If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Rapid consumer-based sensory characterization of requeijão cremoso, a spreadable processed cheese: Performance of new statistical approaches to evaluate check-all-that-apply data
Universidade Federal Fluminense, Faculdade de Medicina Veterinária, 24230-340, Niterói, BrazilUniversidade Estadual de Campinas, Faculdade de Engenharia de Alimentos, 13083-862, Campinas, São Paulo, Brazil
We investigated the performance of multidimensional alignment analysis and multidimensional scaling on phi coefficient values to evaluate check-all-that-apply questionnaire data. We evaluated 6 dairy foods belonging to the category of requeijão cremoso processed cheese (traditional, with starch, or with starch and vegetable fat). We obtained sensory descriptors using trained assessors in descriptive analysis for comparison. A check-all-that-apply questionnaire used with 121 consumers (77 women and 44 men; 18 to 57 yr old) proved to be a suitable alternative for sensory profiling, providing descriptions similar to descriptive analysis and discriminating between products. Multidimensional alignment analysis and multidimensional scaling were efficient and logical approaches for obtaining a deeper understanding of the data, allowing us to clarify the relationships between sensory descriptors and products and contribute to optimizing the different formulations of requeijão cremoso.
Sensory perception is one of the keys to the flavorful and wholesome image that dairy foods continue to enjoy with consumers. In some cases, gross measurement of product quality or consistency may be all that is required, but for most products and market research, more detailed and complex information about sensory properties is needed (
). Outlining the sensory characteristics of products is a common practice in the dairy industry, contributing to business decisions and the development of products according to consumer needs; developing reference products; and investigating the effects of ingredients or processes and sensory changes during storage (
Descriptive analysis is a recognized and established method used by the dairy industry. It employs a trained panel to identify and quantify the intensity of each sensory descriptor, providing a complete sensory profile (
). However, DA requires a considerable amount of time, because a series of training sessions is necessary to ensure reliable results. Faster and more flexible sensory methodologies using consumers are needed (
The application of check-all-that-apply (CATA) consumer profiling to preference mapping of vanilla ice cream and its comparison to classical external preference mapping.
Diabetic and non-diabetic consumers' perception of an apple juice beverage supplemented with a Fraxinus excelsior L. seed extract having potential glucose homeostasis benefits.
). The CATA questions consist of a list of attributes (words or phrases) from which assessors can select those they consider appropriate for describing a product (
). However, when the samples to be evaluated are very similar, or if they are complex from a sensory point of view, traditional statistical methods such as Cochran's Q test and correspondence analysis may not be suitable for assessing the many the differences or correlations between products (
); new approaches that improve the interpretation of results are needed. Multidimensional alignment and multidimensional scaling are options for the interpretation of CATA findings. Multidimensional alignment is a numerical tool that allows researchers to identify associations between samples and attributes by measuring the cosine angles between them (
Multidimensional alignment (MDA): A simple numerical tool for assessing the degree of association between products and attributes on perceptual maps. In 8th Rose-Marie Pangborn Sensory Science Symposium, Florence, Italy. Elsevier,
Amsterdam, the Netherlands2009
). Multidimensional scaling can be used with phi coefficient values to determine which attributes are usually checked together and which are chosen independently, measuring the distance to specific locations in a special configuration (
). It is a typical dairy food manufactured by most Brazilian dairy enterprises regardless of their size, and recently, products similar to traditional requeijão cremoso have been developed that contain starch or vegetable fat or both. Given these differences, we wanted to obtain a deep understanding of the characteristics of the different cheeses, and to suggest improvements from a sensory perspective.
This study had 2 purposes: (1) to investigate the performance of a CATA questionnaire to provide a sensory profile of different requeijão cremoso products (requeijão containing starch; starch and vegetable oil; and a traditional formulation without these ingredients); and (2) to evaluate the use of multidimensional alignment and multidimensional scaling on phi coefficient values as a way of obtaining a deeper, complementary interpretation of the findings, which could be useful for the dairy industry. For comparison, we also performed DA using a trained panel.
). We acquired 6 samples from supermarkets in the city of Rio de Janeiro, Brazil, and kept them in the refrigerator at 9°C (±1°C) until analysis. The samples were traditional requeijão cremoso (1 sample), requeijão cremoso containing starch (3 samples), and requeijão cremoso containing starch and vegetable fat (2 samples), and were numbered I, II, III, IV, V, and VI, respectively. We used the traditional requeijão cremoso as a reference. We used only samples submitted to federal sanitary inspection that could be marketed throughout Brazil. For all sensory analyses, samples were presented in white 50-mL plastic cups identified with 3-digit numbers at a temperature of approximately 5°C.
Descriptive Analysis
Descriptive analysis (DA) was performed in the Sensory Analysis Laboratory of the Federal Fluminense University, according to the principles described by
Itambe (Uberlandia, MG, Brazil), Mane (Rio de Janeiro, RJ, Brazil), Elegê (BRF Brasil, Sao Goncalo, RJ, Brazil), Ninho Nestlé (Ituiutaba, MG, Brazil), Paulista (Danone, Rio de Janeiro, RJ, Brazil), Poços de Caldas (Poços de Caldas, MG, Brazil), Polenghi (Goiatuba, GO, Brazil), Philadelphia (Kraft Foods, Sao Paulo, SP, Brazil), Puranata (Alto da Boa Vista Frutal, MG, Brazil), Leite Moça (Nestlé, Ituiutaba, MG, Brazil), Royal (Mondelez International, Bauru, SP, Brazil), Turma da Mônica (Geleia Homemade, Itupeva, SP, Brazil).
Artificial requeijão cremoso aroma
Aroma associated with Parmesan cheese
None: UHT whole milk, Itambé
Strong: cheese essence, Mane, prepared in the proportion of 12 g of essence to 40 mL of water
Strong: 5 g of Ninho milk powder diluted in 10 mL of Itambé UHT whole milk
Acid aroma
Typical aroma related to the formation of fatty acids by hydrolysis of triglycerides
None: UHT whole milk, Itambé
Strong: natural yogurt, Paulista
Natural requeijão cremoso aroma
Milk aroma with mild salty perception
Low: UHT whole milk, Itambé
High: creamy requeijão cremoso, Poços de Caldas
Sweet taste
Taste perceived by stimulus sugars sucrose, glucose, and lactose
None: UHT whole milk, Itambé
Strong: 5 g of milk powder, Ninho, diluted in 10 mL of UHT whole milk, Itambé
Bitter taste
Taste perceived by stimulating alkaloid chemical compounds
None: UHT whole milk, Itambé
Strong: cream cheese, Polenghi
Acid taste
Taste perceived by stimulus ions released by organic acids
None: UHT whole milk, Itambé
Strong: cream cheese, Philadelphia
Salty taste
Taste perceived by soluble salts such as sodium chloride
Low: 0.1 g of Na Cl + 40 mL of water solution
High: 0.4 g of Na Cl + 40 mL of water solution
Buttery flavor
Flavor associated with unsalted butter
Low: UHT whole milk, Itambé
Strong: unsalted butter, Elegê
Artificial requeijão cremoso flavor
Flavor caused by the substance that gives cheese flavor to food
None: UHT whole milk, Itambé
Strong: requeijão sabor 4 queijos, Puranata
Natural requeijão cremoso flavor
Taste associated with dairy product, combined with light perception of salty and acid tastes
Low: cream cheese, Polenghi
High: creamy requeijão cremoso, Poços de Caldas
Oil film
Fatty sensation in the oral mucosa and lips
None: UHT whole milk, Itambé
Strong: unsalted butter, Elegê
Oral adhesiveness
Adherence to the palate when pressed against the roof of the mouth by the tongue
Low: UHT cream, Nestlé
High: UHT cream, Polenghi
Oral consistency
Speed at which the sample melts in the mouth; faster = less consistent
None: UHT cream, Nestlé
High: UHT processed cheese, Polenghi
Color
Requeijão characteristic generated from the light reflectance of the resulting product and staining
White: natural yogurt, Paulista
Yellow: sweetened condensed milk, Leite Moça
Brightness
Intensity of light reflection
Low: cream cheese, Polenghi
High: sweetened condensed milk, Leite Moça
Consistency
By dividing the product in half with a trowel and speed with which the parties reunite
Low: sweetened condensed milk, Leite Moça
High: cream cheese, Polenghi
Adhesiveness
Power joining the spoon when lying on the surface of the product, using only its own weight, and manually raised at an angle of 90°
Low: sweetened condensed milk, Leite Moça
High: pastry Dulce de leche, Elegê
Gelatinous quality
Feature that the product has when its consistency resembles gelatin
None: UHT cream, Nestlé
High: Royal Vanilla flan prepared according to manufacturer's instructions
Homogeneity
Feature that the product has when its composition or its appearance is completely equal
Low: Strawberry jam, Turma da Mônica
High: UHT cream, Nestlé
Forming wire
Product features to form long strands to be handled with the spatula
None: natural yogurt, Paulista
High: sweetened condensed milk, Leite Moça
Spreadability
The ease with which the product at 10°C is spread on toast with spatula
Low: strawberry jam, Turma da Mônica
High: UHT cream, Nestlé
1 Itambe (Uberlandia, MG, Brazil), Mane (Rio de Janeiro, RJ, Brazil), Elegê (BRF Brasil, Sao Goncalo, RJ, Brazil), Ninho Nestlé (Ituiutaba, MG, Brazil), Paulista (Danone, Rio de Janeiro, RJ, Brazil), Poços de Caldas (Poços de Caldas, MG, Brazil), Polenghi (Goiatuba, GO, Brazil), Philadelphia (Kraft Foods, Sao Paulo, SP, Brazil), Puranata (Alto da Boa Vista Frutal, MG, Brazil), Leite Moça (Nestlé, Ituiutaba, MG, Brazil), Royal (Mondelez International, Bauru, SP, Brazil), Turma da Mônica (Geleia Homemade, Itupeva, SP, Brazil).
To memorize the anchors, assessors went through 18 sessions, totaling 27 h of training. After training, assessors were asked to evaluate 3 samples in 4 replicates, using the definitive sheet for DA. All attributes were evaluated using an unstructured 15-cm line scale, anchored at the extremes by “none/weak” and “very strong” (
). The results were evaluated by 2-way ANOVA and assessors who had discriminatory capacity (pFsample ≤0.50) and good repeatability of the results (pFreplicate ≥0.05) for all attributes were selected for the final stage (
). The final evaluation was performed in 4 replicates, with 11 selected assessors in sensory booths. All samples were presented in monadic and balanced form, according to the methodology described by
A CATA questionnaire was completed by 121 consumers (77 women and 44 men, aged 18 to 57 yr) who were randomly recruited at the Veterinary School of Federal Fluminense University. Prior to testing, each participant signed a consent form. Consumers who were allergic to milk or lactose intolerant and consumers who consumed dairy products <1 time/week were excluded. Mineral water and unsalted crackers were used to cleanse the palate.
The selection of attributes for the CATA questionnaire was based on the list from the DA, removing terms that showed low mean values or no significant difference. A similar approach has been used in recent studies evaluating dairy foods such as yogurts (
Comparison of intensity scales and CATA questions in new product development: Sensory characterization and directions for product reformulation of milk desserts.
The application of check-all-that-apply (CATA) consumer profiling to preference mapping of vanilla ice cream and its comparison to classical external preference mapping.
). Based on these criteria, buttery aroma, sweet aroma, and acid aroma were removed. Color and homogeneity attributes were adapted from the scale used in the DA. In the DA, color ranged from white to yellow, so we evaluated “yellowness” and “whiteness” in the CATA. In the DA, the score for homogeneity varied from low to high, so for the CATA, we revised these extremes to be “homogeneous” and “heterogeneous.” Furthermore, other attributes were revised to be more easily understood by consumers: fatty (oil film), ease of melting in the mouth (oral consistency), viscosity (form wire), and creaminess (consistency).
Consumers were asked to select how many terms they considered appropriate to describe the samples. The position of the terms in the questionnaire was adapted according to recommended procedures from
, who found that consumers had an easier time with a CATA questionnaire when terms were grouped (i.e., odor, taste, appearance). Therefore, descriptors were assembled into 3 major groups: appearance (brightness, yellowness, whiteness, homogeneity, gelatinous quality, heterogeneity, viscosity, creaminess, and spreadability), natural and artificial requeijão cremoso aroma, and oral perception (sweet taste, salty taste, acid taste, bitter taste, natural and artificial requeijão cremoso flavor, buttery flavor, oil film, and ease of melting in the mouth). Then, groups were arranged on the page so that each group occupied all positions the same number of times. Words within attribute groups always appeared in a fixed order. Thus, we obtained 6 different page models.
Statistical Analysis
We analyzed findings from the DA using 2-way ANOVA (samples and assessors and their interaction as sources of variation), followed by Tukey's test for comparison between means (P < 0.05) to check for significant differences between samples for each attribute (
). For the CATA findings, we applied Cochran's Q test to each sensory descriptor to evaluate possible differences between treatments (samples) with binary responses. We calculated correspondence analysis using chi-squared distances on the frequency of each sample and attribute, obtaining a contingency table with orthogonal components and maximizing the sequential representation of variation of data (
Comparison of correspondence analysis based on Hellinger and chi-squared distances to obtain sensory spaces from check-all-that-apply (CATA) questions.
), new approaches recently suggested for evaluating CATA findings. We used MDA to evaluate the association of the sensory descriptors with each sample. By calculating the cosine of the angle formed between each attribute and sample (range −1 to 1), it is possible to determine which attributes have a strong relationship with each sample and obtain complete information about the relationship between products and attributes. Absolute cosines below 0.707 [= cos (45°) = −cos (135°)] indicate very little relationship (
Multidimensional alignment (MDA): A simple numerical tool for assessing the degree of association between products and attributes on perceptual maps. In 8th Rose-Marie Pangborn Sensory Science Symposium, Florence, Italy. Elsevier,
Amsterdam, the Netherlands2009
). We used the phi coefficient to check the correlation between attributes. When applied to all assessors, it allows us to determine which attributes typically appear together and which are used independently. The phi coefficients between pairs of attributes enabled us to construct a matrix of similarities that we submitted to metric MDS. The degree of correspondence among the distances between points implied by the MDS map and the matrix input by the user is measured (inversely) by a stress function. When the MDS map perfectly reproduces the input data, the stress is zero. Thus, the smaller the stress, the better the representation (
Finally, we used multiple factor analysis to compare both sensory configurations (samples and sensory descriptors) described by each methodology, assessing the sensory positioning of products in a single map. A 6-row matrix, in which each row corresponded to each sample data, was constructed with 2 blocks of columns, corresponding to the positions of the samples at the 2 first dimensions of each methodology. The similarity/correlation RV coefficient was used as a quantitative measure of this correspondence (
), and demonstrating the sensory complexity of the samples. Table 2 shows the results of the DA test. We observed significant differences for most attributes, indicating a wide variety of sensory characteristics for this group of products and providing evidence for the different procedures and formulations used by dairy companies to make requeijão cremoso. Sample I (traditional requeijão cremoso) was characterized by increased brightness and wire formation, more intense white color, and high homogeneity. Samples II and III exhibited similar profiles, with intermediate brightness, greater tendency to yellow, and a flavor similar to that of traditional requeijão cremoso. Sample III also had the highest sweetness and lower saltiness intensities. Sample IV had high saltiness intensity, lower brightness, and a strong buttery flavor. Sample V was different from the other samples with respect to the following attributes: gelatinous quality, homogeneity, spreadability and wire formation, artificial requeijão cremoso aroma and artificial requeijão cremoso flavor, sweet aroma, and consistency in the mouth. Finally, sample VI showed a tendency to white and increased consistency, adhesiveness, homogeneity, and oral adhesiveness. We observed no significant differences (P > 0.05) between the samples for the following attributes: buttery aroma, acid taste, and oil film.
Table 2Attribute mean values for each sample of requeijão cremoso in the descriptive analysis using a 15-cm nonstructured intensity scale
I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
Means with different superscript letters in the same column differed (P < 0.05).
a–e Means with different superscript letters in the same column differed (P < 0.05).
1 I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
Principal component analysis explained approximately 76% of the variation when considering the 2 first dimensions (Figure 1). The first dimension was responsible for approximately 50% of the variation, characterized by natural requeijão cremoso aroma, homogeneity, and natural requeijão cremoso flavor. The second dimension was responsible for approximately 26% of the variation, characterized by adhesiveness, buttery flavor, buttery aroma, and bitter taste. Samples I, II, and III were characterized by natural requeijão cremoso aroma and flavor, sweet taste, sweet aroma, forming wire, brightness, homogeneity, and spreadability. Samples IV and VI were characterized by adhesiveness, buttery flavor, buttery aroma, and bitter taste. Sample V was characterized by artificial requeijão cremoso aroma and flavor, oil film, color, and gelatinous quality.
Figure 1Principal component analysis of requeijão cremoso samples using descriptive analysis. I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat. F = factor. Color version available online.
We found no studies describing a DA of requeijão cremoso containing starch and vegetable in the literature, so we used traditional requeijão cremoso for comparison. The attributes used to describe the samples were similar to those reported by
, who studied 4 commercial samples of requeijão cremoso. When comparing those results with the ones obtained in the present study, we noted that forming wire, salty taste, consistency, adhesiveness, and yellow color were important characteristics of this type of processed cheese.
CATA
Cochran's Q test (Table 3) showed differences in all attributes among the samples, except for acid taste and viscosity, emphasizing the efficacy of consumer CATA questions in identifying sensory differences. The terms used in the CATA questionnaire were easy to understand, and some differences from the trained panel were due to intrinsic differences in the methods.
Table 3Frequency mention of sensory attributes associated with each sample of requeijão cremoso by consumers (n = 121) on check-all-that-apply questions
I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
P-value greater than α = 0.05 indicates no significant difference.
I
II
III
IV
V
VI
Sweet taste
9
31
42
9
10
4
<0.001
Salty taste
62
63
35
69
56
60
<0.001
Acid taste
18
18
19
13
25
12
0.125
Bitter taste
13
13
8
12
23
9
0.017
Natural requeijão cremoso flavor
56
23
30
38
6
58
<0.001
Artificial requeijão cremoso flavor
28
43
32
33
69
15
<0.001
Buttery flavor
22
36
37
33
23
24
0.028
Fatty
28
45
36
36
55
34
0.001
Easy melting on mouth
54
52
53
64
21
49
<0.001
Natural requeijão cremoso aroma
46
27
29
33
8
52
<0.001
Artificial requeijão cremoso aroma
17
27
19
13
56
11
<0.001
Brightness
72
41
44
46
19
60
<0.001
Yellow color
10
96
66
41
19
15
<0.001
White color
81
9
29
53
75
72
<0.001
Homogeneity
90
73
78
72
39
86
<0.001
Heterogeneity
2
6
3
6
38
6
<0.001
Gelatinous quality
3
5
1
5
60
10
<0.001
Viscosity
34
27
24
28
23
25
0.508
Creaminess
61
62
66
64
17
78
<0.001
Spreadability
37
28
44
36
11
23
<0.001
1 I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
2 P-value greater than α = 0.05 indicates no significant difference.
A correspondence analysis bidimensional map is shown in Figure 2. For correspondence analysis, we divided the samples into 3 groups: samples I, IV, and VI in the upper left quadrant; samples II and III in the lower left quadrant; and sample V in the upper right quadrant. Considering the frequency for each attribute, shown in Table 3, we were able to obtain a description of samples. Samples I and VI were in almost the same position and had the following properties: brightness, natural requeijão cremoso flavor, and natural requeijão cremoso aroma. These were also the attributes most likely to describe the samples. In the same quadrant, sample IV was homogeneous and creamy, and according to the attributes selected, it was also salty and melted easily in the mouth. Samples II and III had a yellow color, sweet taste, and buttery flavor. Finally, sample V was gelatinous, heterogeneous, and had an artificial requeijão cremoso aroma and flavor. Whiteness of requeijão cremoso was also frequently selected by consumers.
Figure 2Correspondence analysis of requeijão cremoso samples using check-all-that-apply findings. I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat. F = factor. Color version available online.
After developing the correspondence analysis map, we applied MDA to the coordinates of each sample and sensory descriptors. Table 4 shows the relationship between attributes and samples using MDA (cosine values >0.707 suggesting a strong relationship between samples and the attributes). Positive values indicate that the product is strongly characterized by the attribute, and negative values indicate that the product is not characterized by the attribute. Sample I was correlated with viscosity, homogeneity, brightness, and natural requeijão cremoso aroma/flavor (cosines 1.00, 0.74, 0.92, 0.93, and 0.96, respectively); it was not correlated with sweet taste, acid taste, artificial requeijão cremoso flavor, or fatty (−0.71, −0.78, −0.79, and −0.82, respectively). In contrast, samples II and III were characterized as yellow and had a buttery flavor and a sweet taste. The terms spreadability, creaminess, viscosity, homogeneity, brightness, natural requeijão cremoso aroma, easily melted in the mouth and natural requeijão cremoso flavor were the main attributes of sample IV. Sample V was correlated with gelatinous quality, fatty, heterogeneity, artificial requeijão cremoso aroma/flavor, bitter taste, and an acid taste. Sample VI was correlated with viscosity, homogeneity, brightness, and natural requeijão cremoso aroma/flavor. These findings could be interesting for dairy processors, because they can evaluate the attributes that were strongly related to the traditional sample, which did not contain starch and vegetable fat, and use it as a guideline for developing similar products. For example, sample VI, which does contains starch and vegetable fat, had a strong correlation with the same attributes (homogeneity, brightness, and natural requeijão aroma/flavor) as sample I.
Table 4Cosine values between vector pairs (product vector vs. main sensory terms) in the characterization of samples obtained by multidimensional alignment for requeijão cremoso samples
I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
Bold values indicate correlation of the sensory attributes with the respective sample. Positive values indicate positive correlation, and negative values indicate negative correlation.
I
II
III
IV
V
VI
Sweet taste
−0.71
1.00
0.92
−0.06
−0.26
−0.73
Salty taste
0.54
−0.98
−0.98
−0.15
0.46
0.57
Acid taste
−0.78
0.10
−0.30
−1.00
0.93
−0.76
Bitter taste
−0.50
0.27
−0.63
−0.95
1.00
−0.46
Natural requeijão cremoso flavor
0.96
−0.48
−0.09
0.90
−0.72
0.95
Artificial requeijão cremoso flavor
−0.79
0.12
−0.28
−1.00
0.92
−0.77
Buttery flavor
−0.66
1.00
0.94
0.01
−0.33
−0.68
Fatty
−0.82
0.17
−0.23
−0.99
0.90
−0.80
Easy melting on mouth
0.47
0.30
0.65
0.94
−1.00
0.43
Natural requeijão cremoso aroma
0.93
−0.38
0.01
0.94
−0.79
0.91
Artificial requeijão cremoso aroma
−0.70
−0.02
−0.41
−1.00
0.97
−0.67
Brightness
0.92
−0.37
0.02
0.95
−0.79
0.91
Yellow color
−0.70
1.00
0.92
−0.05
−0.27
−0.72
White color
0.57
−0.98
−0.97
−0.12
0.43
0.60
Homogeneity
0.74
−0.04
0.35
1.00
−0.95
0.72
Heterogeneity
−0.51
−0.26
−0.62
−0.95
1.00
−0.48
Gelatinous quality
−0.44
−0.33
−0.68
−0.92
1.00
−0.41
Viscosity
1.00
−0.67
−0.33
0.78
−0.53
1.00
Creaminess
0.59
0.16
0.53
0.98
−0.99
0.56
Spreadability
0.27
0.49
0.79
0.84
−0.97
0.24
1 I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat.
2 Bold values indicate correlation of the sensory attributes with the respective sample. Positive values indicate positive correlation, and negative values indicate negative correlation.
The clear and complete relationship between attributes and samples that can be assessed by MDA is not always available using the bidimensional map obtained by correspondence analysis (
). For instance, sample V did not present a clear association with bitter taste, salty taste, and sour taste when assessed by correspondence analysis bidimensional map. Use of MDA to analyze CATA findings is advantageous and confirms recent observations from a study evaluating low-sodium dry fermented sausages (
We performed MDS using phi coefficients between attributes (Figure 3). Multidimensional scaling provides a map in which terms are distributed in 2 dimensions. The proximity of terms in the bidimensional map indicates their similarity with respect to sensory properties. The results showed a stress value of 0.162 and high validity; according to
, values below 0.1 are considered fair and those up to 0.20 are acceptable. As shown in Figure 3, it is possible to observe a correlation between natural requeijão cremoso flavor/aroma, spreadability, homogeneity, creaminess, easily melted in the mouth and brightness, which were coherent and desirable in a typical processed cheese. We also observed a good correlation for artificial requeijão cremoso flavor/aroma, fatty, viscosity, heterogeneity, gelatinous quality, and bitter taste. These characteristics are not desirable in traditional requeijão cremoso and may have been associated with the vegetable fat and other additives (i.e., flavorings, thickeners) used to mimic the characteristics of the traditional product (
Requeijão cremoso e outros queijos fundidos: Tecnologia de fabricação, controle do processo e aspectos de mercado. Fonte e Editora Ltda.,
São Paulo, Brazil2014
). In addition, we observed an association between yellow color, buttery flavor, sweet taste, and acid taste in the upper left quadrant. The sweet and acid tastes may have been seen as similar by consumers because of a cross-modal interaction. Cross-modal taste and aroma interactions have been reported in the perception of cheese flavor (
These findings may be useful as a guideline for the processed cheese industry. Although one of the principal attributes of this processed cheese is natural requeijão cremoso natural flavor and aroma, it may also present with brightness, spreadability, homogeneity, creaminess, and easy melt in the mouth. Similar products should be formulated to achieve these characteristics. For example, the starch level will affect the texture of the sample, because starch interacts with the water in the cheese, forming a gel and making it firmer (
). Vegetable fat levels will result in insufficient emulsification, illustrated by a highly hydrophobic character and more primary bonds, with a direct effect on fat globule structure (increased diameter). This in turn causes a decrease in protein-protein and protein-fat interactions in the processed cheese matrix, lowering the elasticity of the product and affecting its spreadability and ease of melting in the mouth (
We used multiple factor analysis to analyze the consensus configuration between DA and CATA. Multiple factor analysis explained 75% of the variability using 2 dimensions, in which the first dimension accounted for approximately 52% and the second dimension explained 23%. The RV coefficient values ranged from 0 to 1, with values closer to 1 suggesting adequate correlation between the methods (
Novel and successful free comments method for sensory characterization of chocolate ice cream: A comparative study between pivot profile and comment analysis.
). In this study, the RV coefficient was 0.843 (P < 0.001), indicating that both methodologies provided similar information about the sensory characteristics of the samples.
Figure 4 shows the consensus configuration of the samples associated with different colors according to each sensory methodology. The different positions show clear differences among samples, suggesting that consumers were able to discriminate appropriately between samples. The points corresponding to the 2 methods (CATA and DA) were close, suggesting that sample configurations were very similar for both methods, and that trained assessors and consumers perceived the samples similarly. Overall, our findings reinforced the use of descriptive methodologies with consumers—particularly CATA, because it is useful for small and medium dairy producers to obtain information about the sensory characteristics of their products that belong to the requeijão cremoso category. However, CATA, as with any other descriptive methodology using consumers, cannot be used as a replacement for DA; the latter is more accurate because assessors are extensively trained in the identification and quantification of sensory attributes, enabling them to detect minor differences in formulation (
Figure 4Consensus configuration between descriptive analysis (DA) and check-all-that-apply (CATA) findings for requeijão cremoso samples using multiple factor analysis. I = traditional requeijão cremoso; II, III, and IV = requeijão cremoso with added starch; V, VI = requeijão cremoso with added starch and vegetable fat. F = factor. Color version available online.
Standardized formulation of products in the requeijão cremoso category is lacking in the Brazilian dairy industry. For this reason, CATA can be a valuable, reliable, and economic sensory method of establishing a sensory profile for requeijão cremoso formulations. The use of MDS and MDA based on phi coefficient values allowed us to develop an adequate understanding of the sensory profile of requeijão cremoso samples with added vegetable fat and starch, possibly serving as a guideline for dairy processors who wish to improve the formulation of their products in accordance with consumer perceptions.
CONCLUSIONS
The CATA questionnaire proved to be an effective tool for characterizing different samples of requeijão cremoso, showing results that correlated well with conventional descriptive analysis by trained assessors. Multidimensional alignment and multidimensional scaling based on phi coefficient values were important and useful alternatives for obtaining a deeper understanding of the CATA findings, adding information to the traditional analysis. Adoption of these methods should be encouraged for the processed cheese industry when it conducts sensory profiling with consumers and using a CATA questionnaire. Overall, the findings of the present study may useful for companies that manufacture dairy foods with similar or complex sensory characteristics.
ACKNOWLEDGMENTS
The authors are grateful for the financial support of this research project provided by the National Council for Scientific and Technological Development (CNPq, Brasilia, Brazil) and National Council for the Improvement of Higher Education (CAPES, Brasilia, Brazil).
REFERENCES
Albenzio M.
Santillo A.
Caroprese M.
Braghieri A.
Sevi A.
Napolitano F.
Composition and sensory profiling of probiotic Scamorza ewe milk cheese.
Comparison of intensity scales and CATA questions in new product development: Sensory characterization and directions for product reformulation of milk desserts.
Multidimensional alignment (MDA): A simple numerical tool for assessing the degree of association between products and attributes on perceptual maps. In 8th Rose-Marie Pangborn Sensory Science Symposium, Florence, Italy. Elsevier,
Amsterdam, the Netherlands2009
The application of check-all-that-apply (CATA) consumer profiling to preference mapping of vanilla ice cream and its comparison to classical external preference mapping.
Novel and successful free comments method for sensory characterization of chocolate ice cream: A comparative study between pivot profile and comment analysis.
Requeijão cremoso e outros queijos fundidos: Tecnologia de fabricação, controle do processo e aspectos de mercado. Fonte e Editora Ltda.,
São Paulo, Brazil2014 ([In Portuguese])
Diabetic and non-diabetic consumers' perception of an apple juice beverage supplemented with a Fraxinus excelsior L. seed extract having potential glucose homeostasis benefits.
Comparison of correspondence analysis based on Hellinger and chi-squared distances to obtain sensory spaces from check-all-that-apply (CATA) questions.
30540-4&id=gr1.jpg){kind=link}
30540-4&id=gr2.jpg){kind=link}
30540-4&id=gr3.jpg){kind=link}
30540-4&id=gr4.jpg){kind=link}