, & Liu, 2006). In addition, they reported another popular cheese – Chinese Royal Cheese. Chinese Royal Cheese, also called Gua-nai, is an oriental-type dairy product coagulated with glutinous rice wine (Onyeneho, Partridge, Brunner, & Guan, 1987). It is characterized by sweet, fruity, and slightly alcoholic flavor and soft curd appearance (Jiang, Chen, Xue, & Chen, 2007). A second observation from Fig. 4 is that cluster 5 which was closest to the ‘‘all” cluster. Drivers of liking for this group were ‘‘sour” and ‘‘milky” flavors, and ‘‘slimy” texture. Disliking was driven by ‘‘stale”, ‘‘plastic” and ‘‘earthy” flavors. Consumers in this cluster gave high and similar scores for all the seven cheeses. It could be assumed that all the products evaluated within the present study would be appreciated by this group. Therefore, this group could be considered as the ‘‘positive consumers” which means liking all the cheeses. Cluster 2 is very different from all the other groups. It was located along the positive PC2 axis. The attributes
107
that significantly influenced dislike were ‘‘free fatty acid” and ‘‘milk fat” flavors, and ‘‘smooth” texture. There was no significant preference among consumers in this group, as these consumers scored all cheeses lowly. Taking into account the low preference for cheese, we tentatively considered Cluster 2 as ‘‘negative consumers”. For Cluster 1, ‘‘grainy” texture and ‘‘soured milk” flavor were found to be drivers of liking, whereas ‘‘free fatty acid” flavor was driver of disliking. Cluster 1 scored all cheeses highly, with the exception of Havarti and Cheddar cheeses. These two cheeses were characterized with ‘‘free fatty acid” which is considered as aged cheese flavor. Another typical aged cheese flavor is ‘‘nutty” which is rich in Processed Cheddar cheese. However, consumers in this cluster scored Processed Cheddar cheese highly. This indicated that consumers in cluster 1 did not dislike all the aged cheese flavors, as some of them liked ‘‘nutty” flavor but disliked ‘‘free fatty acid” flavor. Therefore, we considered consumers in Cluster 1 as ‘‘free fatty acid dislikers”. Consumers in this cluster consumed cheese much more frequently, which indicating that some cheese attributes such as ‘‘nutty” could be cultured among Chinese consumers. Cluster 3 appeared to be positively influenced by similar sensory characteristics as Cluster 1, liking ‘‘grainy” texture and ‘‘soured milk” flavor and disliking ‘‘free fatty acid” flavor. However, Cluster 3 discriminated much more between the cheeses, as disliking was also driven by ‘‘umami” and ‘‘salty” flavors and ‘‘firmness” texture. Consumers in this group showed a strong appreciation for soft cheese (Cream and Cottage cheese). Thus this cluster could be considered as ‘‘soft texture lovers”. Cream Cheese particularly appealed to the consumers in Cluster 4, which could be explained by its ‘‘smooth” texture and ‘‘milk fat” flavor. The consumers within this group could be considered as ‘‘Cream cheese lovers”. Previous research has addressed cheese flavor and consumer testing methods. Lawlor and Delahunty (2000) conducted preference mapping and cluster analysis with 10 specialty cheeses in Ireland. Their results indicated diverse flavor differences among cheese varieties and wide consumer preferences for these cheeses. Most recently, Liggett, Drake, and Delwiche (2008) investigated the flavor attributes that drove consumer preference for Swiss cheese.
Fig. 4. External preference map of combined consumer data with descriptive analysis results.
108
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109
Fig. 5. Residual validation variance of the Y-variance of External preference map.
Diacetyl, whey, milk fat, and umami were found to be drivers of liking, whereas cabbage, cooked, and vinegar were drivers of disliking. Our study indicated diverse flavor differences among the seven cheese varieties but a relative homogenous preference of Beijing youth for cheese flavor. Milk-like and yoghurt-like sensory attributes, such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” textures were the favorable characteristics of cheese for most Chinese consumers. 4. Conclusions Descriptive analysis and consumer evaluation showed distinctive flavor profiles and their relationship with acceptability of the seven cheese varieties. 19 of the 24 descriptors were found significantly different among these cheeses. As for cheese consumption behavior, Chinese youth in this study consumed cheese infrequently as compared to the Americans, and hamburger was the most frequently consumed cheese-containing food. The Chinese, once thought to dislike cheese of all kinds, have adjusted their lifestyles to a variety of cheese and products. The potential cheese market in China is enormous and may be larger than once thought possible. Cluster analysis identified five clusters of consumers and greatly enhanced the understanding of consumer assessment. Cheeses with specific sensory profiles could be marketed to specific target consumers. External preference mapping revealed correlations between descriptive and preference data and identified the sensory characteristics of cheeses preferred by the consumer clusters. Overall, most Chinese youth preferred cheeses with mild sensory attributes such as ‘‘soured milk”, ‘‘sour” and ‘‘milky” flavors and ‘‘rate of breakdown”, ‘‘slimy” and ‘‘moist” texture. In addition, ‘‘nutty” as a typical flavor in aged cheese was desirable to some consumer cluster and could be cultivated among Chinese consumers. Acknowledgments This project was financially supported by the Mega-projects of Science Research for the 11th Five-Year Plan (Item No. 2006BAD04A06) of Ministry of Science and Technology of China and National Key Technologies R & D Program of China (Item No. 2006BAD04A13).The authors thank members of the sensory panel
in Kraft Foods Limit (Glenview, IL, USA) and the consumer test members in Beijing for their participation in this study. The authors acknowledge Limin Zheng and Luda Zhang for their assistance in statistical analyses.
References Bo, S., & Zhou, G. H. (2009). The current situation and countermeasures of Cheese in China. Academic Periodical of Farm Products Processing, 181, 73–75 (In Chinese). Caspia, E. L., Coggins, P. C., Schilling, M. W., Yoon, Y., & White, C. H. (2006). The relationship between consumer acceptability and descriptive sensory attributes in Cheddar cheese. Journal of Sensory Studies, 21, 112–127. Dairy Association of China (2007). DAC statistical analysis Data.
X.Y. Zhang et al. / Food Quality and Preference 22 (2011) 101–109 Murray, J. M., & Delahunty, C. M. (2000). Consumer preference for Irish farmhouse and factory cheeses. Irish Journal of Agricultural and Food Research, 39, 433–449. Nicklaus, S., Boggio, V., Chabanet, C., & Issanchou, S. (2004). A prospective study of food preferences in childhood. Food Quality and Preference, 15, 805–818. Onyeneho, S. N., Partridge, J. A., Brunner, J. R., & Guan, J. (1987). Manufacture and characterization of Gua-nai: A new dairy food produced with an oriental-type culture. Journal of Dairy Science., 70, 2499–2503. Pagliarini, E., Monteleone, E., & Wakeling, I. (1997). Sensory profile description of Mozzarella cheese and its relationship with consumer preference. Journal of Sensory Studies, 12, 285–301. Peryam, D. R., & Pilgrim, F. J. (1957). Hedonic scale method of measuring food preferences. Food Technology, 11, 9–14.
109
Schlich, P. (1995). Preference mapping: Relation consumer preferences to sensory or instrumental measurements. In P. Etivant & P. Schlich (Eds.), Bioflavor (pp. 231–245). INRA Editions: Versailles, France. Van Hekken, D. L., Drake, M. A., Molina Corral, F. J., Guerrero Prieto, V. M., & Gardea, A. A. (2006). Mexican Chihuahua cheese: Sensory profiles of young cheese. Journal of Dairy Science, 89(10), 3729–3738. Yang, J. (2007). The development prospects for China dairy industry from the increase of dairy consumption perspective. China Dairy, 3, 11–13 (In Chinese). Yates, M. D., & Drake, M. A. (2007). Texture properties of Gouda cheese. Journal of Sensory Studies, 22(5), 493–506. Young, N. D., Drake, M., Lopetcharat, K., & McDaniel, M. R. (2004). Preference mapping of Cheddar cheese with varying maturity levels. Journal of Dairy Science, 87, 11–19.