A Framework for Establishing Standard Reference ... - ACS Publications

Dec 2, 2015 - Dawning Information Industry Company, Ltd., Beijing 100193, People's ... panel of the intensity scale for sensory attributes), and refer...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/JAFC

A Framework for Establishing Standard Reference Scale of Texture by Multivariate Statistical Analysis Based on Instrumental Measurement and Sensory Evaluation Ruicong Zhi,*,†,§ Lei Zhao,§ Nan Xie,§ Houyin Wang,§ Bolin Shi,§ and Jingye Shi# †

School of Computer & Communication Engineering, University of Science and Technology, Beijing 100083, People’s Republic of China § China National Institute of Standardization, Beijing 100191, People’s Republic of China # Dawning Information Industry Company, Ltd., Beijing 100193, People’s Republic of China ABSTRACT: A framework of establishing standard reference scale (texture) is proposed by multivariate statistical analysis according to instrumental measurement and sensory evaluation. Multivariate statistical analysis is conducted to rapidly select typical reference samples with characteristics of universality, representativeness, stability, substitutability, and traceability. The reasonableness of the framework method is verified by establishing standard reference scale of texture attribute (hardness) with Chinese well-known food. More than 100 food products in 16 categories were tested using instrumental measurement (TPA test), and the result was analyzed with clustering analysis, principal component analysis, relative standard deviation, and analysis of variance. As a result, nine kinds of foods were determined to construct the hardness standard reference scale. The results indicate that the regression coefficient between the estimated sensory value and the instrumentally measured value is significant (R2 = 0.9765), which fits well with Stevens’s theory. The research provides reliable a theoretical basis and practical guide for quantitative standard reference scale establishment on food texture characteristics. KEYWORDS: standard reference scale, hardness, sensory evaluation, instrumental measurement, multivariate statistical analysis, correlation relationship



INTRODUCTION Standard references play a key role in food sensory analysis. Standard reference has been defined as “any chemical, spice, ingredient or product which can be used to characterize or identify an attribute or attribute intensity”.1 Another common frame of reference has been defined as “the background information and reference points (frame of comparison) that assessors mentally refer to when evaluating products”.2 References assist in demonstrating the sensory concepts, determining the intensities of concepts, defining scales, shortening training time, and helping to maintain the original meaning of the language.3 Standard references could be subdivided into three categories according to the context, such as qualitative reference (unifying the understanding of the sensory panel with regard to sensory concepts), quantitative reference for certain sensory attribute (unifying the understanding of the sensory panel of the intensity scale for sensory attributes), and reference for sensory quality (unifying the understanding of the sensory panel of the overall sensory quality). The quantitative standard reference scales are helpful in assessor training and intercomparison testing studies and are mainly developed for sensory attributes of flavor and texture. In International Standard (ISO) 11036 (1994), texture is defined as “all the mechanical, geometrical and surface attributes of a product perceptible by means of mechanical, tactile and, where appropriate, visual and auditory receptors”.4 The mechanical attributes are those related to the reaction of the product to stress. They could be divided into five primary attributes of hardness, cohesiveness, viscosity, springiness, and © XXXX American Chemical Society

adhesiveness and into three secondary attributes of fracturability, chewiness, and gumminess.5 The relationship between sensory and instrumental ratings for the texture reference scales has been reported in many studies, including those reported in refs 6−8. The commonly acceptable texture scales of standard reference are published in ISO 11036,4 including the standard reference scale of hardness, cohesiveness, viscosity, springiness, adhesiveness, fracturability, chewiness, and gumminess. A number of studies modified the ISO standard reference with local foodstuffs. Bourne et al.9 modified the ISO standard reference with Colombian foods. Hough et al.10 developed a set of 13 reference scales with Argentine foods, and Analia et al.8 improved the texture standard reference of Argentina developed by Hough et al. The research of sensory standard reference in China was initiated in 1990s, which was later than in some countries.11 Early on, Shen et al.12 constructed a set of texture references with Chinese food, including hardness, fracturability, chewiness, viscosity, springiness, and gumminess. Recently, most reported works on sensory evaluation of food texture have been limited to one product or one category of products, for example, Yuan and Chang13 on cheese, Wang et al.14 on White Rabbit creamy candy, Chen et al.15 on pork, Sun and Liu16 on instant corn, and Wen et al.17 on Guangdong style mooncake. Duan et al.18 reported the fracturability standard reference scale with a series of Chinese Received: June 29, 2015 Revised: October 25, 2015 Accepted: November 3, 2015

A

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

Figure 1. Illustration of the framework of standard reference scale establishment.

Table 1. Standard References of Hardness term

Colombia (1975)

ISO (1994)

soft

cream cheese (Kraft) egg white cream cheese (Ubaté) frankfurter sausages (Suiza) mozzarella cheese (La Perfecta) peanuts (La Rosa) carrots peanut brittle (Colombia)

cream cheese (Kraft) egg white frankfurter sausages (Mongen David) cheese (Kraft) olives (Cresca) peanuts (Planters) carrots peanut brittle (Kraft)

creamy candy (Colombia)

rock candy

hard

Argentina (2011) ́ cream cheese (La Serenisima) egg white frankfurter sausages (Tres Cruces) olives (Nucete) peanuts (Kelloggs) chocolate (Suchard) candy (Arcor)

China (1993) K fruit gel (Shanhua) egg white ham sausage (Zhengrong) Xiao-Jiling stomachic (Laixiu) olives (Yindun) peanuts carrots Royal jelly milk chocolate (Shenfeng) rock candy

groups of reference samples. Finally, the traceability of references was verified by psychophysics law (Stevens’s theory). Although there are several existing hardness standard reference sets (the comparison of existing hardness standard reference is listed in Table 1), some of the products are difficult to buy in China, and they are not familiar to the Chinese people. Moreover, the products in the reference sets studied in Shen et al. are out of date, most of the manufacturers having disappeared. Therefore, the existing hardness standard references should be improved with recent well-known Chinese foods, which makes it more reasonable in practice. In this study, the intensity scale of hardness standard reference (texture attribute) is developed within the framework of multivariate statistical analysis method with Chinese foods. The reasonableness of the framework method is evaluated. The research provides a reliable theoretical basis and practical guide for quantitative standard reference scale establishment on food texture characteristics.

foods. Most of the studies verified the relationship between sensory and instrumental measurements of texture. Few of them applied instrumental measurement to select reference samples. The instrumental measurement values were not paid sufficient attention to find typical products for certain attributes. It is worth mining essential characteristics of reference products by both instrumental and sensory measurements. In this study, a framework for establishing standard reference scale by multivariate statistical analysis according to instrumental measurement and sensory evaluation is proposed (Figure 1). Multivariate statistical analysis is conducted to rapidly select typical reference samples with characteristics of universality, representativeness, stability, substitutability, and traceability. Clustering analysis and principal component analysis (PCA) were utilized to verify the representativeness of the samples; relative standard deviation values focused on verifying the stability of the samples, and ANOVA was conducted to verify the differences between B

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

C

vegetable product and pickle egg products sugar aquatic products other

bean products dairy products

roasted nuts

preserved fruit product

candy

potato and puffed product

cake

cookies canned

meat products

processed grain

category

Table 2. Food Products food products Mankattan golden bread (Mankattan), Bimbo classic plain sliced bread (Bimbo), Mankattan classic white bread (Mankattan), Dali-Yuan egg milk fragrant cake (Dali), Mankattan sun cake (Mankattan), Master Kong muffin (Master Kong), Panpan mini bread (Panpan) Harbin sausage (Dazhong), King of King ham (Shineway), Shineway sandwich sausage (Shineway), Vienna sausage (Shineway), Shineway chicken ham (Shineway), Jinluo fish sausages (Jinluo), Jinluo small roasted sausage (Jinluo), Shineway fish sausage (Shineway), Mother’s beef stick (Chengjia) Danco waffle (Danco), Alliance soft cookies (Guanghe), Mini chips ahoy (Kraft), Ginger cracknel (Jiapinjia) Yiming canned sour gherkin (Yiming), Yiming canned Chinese yam (Yiming), Yiming canned chestnut (Yiming), Meikeduo canned pear in syrup (Meikeduo) Yushiyuan pea flour cake (Yushiyuan), Fujunshan haw jelly (Likang), Yushiyuan haw jelly (Yushiyuan), Daoxiangcun crisp cookie (Daoxiangcun), Jili vegetarian cake (Jili), Jinguangyuan mung bean cake (Jinguangyuan) Nong Shim shrimp (Nong Shim), Mars crisp bread (Mars), Oishi prawn crackers (Oishi), Wangwang mini bun (Wangwang), Hsu Fu Chi maiqiaobo wafer (Hsu Fu Chi), Jinfuwang rice crackers (Jinfuwang), Uncle Pop egg yolk battercake (Uncle Pop), Miduoqi grilled bread (Miduoqi), Wangwang xianbei (Wangwang), Fuwa brown rice roll (Fuwa), Hsu Fu Chi wagner thin crisp (Hsu Fu Chi), grilled bread slice (Bread Talk), Orion roasted potato wish (Orion), Hsu Fu Chi pellet pancakes (Hsu Fu Chi) Hsu Fu Chi grape pudding (Hsu Fu Chi), Sister Ma fruit party jelly (Kangbeier), Labi-Xiaoxin fruit jelly (Labi-Xiaoxin), Oishi orange marshmallow (Oishi), Lizhou marshmallow (Lizhou), marshmallow (Quality & Value), Hsu Fu Chi black sesame crisp candy (Hsu Fu Chi), Bear Doctor comprehensive fruit gum (Hsu Fu Chi), Wangwang QQ gum (Wangwang), Bear Doctor orange gum (Hsu Fu Chi), Sister Ma corn jelly (Kangbeier), Caramal, Yake gum (Yake), Sister Ma golden-bar chocolate (Kangbeier), Sister Ma apple jelly (Kangbeier), fruit chew, Fruit-Tella blueberry yogurt jelly (Perfetti), Hsu Fu Chi round chocolate (Hsu Fu Chi), Alpenliebe centered toffee (Perfetti), Sugus (Swiss chocolate company, Suchard), White Rabbit milk candy (Guanshengyuan) Buy Well crisp plum (Buy Well), Yushiyuan chestnut (Yushiyuan), Ubite preserved apricot (Ubite), Orchard Farmer raisin (Orchard Farmer), Great Value red raisin (Wal-mart) Orchard Farmer peanuts (Orchard Farmer), Great Value crispy peanuts (Wal-mart), Orchard Farmer hazelnut (Orchard Farmer), Fenglin salt baked peanut (Fenglin), Cofco almond (Cofco) Baiyu tofu, Baiyu tenacious tofu, Joy dried tofu (JoyTofu) Junlebao yogurt (Junlebao), Yili pureday yogurt (Yili), Mengniu distingue spreadable cheese (Mengniu), Yili cheese (Yili), Ritz cube snack (Kraft) Heye pickled peppers bamboo shoot (Heye), Fuchunlong wild mountain bamboo shoots (Fu-yang Qianxi), Buy Well green pea (Buy Well) Yurun iron marinated egg (egg white, Yurun), Henghui dried egg (Henghui) Beijing Ershang rock candy (Beijing Ershang) Lujiazhuang grilled fresh squid (pepper, Dongfu) Zhizhonghe Chinese herbal jelly (Zhizhonghe)

structural state

organic organic gelatinous organic gelatinous

organic, gelatinous gelatinous

organic

organic

organic, gelatinous

porous

porous

porous organic

organic

porous

Journal of Agricultural and Food Chemistry Article

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry



following aspects: first, a high degree of flexibility for both the panel leader and the assessor; second, it is less susceptible to “end-effects” than those methods that employ continuous or discontinuous response scales. However, free scale usually raises “assessor effect”22 due to the different value scope and scoring habits of different assessors. In this study, the intensities were scored on a 15 cm unstructured line scale using free magnitude estimation. It was performed as follows: (1) assign values to intensity descriptors on the scale; (2) score the intensity of the texture attributes on the scale; (3) normalize the evaluation values of intensity descriptor and texture attribute and obtain the sample value of attribute under the 15 cm scale. The evaluation is repeated three times, and the average results were calculated for further processing. Table 3

MATERIALS AND METHODS

Criteria for Selection of Reference Products. Five basic principles for selecting reference products are proposed, those being universality, representativeness, stability, substitutability, and traceability.19 The first four principles have been improved according to refs 4 and 5. Universality. The reference samples should be raw or simple machining products and be generally familiar and well-known products. Representativeness. The product should possess the desired intensity of the textural attribute, and this attribute shall not be overshadowed by other textural attributes; for instance, hardness reference products should be based mainly on dry matter, with little water or oil content, avoiding the influence of other textural attributes. Stability. The reference samples should have a constant quality; different batches should have a good reproducibility and undergo minimum change upon small temperature variations. Substitutability. Under special circumstances, if some selected products are difficult to obtain, other analogues should be found. Traceability. Best practice is to establish the correlation between sensory properties and measurable physical quantities. Then the value of sensory attributes through the measured value should be estimated as much as possible to ensure the desired intensity reference products. To some extent, it has been achieved that the sensory results were traced by instrument measurement. Sample Preparation. According to the principles, all of the products in this study were bought from markets of Beijing and stored in a refrigerator. More than 100 food products were selected in 16 categories, and the structural state of the products includes porous, organic, and gelatinous. Details of the products are listed in Table 2 (type, band, product, structural state). The products were processed in the following way: make to 1 cm3 cubes for products with regular shape; minimal unit was used for products with irregular shape. The preparation process, size, shape, and temperature should be uniform, and the evaluation temperature should be the edible temperature. Instrumental Analysis. Food samples were instrumentally evaluated using the TA-XT Plus (Stable Micro Systems, Godalming, UK), which could analyze various texture attributes such as hardness, adhesiveness, viscosity, and chewiness. The measurement parameters were set as follows: model is set to “TPA”, stainless steel ball probe is P/50, moving speed is 1.0 mm/s, compression ratio is 0.7, loading weight is 5 g. All of the appliances should be clean and dry. Every test was repeated 10 times. Parameters of mechanical attributes were obtained from the instrumental curves. Sensory Evaluation. The sensory panel was composed of 10 panelists, who were recruited and trained according to ref 20. The panel was trained three times per week (for 3 weeks) before formal sensory evaluation experiments by standard sensory analysis methods, ranking31 and magnitude estimation method,30 to unify the understanding of the panel of the intensity scale for sensory texture attributes. Sensory evaluation was carried out in a sensory laboratory in accordance with the requirement of ref 21. The sensory evaluation environment should be controlled uniform. At the beginning of the experiment, the panel leader explained the term and definition of the texture attribute to assessors and demonstrated the method of sensory analysis. Samples were labeled with three random digits and provided to assessors randomly. The sensory evaluation of texture attributes consists of two stages. In the first stage, ranking31 was used to sort samples from the weakest intensity to the highest intensity. Each assessor performs the ranking three times to guarantee the consistency of the evaluation results. Samples would be eliminated or replaced if their sensory evaluation results deviated from the natural intensity. In the second stage, an improved free magnitude estimation method was used to score the intensities of the texture attribute. Magnitude estimation method30 is a psychophysical scaling technique where assessors assign numerical values to the estimated magnitude of an attribute, and the only constraint placed upon the assessor is that the values assigned should conform to a ratio principle. Magnitude estimation can offer advantages over other scaling methods from the

Table 3. Answering Table of Line Scale Free Magnitude Estimation

is an example of an answering table of linear scale free magnitude estimation. Statistical Analysis for Rapid Screening of Reference. Multivariate statistical analysis was utilized to select typical reference samples rapidly according to the instrumentally measured ratings. Fist, preprocessing should be conducted by eliminating outliers from the instrumental data set. In the preprocessing stage, a Q test was conducted to identify the outlier of the instrumental data. To perform the Q test, calculate the quantity Q, which is the ratio of the difference between the value under suspicion and the next closest value to the difference between the highest and lowest values in the series, that is Q=

|xa − xb| R

where R is the range of all data points, xa is the suspected outlier, and xb is the data point closest to xa. Compare Q with the critical value, and if Q is greater than the critical value, the suspect measurement may be rejected. The Q test is a widely used method for rejection of discordant data. The remaining data were utilized for further multivariate statistical methods. For each texture attribute, the general procedure consists of five sessions: (1) clustering analysis (CA) is conducted to group all of the samples into several categories; (2) PCA is utilized to process the samples in the same category, so as to select typical samples that could best represent the characteristics of each category; (3) relative standard deviation (RSD) is used to verify the stability of selected reference samples; (4) variance analysis (ANOVA) is calculated to determine the final reference samples with high discriminative characteristics between different scale levels; (5) finally, verify the correlation between physical quantity value and sensory evaluation value according to Stevens’s law. CA and PCA aim to verify the representativeness of the samples; the RSD value focuses on verifying the stability of the samples; ANOVA aims to verify the differences between scales of reference samples; and Stevens’s theory focuses on verifying the traceability of reference samples. The data were analyzed using statistical analysis software SPSS 18.0, MATLAB2012b, and Excel 2010. D

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

Figure 2. Scatter plot of K-means clustering analysis. The illustration shows the scatter of all the food products listed in Table 2, but some of the samples are overlapped. Clustering Analysis. CA is a general term for procedures that group variables or cases according to some measure of similarity. The variables within a cluster are highly associated with one another, whereas those in different clusters are relatively distinct from one another.23 The most widely used clustering analysis includes hierarchical clustering, K-means clustering, fuzzy clustering, dynamic clustering, etc. In this study, K-means is utilized to categorize the reference samples into several classes according to the instrumental measurements. The K-means clustering aims at partitioning the N observations into C clusters U = {U1,U2, ..., UC} in such a way that the criterion

Relative Standard Deviation. The RSD is a measure of a statistical estimate’s reliability obtained by dividing the standard error by the estimate. RSD is widely used in analytical chemistry to express the precision and repeatability of an assay.28 The equation of the RSD is the ratio between standard deviation and mean of the data given as %RSD =

where Ŝ is equal to the standard deviation and x̅ is equal to the mean value. A lower percentage indicates a lower variability in the data set. Equally, a higher percentage indicates the data set is more varied. Analysis of Variance. ANOVA is a collection of statistical models used to analyze the differences between group means and their associated procedures (such as variation among and between groups).29 The instrumental measurement and the sensory evaluation results of the reference samples are analyzed by one-way ANOVA. The selected reference samples are validated whether there are significant differences among classes. Correlation between Instrumental Measurement and Sensory Evaluation. Stevens proposed that the correlation between physical quantity value and sensory evaluation value could be expressed as30

C

arg min ∑ U



(x i − mc)T (x i − mc)

c = 1 xi ∈ Uc

is minimized. In this formula, xi is the observation of a sample and mc is the mean of cluster Uc.24 Generally, the number of clustering classes could be determined corresponding to the number of scale levels. Principle Component Analysis. PCA is a method of extracting structure from the variance−covariance or correlation matrix. Its objective is the interpretation of data relationships.25,26 PCA constructs linear combinations of the original data with maximal variance. The first component will account for the greatest portion of the variance, the second for the second largest portion, and so on, until all of the variance has been accounted for. The optimal projective axis could be determined by maximizing the covariance matrix

S = kI n where S is the perceived intensity, I is the stimulus (concentration of a chemical substance or physical variable), k is a constant that reconciles the units of measurement used for S and I, and n is the exponent of the power function and the slope of the regression curve for S and I when they are expressed in logarithmic units. In practice, Stevens’s equation is generally transformed into decimal or natural logarithms:

N

Cov =

Ŝ × 100 x̅

∑ (x i − m)(x i − m)T i=1

where m is the mean of all the samples. The score plot and loading plot could be used to visualize the relationship between the observations. The score plot is a plot of the relationship between the objects in the projected space. The loading plot is a plot that shows the relationship between the original variables and the principal components themselves.27

ln S = ln k + n ln I In theory, the correlation between sensory and instrumental value through the use of logarithmic transformation presented a linear relationship. E

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry



Figure 3. Scatter plots PCA a−h denote the eight classes, respectively.

Representativeness Verification. All of the samples were classified into nine groups by K-means clustering. (In ISO 11036, there are nine levels for hardness standard reference scale.) The scatter plot in Figure 2 shows the distribution of the nine groups, in which the x-coordinate denotes the distance of samples from the center of the clustering center it belongs to and

RESULTS AND DISCUSSION Quantitative standard reference scale of hardness texture attribute is established by the multivariate statistical method according to TPA test values, and the effectiveness of the rationality of the hardness standard reference is verified by sensory evaluation according to the psychophysics law. F

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry the y-coordinate denotes the instrumental values of hardness. Samples categorized in the same class are set to the same color. PCA is utilized for the instrumental measurement data set of all sensory attributes for each clustering class, and the hardnessrepresented samples were selected. The original dimensions of the samples are reduced to 2, and the scatter plot (black circle) together with the projective direction (red dashed line) of hardness are shown in Figure 3. Samples close to the projection direction have high representativeness of hardness. There is only one sample in the ninth class; therefore, PCA is conducted on the other eight classes. The total contribution of the first two principal components is >90%; they could represent most information of the original samples. The samples selected after CA and PCA are listed in Table 4. There are 14 samples with 9 hardness classes. Stability and Difference Verification. Samples with better stability are expected to have lower RSD values. Table 4

Table 5. Comparison between Selection Results and ISO Standard Reference of Hardness

1 2 3 4 5 6 7 8 9

reference sample

mean

RSD (%)

1 1 1 2 3 4 5 6 6 6 7 7 8 9

Baiyu tofu Mankattan golden bread Mengniu distingue spreadable cheese Labi-Xiaoxin fruit jelly Jinluo fish sausages Henghui dried egg Yiming canned chestnut Great Value crispy peanuts Orchard Farmer peanuts Orchard Farmer hazelnut Hsu Fu Chi pellet pancakes Cofco almond Sugus Beijing Ershang rock candy

238.2 185.72 130.56 1609.7 4951.46 4959.4 10729.3 21388.51 20949.26 22951.46 38431.10 44488.53 55737.14

16.6 17.8 45.5 14.4 4.8 18.3 13.1 16.4 25.8 19.7 21.2 18.2 14.2

Baiyu tofu, Mankattan golden bread Labi-Xiaoxin fruit jelly Jinluo fish sausages Henghui dried egg Yiming canned chestnut Great Value crispy peanuts, Orchard Farmer hazelnut Cofco almond Sugus Beijing Ershang rock candy

cream cheese egg white frankfurter sausages cheese olives, green peanuts carrots peanut brittle rock candy

Traceability Verification. The correlation relationship between sensory and instrumental values was verified by linear regression method. A scatter plot is frequently used to visualize the correlative relationship with the natural logarithm of instrumental measurements as independent variable and the natural logarithm of sensory evaluation as dependent variable. Generally, two variables are significantly correlated if the regression coefficient is >0.8, which means that the relationship between sensory evaluation and the traceable instrumental measurement satisfies Stevens’s law. The relationship between the natural logarithm of instrumental measurements and sensory evaluation is shown in Figure 4.

Table 4. Selected Hardness Reference Samples Based on CA and PCA class

ISO standard reference

selected hardness sample

shows the mean and RSD values of 10 times the TPA measurement values for the selected samples. Samples with RSD values 70000

Mean of the sensory evaluation. bMean of the TPA measurement. (2) Munoz, A. M.; Civille, G. V. Universal, product and attribute specific scaling and the development of common lexicons in descriptive analysis. J. Sens. Stud. 1998, 13, 57−75. (3) Murray, J. M.; Delahunty, C. M. Selection of standards to reference terms in a cheddar-type cheese flavor language. J. Sens. Stud. 2000, 15, 179−199. (4) ISO 11036. Sensory analysis − Methodology − Texture profile, 1994. (5) Szczesniak, A. S. Classification of textural characteristics. J. Food Sci. 1963, 28, 385−389. (6) Szczesniak, A. S.; Brandt, M. A.; Friedman, H. H. Development of standard rating scales for mechanical parameters of texture and correlation between the objective and the sensory methods of texture evaluation. J. Food Sci. 1963, 28, 397−403. (7) Meullenet, J.; Lyon, B. G.; Carpenter, J. A.; Lyon, C. E. Relationship between sensory and instrumental texture profile attributes. J. Sens. Stud. 1998, 13, 77−93. (8) Garcia Loredo, A. B.; Guerrero, S. N. Correlation between instrumental and sensory ratings by evaluation of some texture reference scales. Int. J. Food Sci. Technol. 2011, 46, 1977−1985. (9) Bourne, M. C.; Sandoval, A. M. R.; Villalobos, M. C.; Buckle, T. S. Training a sensory profile panel and development of standard rating scales in Colombia. J. Texture Stud. 1975, 6, 43−52. (10) Hough, G.; Contarini, A.; Muñoz, A. Training a texture profile panel and constructing standard rating scales in Argentina. J. Texture Stud. 1994, 25, 45−57. (11) Sun, H. Y.; Yuan, H. T. Development of study on sensory evaluation in food. Serv. Agric. Technol. 2009, 26, 125−126. (12) Shen, Y. X.; Jiang, D. H.; Zeng, N. W. Method for sensory evaluation of food texture. J. Shanghai Fish. Univ. 1993, 2, 135−142. (13) Everard, C. D.; O’Callaghan, D. J.; Howard, T. V.; O’Donnell, C. P.; Sheehan, E. M.; Delahunty, C. M. Relationships between sensory and rheological measurements of texture in maturing commercial cheddar cheese over a range of moisture and pH at the point of manufacture. J. Texture Stud. 2006, 37, 361−382. (14) Wang, J.; Zhou, R.; Li, S.; Huang, J. Studies on the sensory and textural characteristics of white rabbit creamy candy, China Academic Journal Electronic Publishing House. 2010, 10, 239−242. (15) Chen, L.; Wang, J.; Li, X. Regression analysis of instrumental texture characteristics and sensory characteristics of pork, Transactions of the Chinese Society of Agricultural Engineering, 2010, 26, 357−362. (16) Sun, H. T.; Liu, J. S. Sensory assess and instrumental analysis of instant corn texture. Food Sci. Technol. 2011, 36, 288−291. (17) Wen, B.; Zhang, M. W.; Zhang, Y.; Wei, Z. C.; Zhang, R. F.; Tang, X. J.; Deng, Y. Y. Journal of the Chinese Cereals and Oils Association 2012, 27, 91−96. (18) Duan, H. L.; Gu, S. Q.; Zhao, L.; Lu, D. X. Establishment of fracturability standard reference scale by instrumental and sensory analysis of Chinese food. J. Texture Stud. 2014, 45, 148−454. (19) GB/T 29604. Sensory analysis − General guidance for establishing references for sensory attributes, 2013.

references are similar to the ISO standard reference. Egg white is replaced by jelly, carrot is replaced by processed almond, and some samples are replaced by samples with Chinese brands that are easily obtained. The standard reference is classified by a ninepoint rating scale. These scales are illustrative of only the basic concept of using familiar reference products to quantify the intensity of hardness. These scales reflect the range of intensities of the mechanical attributes normally encountered in foodstuffs intended to be profiled. The reference samples established by Shen et al. also have nine-point rating scales. Some of the samples are difficult to gather now, for example, K fruit gel (Shanhua), ham sausage (Zhengrong), Xiao-Jiling stomachic (Laixiu), and Royal jelly milk chocolate (Shenfeng), as the manufacturers could not be found. Tofu is very popular in China and easy to buy in the market. Zhengrong ham sausage is replaced by Jinluo fish sausage, which is produced by the Jinluo Group. Olives are not familiar to the Chinese people, and carrot is a vegetable whose quality changes in different seasons. In the standard reference samples proposed in our work, produced chestnut and almond are determined, their quality being more stable. Overall, a new series of hardness standard reference samples are determined for Chinese foods. The new procedure of selecting standard reference samples by multivariate statistical analysis according to instrumental measurement and sensory evaluation could be conducted to establish new scale reference samples for texture properties, such as viscosity, adhesiveness, fracturability, chewiness, and springiness.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Funding

This work is supported by the program on General Administration of Quality Supervision, Inspection and Quarantine of China (No. 201110211), the National Natural Science Foundation of China (No. 31201358), the China National High Technology Research and Development Program 863 (No. 2011AA1008047), and Foundation of Basic Scientific Research of China (No. 06116070). Notes

The authors declare no competing financial interest.



REFERENCES

(1) Rainey, B. A. Importance of reference standards in training panelists. J. Sens. Stud. 1986, 1, 149−154. H

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry (20) ISO 8586-1. Sensory analysis − General guidance for the selection, training and monitoring of assessors, Part 1: selected assessors, 1993. (21) ISO 8589. Sensory analysis − General guidance for the design of test rooms, 2007. (22) Zhao, L.; Liu, W. Practical Guidance of Sensory Evaluation; Chinese Light Industry Press: Beijing, China, 2011; Chapter 2, pp 46−55. (23) Resurreccion, A. V. A. Consumer Sensory Testing for Product Development; Aspen Publications: Gaithersburg, MD, USA, 1998; pp 197−200. (24) Macqueen, J. B. Some methods for classification and analysis of multivariate observations. In Proceedings of the 5th Berkeley Symposium on Mathematical Statistics and Probability; University of California Press: Berkeley, CA, USA, 1967; Vol. 1; pp 281−297. (25) Popper, R.; Heymann, H.; Rossi, F. Three multivariate approaches to relating consumer to descriptive data. Relating Consumer, Descriptive, and Laboratory Data 1997, 39−61. (26) Lachenmeier, D. W. Rapid quality control of spirit drinks and beer using multivariate data analysis of Fourier transform infrared spectra. Food Chem. 2007, 101, 825−832. (27) Næs, T.; Brockhoff, P. B.; Tomic, O. Statistics for Sensory and Consumer Science; Wiley: Hoboken, NJ, USA, 2010; pp 210−217. (28) Dodge, Y. The Oxford Dictionary of Statistical Terms; The International Statistical Institute: The Hague, The Netherlands, 2003; pp 56−59. (29) Fisher, R. Studies in crop variation. I. An examination of the yield of dressed grain from broadbalk. J. Agric. Sci. 1921, 11, 107−135. (30) ISO 11056. Sensory analysis − Methodology − Magnitude estimation method, 1999. (31) ISO 8587. Sensory analysis − Methodology − Ranking, 2006.

I

DOI: 10.1021/acs.jafc.5b03152 J. Agric. Food Chem. XXXX, XXX, XXX−XXX