You are on page 1of 15

Food Quality and Preference 18 (2007) 425439 www.elsevier.

com/locate/foodqual

Food quality certication: An approach for the development of accredited sensory evaluation methods q
rez Elortondo F.J. Pe
a b

a,*

n a, I. Etayo a, M. Molina , M. Ojeda a, M. Albisu a, J. Salmero

lisis Sensorial, Facultad de Farmacia, Universidad del Pa s Vasco/Euskal Herriko Unibertsitatea, Laboratorio de Ana Paseo de la Universidad, 7, 01006 Vitoria, Gasteiz (Alava), Spain n de Origen Queso Idiazabal, Granja Modelo de Arkaute, Apartado 46, Arkaute (Alava), Spain Consejo Regulador de la Denominacio Available online 20 May 2006

Abstract The standardization and accreditation of sensory quality evaluation methods is a pressing need for the certication of food products, particularly for foods and beverages with specic sensory characteristics, such as those with a protected designation of origin (PDO). This study describes a proposal for the development of accredited sensory evaluation methods, showing as an example an application from a specic certied product (PDO Idiazabal cheese). A training and qualication process for expert panelists is required. In cheese, panelists score quality of overall sensory parameters (shape, rind, paste colour, eyes, odour, texture, avour and aftertaste) on a scale, based on how close the product lies to a specic quality standard. Panelists justify the quality scores given on the basis of the absence/ presence of specic characteristics in the product and/or the presence of defects. Training requires the prior establishment of references for both characteristics and defects. Qualication trials determine whether or not the expert panelists (both individually and as a panel) are appropriately qualied to carry out the sensory evaluation. This work also shows the quality control maintenance of qualications for the expert panellist. This approach could be generalized to any type of food and beverage as a reference for the accreditation of sensory quality evaluation methods according to ISO 17025. In this way, each product manufacturer would be able to dene its quality standard and, on the basis of this standard, carry out the sensory evaluation using a panel specically trained for this purpose. 2006 Elsevier Ltd. All rights reserved.
Keywords: Sensory quality evaluation; ISO 17025 sensory method accreditation; Food quality certication; Protected designation of origin (PDO); Idiazabal cheese

1. Introduction The scientic strategy most acceptable for sensory quality evaluation takes into account the relation from two types of data: tests with consumers (aective class, hedonic type test) and trained analytical panels (descriptive class, analytic type test). The relation between the two makes it possible to determine sensory proles best adapted to the concept of the product quality in the target market,
A poster of this paper was accepted and presented during the European Conference on Sensory Science of Food and Beverages A sense of identity in Florence (Italy) from 26th to 29th September 2004. * Corresponding author. Tel.: +34 945 013075; fax: +34 945 013014. rez Elortondo). E-mail address: franciscojose.perez@ehu.es (F.J. Pe 0950-3293/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodqual.2006.05.002
q

enabling large companies to establish control activities, improve quality, and develop new products. This strategy, however, has limitations with respect to the possibility of it being applied by small producers with typical sensory characteristic products, such as those with protected designation of origin (PDO). The integration of the two types of information requires high level technical work, where multidisciplinary research is needed (psychology, marketing, dynamic group techniques, sensory methodology, statistical multivariate analysis). There is a standardized method for the selection, basic training and development of sensory proles, but a lack of standards for consumer tests, this being one of the main needs of sensory analysis standardization.

426

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

In laboratories that only use descriptive techniques, typically the number of positive descriptors is limited (key attributes) and it is doubtful that it is even possible to achieve a product more in line with a given sensory prole, but where quality is insucient due to the presence of a defect. Finally there are some products, which, although they should be market orientated to ensure successful sales, their sensory attributes are inuenced by certain origins or processing factors. This is the case of PDO products that need to gain the acceptance of consumers while at the same time respecting their sensory typicalness. In many cases, these producers are small enterprises, where it is economically impossible for them to develop and introduce systematic quality control on the end product from hedonic and analytical information. Questions arise such as the followings: Apart from the opinion of the consumer, should the inuence of technical, cultural, social and environmental factors be taken into account to obtain dierentiated products? Would it be appropriate to propose a sensory analysis method validated and recognised (accredited), allowing a direct evaluation of the quality previously dened for a product, by means of a panel of expert panelists trained specically for the purpose? Using this method, should we take into account the possible defects known in the product that are important for its identity? How do we integrate said possible defects in the quality evaluation? Generally, in quality certication of PDO products, regulations underline the fact that these products should present sensory characteristics in accordance with the judgement of a Sensory Committee. These panelists, who may be labelled expert judges of a specic product,

should evaluate not only the presence of certain specic sensory characteristics in the product, but also the absence of possible defects. The members of these panels are people with a great deal of experience in tasting a specic product, but in many cases, neither use standardised methods nor have they been given systematic training. Many of the PDO products still lack a Sensory Committee, and this is largely due to the lack of references published on how to conduct such a task. Sensory evaluation comprises a set of techniques for accurate measurements of human responses to foods and minimizes the potentially biasing eects of brand identity (Lawless & Heymann, 1998). Sensory characteristics comprising appearance, odour, avour and texture are included within the quality of food products. There are a limited number of internationally accepted standards on general methods in sensory analysis, such as: general guidance (ISO 6658, 1985), assessors selection and panel training procedures (ISO 8586, 1993) and sensory tests (ISO 5495, 1983; ISO 4120, 1983; ISO 6564, 1985; ISO 10399, 1991; ISO 1036, 1994). These standards permit the selection, basic training of panelists and general application of discriminative and descriptive sensory methods. Using these standards it is possible to detect and describe sensory dierences between food products but not to establish and evaluate their quality or dene training procedures for experts in specic foods. Accreditation means the demonstration of technical competence. Sensory laboratory accreditation would have a future and perhaps could be realized in the next two decades (Moskowitz, Mun oz, & Gacula, 2003). ISO 17025 (1999) was developed in order to obtain an accepted system for comparable evaluation of testing activities. According

Quality definition of the product

Selection and basic training of panelists

Specific development of the method

Specific training of panelists

Individual qualification of the expert panelists

Validation of the method (proficiency tests included)

Quality control maintenance of expert panelists and of validated method

General quality management activities

Fig. 1. Main activities for the accreditation of a sensory quality evaluation method on a specic food product.

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

427

to the European Community regulations for PDO products, the certifying organizations of these food products have to comply with the European Standard EN45011 (DOCE, 1992). Consulting the mentioned standard, it becomes obvious that the analysis for certifying these products, sensory included, should be carried out in laboratories according to the standard ISO 17025. One of the most important documents nowadays for standardization of sensory activity is the European publication reference EA-4/09 Accreditation for sensory testing (ENAC, 2003) that supplements ISO 17025 and provides specic guidance on the accreditation of sensory laboratories. This document gives detailed guidance for the interpretation of ISO 17025 for those undertaking sensory examinations. Accreditation bodies will only accredit laboratories for sensory methods that have been fully documented and validated. Some laboratories have accredited general discriminative, descriptive and consumer sensory tests, but there is no previous experience in the accreditation of sensory quality evaluation methods of specic products. The aim of this study is to describe, for example, a sensory evaluation method that serves as a basis for the sensory quality certication of a specic product (PDO Idiazabal cheese), as well as the training and qualication process for expert panelists required for implementation. The paper presents the experience of four years of work and shows a general perspective across all the activities (Fig. 1) developed in an accredited sensory laboratory (June, 2005) according to ISO 17025:http://www.enac.es/ html/anx/le/UPV_EHU_LE1020_REV1.pdf. 2. Material and methods 2.1. Expert panelists for sensory quality evaluation of a specic product ISO standards on the selection and basic training of panelists may be used as a reference during the initial training stages of experts, although such standards do not establish the required level needed to pass these preliminary tests. Afterwards, expert judges should be trained specically on the sensory quality of the product under study (characteristics and defects) and for the specic application of the evaluation method. 2.2. Installations, equipment, chemical and foodstus specic references An accredited sensory analysis laboratory (SAL) should have a meeting room according to the ISO standard, standardized booths and an isolated sample preparing area. The other equipment in a SAL is not sophisticated, in order to comply with the accreditation standards. This equipment is basically limited to thermo hygrometers and continuous read-out thermometers, refrigerators, tempered sample chambers, weighing scales, distilling equipment and/or water deionisation and volumetric material, chemi-

cal compounds and foodstus used for the standardisation of references. The harmonized use of common references by dierent laboratories represents in itself a beginning of standardization and currently it could be valid in the accreditation of specic sensory evaluation methods. However, as pointed out by Issanchou, Schlich, and Lesschaeve (1997) each product needs its own adequate references which can be completely dierent from one type of cheese to another, thus making it impossible to generalize a set of standards for every variety. 2.3. Specic samples of the product During the specic training of panelists, a wide variety of product samples is essential in order to accommodate the highest possible quality variability in the product. On a daily basis, the systematic evaluations make it possible for the laboratory to identify interesting product samples for dierent purposes, such as harmonize and improve the consensus among the panel members, introduce quality control tests, and increase the experience of the panelists for the identication of defects. 2.4. Data treatment A coding and internal recording system should ensure traceability of the results and all the information from the activities of the SAL. Data treatment is applied at very dierent levels: qualication and follow-up of the panelists performance, method validation, quality control tests, prociency testing schemes, treatment of the results and test reports. Generally, statistical descriptive analysis and average comparison tests are appropriate; more sophisticated multivariate analyses are not essential. 3. Results and discussion 3.1. Sensory quality denition of the product Sensory analysis is a research discipline that should become one of the main driving forces to ensure the quality and protection of PDO food products (Bertozzi & Panari, 1993; Perez Elortondo, Barcenas, Casas, Salmeron, & Albisu, 1999). To establish the sensory quality of a PDO product and dene a method for its evaluation, it is necessary to have in-depth knowledge of the sector involved (type of company, production systems, and product variability) and to compile scientic and market information related to the type of product under study. A working group on sensory analysis should be set up to reach a consensus of opinion on progress, both in terms of sensory quality denition and the evaluation method. The members of this group are a decisive element for ensuring the success of the work. It is very important to be able to rely on people with acknowledged prestige and experience in the sector

428

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe Table 1 Sensory standard denition of Idiazabal cheese used in quality evaluation rez Elortondo, 1993) (adapted from Pe Variable Shape Optimal situation (reference) Cylindrical, well-proportioned, 812 cm high, 1030 cm diameter and 0.93 kg weight. Markedly at faces. Slightly convex sides. Rounded edges on small cheeses, sharp edges on larger ones Hard. Smooth, without trace of outside agents. Slight signs of the cloths used. Homogenous colour, from a pale yellow or whitish grey to a darker grayish-brown in the case of smoked cheeses. Absence or light marks from the trays on one of the faces Homogeneous. Variable (from ivory to straw yellow). Matt. Narrow and slightly darker border Random distribution. Absence or not very numerous. Mostly of irregular shape. Smaller than a grain of rice. Absence of cracks Intense odour, ewes milk, clean and sharp. Slightly pungent. Variable acidity (intensity null to medium) and sweetness (intensity null to medium). Medium intensity smokiness with smoked cheeses Slight elasticity. Medium rmness. Weak granularity Characteristic avour, balanced and intense: character of mature ewes milk, somewhat of natural rennet avour, clean and consistent. Slightly pungent. Variable sweetness and acidity. Absence of bitterness. Medium saltiness. Mild to medium smokiness in smoked cheeses Continuity with respect to the characteristic avour. Persistent. Pronounced

and that have something to say. Members of this working group should include product processors, restaurateurs, and technicians in sensory analysis, gastronomic critics and consumers. In order to achieve an initial sensory quality denition draft and a consensus over the evaluation method, experience has taught us that several meetings are needed, lasting up to 3 h and held weekly over a period of approximately 612 months, depending on the complexity of the product and on the number of companies involved. These meetings should include sensory evaluation sessions to discuss the sensory variability of the product, to generate and select sensory descriptors, and to record any possible defects. At this stage, it is important to identify the essential sensory attributes linked to the origin and/or particular systems of production. It is advisable to corroborate the identication of these essential attributes by means of a parallel research procedure, using descriptive methods by an analytical trained panel. These key attributes distinguish the PDO product and ensure its typicalness and protection in terms of possible imitations. The result of this work (sensory quality denition standard) should be conveyed to the production sector for acceptance as a sensory reference for the control and improvement of the product quality. Follow-up of adjustment to this initial sensory quality standard should continue, because changes may occur in the consumer preferences, in the production systems and/or new knowledge in research sensory characterisation may also be available. As an example this procedure made it possible to establish the standard sensory quality denition of a PDO product such as Idiazabal cheese (Table 1) a decade ago. Nowadays this denition is the basic reference which is used for applying the accredited sensory evaluation method developed by our SAL. 3.2. Selection and basic training of panelists 3.2.1. Selection of panelists SAL comprises three selection stages: recruiting, selection and preliminary formation. SAL includes two ways of recruiting: internal (when candidates belong to the University), or external (when candidates are expert technicians, restaurateurs or consumers). Initial selection of new judges is carried out by questionnaire, to determine personal aspects, such as the aptitude for foods, communication, health, availability and other factors of a general nature. Preliminary formation trains new judges by familiarizing them with the discipline of sensory analysis. Explanations are given based on theory, and selection tests are conducted. Test results, used as a basis for selecting panelists, address the following objectives: to detect inabilities, determine sensory sensibility and evaluate the potential of the candidates in describing and communicating their perceptions.

Rind

Paste colour Eyes

Odour

Texture Flavour

Aftertaste

Table 2 shows the tests and selection criteria for new panelists. The type of test and concentrations of the references are based on ISO standards, although some selection criteria are internal to our SAL due to the absence of specications in the standards. Selection tests are of a basic and general nature, regardless of the type of product to be tested by the panelists. Each test is conducted twice. SAL establishes that the judges achieving a P75% success rate out of the total number of tests conducted are to be accepted, and will proceed on to the following basic training stage. Those obtaining a success rate between 60% and 75% may repeat once the tests failed. If after this test they achieve a 75% success rate, then they are to be accepted; lower success rates mean rejection from participating in the subsequent basic training. 3.2.2. Basic training of panelists Table 3 shows the tests and criteria required for the panelists to pass the basic training tests, regardless of the type of food being tested. The purpose of this training is twofold. First, the training provides judges with the basic elements of the standardized techniques used in sensory analysis, developing their aptitude to detect, recognize and describe the sensory stimuli involved. Second, the training allows judges to become competent in applying these basic techniques to any food item. The training programme encompasses both theory and practice. As with the selection tests, the type of basic training test and the concentrations of the references are based on ISO standards,

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe Table 2 Tests and criteria for selection of panelists 1. Viewing colours test (Ishihara test): If the candidate views colour normally, he/she will attain certain numbers. Full points are required 2. Colour recognition test. Ranking test: Candidates are presented with ve random colour strips from the Yolk Colour Fan for colours of egg yolk. They are asked to put the strips in ascending order in terms of the degree of colour intensity. A maximum inversion is required in adjacent pairs 3. Taste sensitivity test (taste identication). 13 dissolutions are presented to judges. Each judge will taste, successively, the content of each dissolution, following the order in which they are presented, without going back to the samples previously tasted. A minimum success rate of 80% is required 4. Sapid substance recognising tests 4A. Duo-trio test: Samples representing basic tastes, at concentrations higher than the recognition threshold. A sample of each type is presented to each candidate, and they are left to familiarize themselves with the same. Then they are presented, at random, with two series of the same samples. Candidates should pair them o with the original samples. A minimum 80% success rate is required in the pairing o test 4B. Triangular test: The concentration of the product being tested should be at a supraliminal level. The candidate should detect which sample is dierent. Four trials will be presented. A minimum 75% success rate is required 4C. Ranking test: Panelist are asked to classify samples in order of growing intensity. The order of presentation will be the same for all candidates. A maximum inversion is required in adjacent pairs 5. Olfactory substance recognition tests 5A. Ranking test: Panelists are asked to classify samples in order of growing intensity. The order of presentation will be the same for all candidates. A maximum inversion is required in adjacent pairs 5B. Odour description test: Samples containing dierent smelling substances, at concentrations higher than the recognition threshold. Candidates must mentally classify the sensory perceptions identied and note down the family of odours and/or descriptors. 3 points (identication or correct description); 2 points (description in general terms); 1 point (identication or description of an appropriate association following discussion); 0 for no answer. Candidates must score at least 65% 6. Texture recognition tests 6A. Ranking test: Solid samples (foods). Panelists are asked to order the samples in ascending order, in terms of elasticity and granularity. A maximum inversion is required in adjacent pairs 6B. Test for describing texture: Solid samples are presented. Panelists are asked to describe their properties as regards texture, noting down the term and/or terms most frequently linked with its texture. 3 points (identication or correct description); 2 points (description in general terms); 1 point (identication or description of an appropriate association following discussion); 0 for no answer. Candidates must score at least 65% Table 3 Tests and criteria for basic training of panelists

429

1. Training tests in detecting and recognizing smell and sapid substances 1A. Two duo-trio tests: (a) Samples representing odours, at concentrations higher than the recognition threshold. A sample of each type is presented to each candidate, and they are left to familiarize themselves with the same. Then they are presented, at random, with two series of the same samples. Candidates should pair them o with the original samples. A minimum 80% success rate is required in the pairing o test (b) Four sets of three sapid samples with the combinations AAB, ABA, BAB, BBA are presented. Samples A and B correspond to two dierent branches of the same product. They are asked to examine the samples in each set from left to right so that the reference sample (the rst on the left) is examined rst, and to indicate which of the other two samples is identical to the reference. A minimum 75% correct answer will be required, both in the sapid and in the smell tests 1B. Pairs comparison test. Panelists are presented with two pairs of samples for each characteristic to be evaluated in terms of avour and odour. Panelists must compare, within each pair of samples, which of them presents a higher degree of intensity in terms of a particular attribute (forced judgement). A minimum 75% correct answer will be required, both in the sapid and in the smell tests 2. Training in the use of scales 2A. Use of one-dimensional descriptors: Olfactory and sapid samples are presented at dierent concentrations. To enable the panelists to become acquainted with the concepts of classication, with the help of dierent scales (structured continuous and non-structured scales, categories or discontinuous scales), they will order the odour and avour stimuli in terms of the intensity of each particular attribute. They will be asked not to invert the order of the samples in more than one of the characteristics evaluated, considering both the odour and avour scales separately 2B. Use of multidimensional descriptors. Odour, avour and texture samples are presented, at dierent concentrations. Using the same scales as in the previous sessions, panelists will put the stimuli from the samples of odour, avour and texture in order, in terms of the intensity of each particular attribute. They must not invert the order of the samples by more than one of the characteristics evaluated, considering odour, avour and texture separately 3. Training in developing and using descriptors The purpose is to make the panelists aware of the concept of prole. In order to develop the vocabulary, 2 brands of a food are evaluated. By means of the Interlinked Kelly Method, panellists will describe the dierences and similarities between the samples. Once the test has concluded and following an open discussion between the panel and the panel leader, the terms which the group considers appropriate for the sensory description of the products are chosen. Hedonic and intensity terms will be removed. Synonyms or terms presenting similar meanings are to be grouped together 3A. Sensory evaluation Panelists will evaluate the food under study in the previous test, noting points for intensity, using the attributes nally agreed on. The evaluation method for the descriptors is established and references will be used at some point on the scale. Three repetitions per brand is presented to each panelist. The panel leader will establish the prole for each product, using the results to illustrate the value of the descriptive analysis. These proles will be shown to the panelists so that they can become acquainted with their use So that the panel as a whole becomes acquainted with discriminating parameters, a comparison of averages is applied (t-Student test) on each attribute. In order to verify the discriminating capacity of each member of the panel, a t-Student test will be applied to the results from each panelist for each discriminating parameter. An individual discriminating capacity will be required of at least 50% of the discriminating parameters considered by the panel as a whole

although some selection criteria are internal to our SAL in view of the absence of specications in the standards. During the basic training stage, the panel leader carries out a follow-up evaluation to assess the interest and dedication of each panelist and on their sensory capacities and correct application of the sensory methodology, recording these in the laboratory record. The judges who have a success rate of at least 75% out of the total number of tests in duplicate are considered to have passed the basic training stage. Those who do not achieve this success rate may repeat once the tests not passed. Finally, the labora-

430

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe Table 4 List of defects in appearancea Shape Blowing Low Convex Sunken Inclined Rind Colouring Not developed Cracks Extraneous agents Marks Mouldy Dirty Paste colour Excess of rind halo White Dark Irregular Internal moulds Crystals Holes Caverns Large Cracks Badly distributed Numerous Rounded

tory management personnel decides, which of the candidates continue in future tests, based upon interest, availability and training results. SAL considers these judges (food panelists) to be qualied to conduct simple standardized sensory analysis, basically discriminating tests, and to participate in the development of descriptive sensory methods for specic foods. 3.3. Specic training of panelists: PDO Idiazabal cheese as example So that a panelist can evaluate the sensory quality of Idiazabal cheese, he must be given specic training and

References: catalogue of photographs.

attain the qualication of expert judge in a specic product (Idiazabal cheese).

Table 5 Denition and references for characteristics of odour, avour and aftertaste Characteristic Odour Sheeps milk Sharp Pungent Acid Sweet Smoky (only smoked cheeses) Flavour Sheeps milkd Natural rennetd Pungent Sweet Acid Denition Smelly sensation frequently associated with acidied sheeps milk Sensation of penetration in the nasal cavity Sensation perceived inside the nasal cavity in the form of irritation, burning or stinging Sensory property of pure or mixed substances, the olfactation which is reminiscent of an acidic sensation Sensory property of pure or mixed substances, the olfactation which is reminiscent of a sweet sensation Odour produced following combustion of wood Reference 15 mL acidied raw sheeps milk 36 h at room temperaturec Add 5 mL H2Oa to 40 lL propionic acid Dissolve 80 lL isobutyric acid to 10 mL with H2Ob Dissolve 50 lL acetic acid to 10 mL with H2Ob Dissolve 0.003 g maltol to 10 mL with H2Ob Dissolve 5 lL commercial phenolic type liquid smoke aroma to 10 mL with H2Ob 15 mL acidied raw sheeps milk 36 h at room temperaturec Add 50 mL H2O a 0.5 g rennet paste. Stir for 15 min and lterc Add boiling 100 mL H2O to 0.5 g cayenne pepper seeds. Leave for 5 min and lter. 1 mL supernatant is added to 100 g curd To 1.5 g D(+) sucrose are added 10 mL H2O. Add 6 mL BS to 100 g curd To 75 g curd 25 g of natural yoghurt are added

Bitter (absence) Salty Smoky (smoked cheeses)d Aftertaste Continuity

Olfactive-taste sensation frequently associated with acidied sheeps milk Combination of typical aromas of traditional rennet Sensation perceived in the mouth as an irritation, burning or stinging Elemental taste perceived in the mouth with watery solutions of diverse solutions such as sucrose or fructose Elemental taste perceived in the mouth with watery solutions of diverse organic substances such as citric or lactic acid Elemental taste caused by diluted watery solutions of diverse solutions such as quinine or caeine Elemental taste caused by diluted watery solutions of diverse solutions such as sodium chloride Sensation produced following the combustion of wood

Not applicable To 2.35 g anhydrous sodium chloride add 50 mL H2O. Add 2.4 mL BS to 100 g curd Dissolve 5 lL commercial phenolic type liquid smoke aroma to 10 mL with H2Ob Not applicable

Persistent

Notable
a

Olfactive-taste sensation maintained after removing the product that does not dier from the sensations perceived when it was in the mouth Olfactive-taste sensation very close to the sensation perceived when the sample of cheese was in the mouth, remaining localized a certain time Olfactive-taste sensation with high intensity perceived after removing the sample of cheese from the mouth

Not applicable

Not applicable

5 mL of base solution (BS). 10 mL of base solution (BS). c 15 mL of base solution (BS). d References of acidied raw sheeps milk avour, natural rennet and smoky, may not be ingested. They are prepared and presented as odour references.
b

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

431

3.3.1. Training for appearance The panel leader explains the sensory denition and the evaluation method of each of the characteristics (Table 1)
Table 6 Denition and references of texture characteristics Characteristic Elasticity Denition

and defects (Table 4) for the four appearance parameters (shape, rind, paste colour, and eyes). Theoretical training is backed by a specic catalogue of photographs (references)

References Null: soft butter/raw carrot Medium: stued olive High: frankfurter sausage Null: melted cheese Medium: frankfurter sausage High: cooked carrot (5 min) e Null (oury): potato pure Medium (granular): small couscous grain High (coarse): medium couscous grain

Aptitude of a cheese sample to rapidly regain its initial thickness after compression and deformation Resistance of a cheese sample to a very slight opening and shutting of the jaws Perception at the end of the sensation of the dimensions and shape of particles in the sample. E.g., rounded grains (more or less hard when bitten into)

Firmness

Granularity

Table 7 Group of defects and references for odour, avour and aftertaste Defects Odour Acid Vinegar Acid (in excess) Rancid Butyric (in excess) Rancid Denition Smelly sensation associated with white wine vinegar Organoleptic sensation of pure or mixed substances with an olfactation reminiscent of an acid sensation Smelly sensation reminiscent of butyric acid Smelly sensation, characteristic of fatty substances, following deterioration reactions (rancidness) Smelly sensation reminiscent of an unventilated sheep cowshed Smelly sensation reminiscent of substances presenting faeces and fermentation products with enteric bacteria Smelly sensation reminiscent of deteriorating protein rich foods caused by microbial growth, leading to unpleasant odours Elementary taste perceived in the mouth with watery solutions of diverse organic substances such as lactic acid Flavour sensation reminiscent of white wine vinegar Flavour sensation reminiscent of a sheep cowshed Flavour sensation associated with lamb rennet paste in excess Flavour sensation reminiscent of substances present in faeces and fermenting products in enteric bacteria Flavour sensation associated with deteriorating protein rich foods, caused by microbial growth, leading to unpleasant odours Elementary taste produced by diluted watery solutions of diverse substances such as quinine or caeine Abrasive Pungent Irritating action perceived in the mouth and pharynx mucous, possible accompanied by a sensation of warmth Sensation perceived in the mouth as an irritation, burning or stinging Elementary taste produced by diluted watery solutions of diverse substances such as sodium chloride Reference White wine vinegar directly into 50 lL bottle, adding 10 mL H2Oa Dissolve 60 lL acetic acid to 10 mL with H2Oa To 15 lL butyric acid add 10 mL vaseline oila 5 g D.O.P Idiazabal cheese (>12 months ripening). Slice and keep at room temperature for about 6 days 5 g rear sheep wool 5 g sheep dung 5 g putrid cheese. Add 3.5 mL BS to 100 g curd

Animal

Cowshed Faecal Putrid

Flavour/aftertaste Acidic

Acid (in excess)

Add 65 g curd to 35 g natural yoghurt

Vinegar Animal Cowshed Natural rennet (in excess)c Faecalc Putridc


c

To 100 g curd add 3.3 mL white wine vinegar 5 g rear sheep wool To 0.5 g rennet paste add 35 mL H2O. Stir for 15 min and lterb 5 g sheep dung 5 g putrid cheese

Bitter Pungent

To 0.35 g anhydrous caeine add 50 mL H2O. Add 4 mL BS to 100 g curd To 5 mL distilled alcohol add 15 mL H2O. 20 mL BS Add 50 mL boiling H2O to 0.5 g cayenne pepper seeds. Keep for 5 min and lter. 2 mL oating add to 100 g curd To 2.35 g anhydrous sodium chloride add 50 mL H2O. Add 3.5 mL BS to 100 g curd

Salty (in excess)


a b c

10 mL of base solution (BS). 15 mL of base solution (BS). References for avour/aftertaste, natural rennet, faecal and putrid, are not ingested. Prepared and presented as odour references.

432

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

representatives of each of the parameters. Once the panelists acquire sucient theoretical knowledge, they participate in two types of training sessions: 1. Description of appearance: Panelists describe cheeses in qualitative terms, taking into account both the optimum characteristics and possible defects. 2. Evaluation of appearance: quantication on scale (shape, rind, paste colour, and eyes). 3.3.2. Training for odour, avour, aftertaste and texture The panel leader gives a theoretical description of each of the olfactory-taste and texture characteristics and defects. The use of standardized references is an essential point in order to ensure that the panelists memorize these sensations (Tables 58). Once the panelists have acquired the necessary knowledge on odour, avour, aftertaste and texture descriptors and defects, dierent samples of PDO Idiazabal cheese are presented and two types of training session procedures
Table 8 Denition and references for defects in texture Defect Elastic Denition Sample quickly recovers its initial thickness after being compressed and deformed by the thumb Easily deformed when compressed between the tongue and palate Resilient to being cut or compressed when teeth clenched After masticating, rounded, very ne sized particles noted in the mouth After masticating, separated, large sized, ne sized particles noted in the mouth After masticating, separated, large sized particles are noted in the mouth that is more or less hard on the teeth After masticating, angular shaped separated particles noted in the mouth which, when crushed, let out an audible crunch Lack of smoothness and unpleasant sensation perceived as the sample comes into contact with the surface of the mouths soft tissues Forms a paste with saliva and continuously melts giving a perception of presence in the mouth Releases some amount of liquid following mastication Releases a fair amount of liquid following mastication Cohesive when mixed with saliva, adheres to the surface of the soft parts of the mouth Slowly deforms in the mouth before breaking up, without recovering its initial shape Easily breaks into pieces when the sample is compressed with ngers or incisors Absorbs saliva Dissolves in contact with saliva and disappears in the mouth Reference Frankfurter sausage Melted cheese Raw carrot e Potato pure Breadcrumbs Medium grained couscous Sugar

are conducted: qualitative description of characteristics and defects; and quantication on scale of overall parameters (odour, texture, avour and aftertaste). 3.3.3. Training by harmonization tasting Prior to each harmonization session, panelists undergo a refamiliarization (i.e. reminder) session to refresh their knowledge of the dierent characteristics and/or possible defects of PDO Idiazabal cheese using the references. Afterwards, they evaluate the dierent product samples. In these harmonization sessions, panelists discuss openly among themselves, with the intervention of the panel leader as moderator. 3.4. Preliminary individual qualication of panelists: PDO Idiazabal cheese as an example As a preliminary stage of qualication for quality quantication on scale, a study of the discriminating capacity, repeatability and reproducibility for each panelist is carried out in two sessions. The sensory analysis laboratorys (SAL) criteria are the following: Discriminating capacity: a panelist is qualied in terms of discriminating capacity if a compatibility rate greater than 1 is overcome (as a value of uncertainties, the reference value of standard deviations is taken) in 50% of the discriminating parameters of the values in the two sessions as a whole. The compatibility rate is dened as j x1  x2 j I c q 2 I1 I2 2 taking as I1 and I2 the reference values. Repeatability: the standard repeatability deviation is calculated as the average of the standard deviations in the session with each cheese. A person is considered repeatable when the standard repeatability deviations are less than or equal to the reference value in 50% of the total number of parameters analysed. Reproducibility: the standard reproducibility deviation is calculated as the square root of the sum of the variances between sessions plus the variance due to repeatability. The variance between sessions is considered as the variance corresponding to the average of the values recorded in the sessions. A panelist is considered as qualied in reproducibility when in 50% out of the total number of parameters, the standard reproducibility deviation value is less than or equal to the reference value taken. As an example, Table 9 shows the qualication results of one of the judges. During two sessions (eight cheeses per session), panelists evaluated four dierent Idiazabal cheeses (ICh), where two of them were evaluated three times in each session. In this example, results show that this panelist is qualied to reproducibility, repeatability and discriminat-

Soft Hard Floury Sandy Lumpy

Crystals

Coarse

Pippin apple

Melting

Melted cheese

Moist Liquid Doughy Plastic

Boiled white of egg Granny Smith apple Boiled egg yolk Soft toee

Brittle Dry Soluble

Toast Toast Sugar

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe Table 9 Means, standard deviation, reproducibility, repeatability and discriminating capacity of one panelist Shape Rind ICh 2 2 0 ICh 2 2 0 0 0 0 0 0 Paste colour ICh 1 4 0 ICh 1 3 0 0 0 0 0 0 ICh 2 3 0 ICh 2 3 0 0 0 0 0 0 Eyes ICh 1 3 0 ICh 1 3 0 0 0 0 0 0 ICh 2 1 0 ICh 2 2 0 0 0 0 0 0 Texture ICh 1 4 0 ICh 1 5 0 0 0 0 0 0 ICh 2 4 0 ICh 2 4 0 0 0 0 0 0 Odour ICh 1 4 0 ICh 1 4.25 0.5 0 0.50 0.25 0.35 0.43 ICh 2 4 0 ICh 2 3 0 0 0 0 0 0 Flavour ICh 1 4 0 ICh 1 4 0 0 0 0 0 0 ICh 2 3.50 0.58 ICh 2 2.25 0.50 0.58 0.50 0.54 0.05 0.54 Aftertaste ICh 1 4 0 ICh 1 4 0 0 0 0 0 0

433

Panelist: reproducibility and repeatability Day 1 ICh 1 ICh 2 ICh 1 Mean (N = 3) S sample Day 2 Mean (N = 3) S sample S S S S S session 1 session 2 repeatability between session reproducibility 3.25 0.50 ICh 1 3 0 0.50 0 0.25 0.35 0.43 94 94 3 0 ICh 2 4 0 0 0 0 0 0 3.75 0.50 ICh 1 4 0 0.50 0 0.25 0.35 0.43

ICh 2 3.25 0.50 ICh 2 2 0 0.50 0 0.25 0.35 0.43

% Reproducibility % Repeatability

Panelist: discriminating capacity ICh 1 ICh 2 Mean Ic % Discriminating 3.12 0.53 67 3.5

ICh 1 3.87 2.65

ICh 2 2

ICh 1 3.5 0.70

ICh 2 3

ICh 1 3 2.12

ICh 2 1.5

ICh 1 4.5 0.70

ICh 2 4

ICh 1 4.12 0.88

ICh 2 3.5

ICh 1 4 1.59

ICh 2 2.87

ICh 1 4 1.94

ICh 2 2.62

Panel as a whole: discriminating parameters ICh 1 ICh 2 ICh 1 Mean (N = 10) S panelists t calculated t tabulated 3.10 0.22 1.09 2.23 3.36 0.73 3.56 0.72 7.06 2.10 OK

ICh 2 1.61 0.49

ICh 1 3.37 0.46 3.47 2.10 OK

ICh 2 2.43 0.72

ICh 1 3.07 0.33 11.97 2.10 OK

ICh 2 1.29 0.33

ICh 1 3.77 0.57 1.49 2.10

ICh 2 3.45 0.38

ICh 1 4.27 0.42 2.35 2.10 OK

ICh 2 3.82 0.43

ICh 1 3.77 0.58 3.45 2.10 OK

ICh 2 2.96 0.46

ICh 1 3.86 0.76 3.25 2.14 OK

ICh 2 2.97 0.40

ICh 1: Idiazabal cheese 1; ICh 2: Idiazabal cheese 2; N: Number of times that every kind of cheese is evaluated during each session; S: Standard deviation; Ic: Compatibility race.

ing capacity. That is, 94% of standard reproducibility and repeatability deviations are less than the 0.5 reference value; and the compatibility rate is greater than 1, form more than 2/3 of the discriminating parameters. In the example (Table 9), shape and texture parameters were not considered discriminating parameters by the panel as a whole. 3.5. Qualication of panelists as experts: PDO Idiazabal cheese as an example Qualication of panelists as experts (quality quantication on scale and defects identication) is done on the basis of their performance history. The competence of panelists in formation is measured after a minimum of 10 sessions with eight cheeses each (80 cheeses). 3.5.1. Qualication for quality quantication on scale A comparison is carried out between the panelist rating and the average rating obtained from the qualied panel as a whole. The panelist is qualied for quality quantication on scale if during 10 consecutive sessions the panelist shows less than 15% of deviations (dierence with the average of

the qualied panel bigger than 1) considering all the parameters evaluated in all the cheeses (eight cheeses per session eight parameters per cheese = 64 comparisons per session). 3.5.2. Qualication for defect identication A comparison is carried out comparing defects identied by each panelist with those defects identied by the qualied panel. The SAL has established that a cheese presents a defect when at least ve out of the seven qualied panelists point this out. It is considered that a panelist is qualied to identify a specic defect if he/she shows a competence according to the following criteria: Number of cheeses with the defect that the panelist has evaluated and has pointed out/total number of cheeses with said defect evaluated by the panelist 100 > 66.6%. For example, when a defect has appeared on three occasions, then the panellist is considered to be qualied when he/she has noted the defect at least twice out of the possible three times. When a panelist is qualied for more than

434

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

66.6% of the total number of valuable defects, he is acknowledged to have the capacity to join the PDO Idiazabal cheese panel as an expert. In order to reinforce their performance, harmonization tasting procedures are undertaken, at two-week intervals to increase the historical experience in identifying defects of the previously qualied panelists. Experience has shown us that at least one year of activity is needed for a panelist to demonstrate sucient historical experience to be considered qualied as an expert for evaluating the quality of this complex specic product. This long period is necessary due to the limitations in food sensory analysis, such as the lack of stable references and sensory fatigue. 3.6. Sensory quality evaluation method: PDO Idiazabal cheese as an example Sensory evaluation is carried out in standardized booths. Sensory sessions usually starting at 10:00 h, and may extend for a maximum of 2 h. In the case of the sensory quality evaluation of Idiazabal cheese, the maximum number of test samples to be evaluated in each session is 8, being the temperature of the samples 16 2 C. Samples are prepared according to the specic procedure published by Lavanchy et al. (1993). Sample replication is not considered in routine tests. When a well-trained panel generates data with considerably lower variability and replications are also desirable, but not essential, and if replications cannot be accommodated, the data is nonetheless still valid (Moskowitz et al., 2003). 3.6.1. Tasting room preparation Prior to each session, checks are made on the temperature (21 2 C) and relative humidity (60 20%) in the booths, and on the correct working order of the taps and warning devices. In each booth, a paper serviette, a plastic cup, a bottle of water with low mineralization and a solid food (apple, biscuit without salt) are placed. 3.6.2. Conditions for panelists Samples are to be analysed by seven of the expert qualied judges for this specic product. In general, a small number of panelists are adequate if they are well-trained (Moskowitz et al., 2003). Prior to sensory evaluation panelists should take the following precautions into account: personal hygiene and clothing, punctuality; to arrive at least 30 min before the test. Panelists should abstain from consuming any strongly avoured products and from using cosmetics with very intense or residual odours, and should also be rested and clear-headed. Panelists should not take part in the test if: pregnant, suering from a headache, nasal congestion or using drugs. If judges nd themselves in any of these conditions, they should advise the panel leader 24 h in advance.

3.6.3. Scales of evaluation Quality evaluation is conducted by taking into account the extent to which each sample complies with the optimum state of PDO Idiazabal cheese in each of the eight parameters analysed (shape, rind, paste colour, eyes, odour, texture, avour and aftertaste). A 7-point category scale is used, where 1 = null, meaning that the sample is considerably removed from the optimum state for each parameter; 7 = maximum, full compliance with the optimum state. To carry out evaluation, each judge has documents in the test booth showing: the optimum states for the dierent parameters (Table 1); the denition, evaluation technique and references used in characteristics (Tables 5 and 6) and defects (Tables 7 and 8) of olfactory-taste and texture; a catalogue of photographs illustrating the optimum characteristics (Table 1) and possible defects (Table 4) of appearance; and the evaluation criterion (Fig. 2). 3.6.4. Procedure Prior to sensory evaluation, a harmonization tasting with one or two samples of PDO Idiazabal cheese is carried out. These samples are to be dierent from the routine samples. In order to avoid problems with order and presentation, this order shall be randomized and dierent for each judge. In the test booths, communication between the judges and the panel leader is done by a system of warning lights. Panelists giving points below the average value (<4 on the scale) mark the identied listed defect/s. Evaluating allows noting down any defect that, despite not being listed, has been identied by any of the judges for any of the samples. These new defects will not be taken into consideration when drawing up the test reports for the current year. In order to include a new defect on the list, the number of judges coinciding on the same defect, in the same cheese and at the same session, should be equal to or greater than 5. Such defects on the list that have not been recorded in any test report in the course of a year are removed from the following years list. Depending on the extent to which the sensory parameters appearing in Table 1 are complied with, and the possible presence of defects, panelists are to give points to the samples in line with the criteria shown in Fig. 2. Test reports are according to ISO 17025 and they are sent to the protected designation of origin (PDO) Idiazabal cheese, including the average points given by the panel for each parameter and any defect detected. Nowadays, these results permit the Regulatory Council of Idiazabal cheese to certify, from the sensory viewpoint, the identity and quality of the products made by 104 registered cheese producers. To do so, this Certication Body establishes the minimum limits for quality in terms of points and the seriousness of the possible defect (not all the parameters are considered equally important). It is important the panelists do not know where the limits laid down by the Designation stand, so that they will not be inuenced by factors of a psychological nature. This system enables producers to

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

435

Tree diagram for ascribing points To be applied for each parameter and each cheese

Do you note-detect any defect to be pointed out?

No

Yes

To consider all the characteristics

Do you consider this/these defect/s to be slight?

To be all present and appropriate?

No

Yes

No

Yes

7
One absent-inappropriate?

3
Do you consider this/these defect/s to be medium?

No

Yes

No

Yes

6
Various absent-inappropriate?

2
Do you consider this/these defect/s to be serious?

No

Yes

Yes

5
Are the majority absent-inappropriate?

Yes

4
Fig. 2. Tree diagram for evaluating sensory quality (evaluation criteria).

detect points to be improved upon in the minimum sensory quality when making their product. Mun oz, Civille, and Carr (1992) discuss dierent approaches to sensory quality assessment. A reasonable compromise between the quality-rating method and a descriptive approach was proposed by Beckley and Kroll (1996). The centrepiece of this procedure is a scale for overall quality. The advantage of this method is its outward simplicity in using an overall rating and recognizing that there are situations where products will not exactly match the gold standard but are still acceptable to ship. In the future, in spite of the open discussion between those in favour or otherwise of sensory quality assessment by experts, accreditation will become the tool for evaluat-

ing the validity of the methods applied in sensory laboratories. 3.7. Validation of the method: PDO Idiazabal cheese as an example In a similar manner to the preliminary individual qualication of panelists for quality quantication on scale, the repeatability, reproducibility and discrimination capacity is considered as the validation criteria of the method. Maximum uncertainties of repeatability and reproducibility acceptable by the SAL should be previously established. These values can be calculated from three sessions with three dierent Idiazabal cheeses, where every panelist

436

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe 0.20 0 0.10 0.43 0.44 0 3.43 0 2.86 0 0.33 0.33 Mean day 1 Mean day 2 S repeatability Sbs S reproducibility 3.14 3.14 3.14 3.43 0 4 0.20 4.28 0.10 0.15 0.17 4.57 4 0.40 2.86 0.20 2.71 0.30 0.27 0.41 2.86 2.71 0 2 0 2.43 0 0.08 0.08 2 2.43 0 3.57 0 3 0 0.25 0.25 3.57 3 3.57 2.86 0.10 3.86 0 3.14 0.05 0.37 0.38 ICh 1: Idiazabal cheese 1; ICh 2: Idiazabal cheese 2; Ss: standard deviation of each session; Sbs: standard deviation between sessions. 3.71 3 0.10 3.71 0.10 2.86 0.10 0.42 0.43 3,43 2.86

evaluates each Idiazabal cheese sample three times. After this study, the SAL have proposed 0.5 (repeatability) and 0.8 (reproducibility) as maximum uncertainties accepted. From these criteria, validation of the method (reproducibility and repeatability) is carried out in three later sessions, in the same way as the uncertainties study was done. The discrimination capacity is analysed in a later session where each panelist evaluates both a sample of Idiazabal cheese and a sample of non-Idiazabal cheese, doing each twice. As an example, Table 10 shows the validation results for the parameter avour. This parameter can be validated because the repeatability and reproducibility values obtained by the qualied panel are less than the maximum uncertainties acceptable proposed by the SAL. Also, the student-t test shows the discriminating capacity of the method for the parameter avour. In a similar manner a historical analysis and validation (repetition of cheeses with defects) has been carried out for the identication of defects by the panel as a whole. Prociency/accreditation is a very current and pressing issue being addressed by sensory professionals worldwide. The SAL validation dossier also includes prociency tests with other laboratories. Prociency testing is the use of inter-laboratory test comparisons as part of the process to assess the ability of laboratories to competently perform test for which accreditation is held. Nowadays, prociency testing schemes in Sensory Science are available (Moskowitz et al., 2003). The rst attempt to address sensory prociency was initiated in 1999 by the PROFISENS European Union project, where technical requirements for conducting sensory prociency tests and the management systems requirements are provided (Lyon, 2001). Other relevant publications of this project are the papers

Aftertaste Flavour Odour Texture Eyes Table 11 Quality control for quantication on scale of sensory parameters Paste colour Rind Shape

Table 10 Example of results of validation (mean of the seven panelists as a whole) for the parameter avour Repeatability and reproducibility ICh 1 Session 1 Session 2 Session 3 3.71 3.86 4 ICh 2 ICh 3 S session 0.142 0.082 0.082 0.103 0.237 0.258 3.43 3.57 4 4 4.14 4 S repeatability S between session S reproducibility

Discriminating capacity Idiazabal cheese Mean (N = 14) S panelists t calculated t tabulated 4.14 0.86 2.95 2.064 OK ICh 1: Idiazabal cheese 1; ICh 2: Idiazabal cheese 2; S: Standard deviation; N: Number of times the sample has been evaluated. Other PDO cheese 3.36 0.50

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

ICh 1 ICh 2 Ss

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

437

by McEwan, Hunter, Van Germert, and Lea (2002b, 2002a) that describe measuring performance panel issues for ranking and prole testing. The SAL has established systematic quality controls for maintaining both method validity and the individual qualication of the panelists. 3.8. Quality control activities: PDO Idiazabal cheese as an example 3.8.1. Maintaining method validity Quality control is conducted by studying the reproducibility and repeatability of the panel as a whole, in order to maintain method validity. To do so, at two monthly intervals, a cheese is analysed, in duplicate, at the same session and afterwards at a subsequent session. The standard deviations for repeatability and reproducibility for each parameter are calculated. These deviations should be less than the SALs maximum acceptable uncertainty for more than 80% of the parameter analysed. As example, Table 11 shows quality control activity for the quantication on scale carried out by the SAL in 2005. This control was positive because both standard deviation for repeatability and reproducibility were less than SALs maximum acceptable uncertainties in 100% of the parameters analysed. On an every two-week basis, quality control for the identication of defects is carried out using a systematic repetition of cheese samples, where defects had been idenTable 12 Individual report of a panelist after a session Data Cheese code 256 Deviation? Panel mean 789 Deviation? Panel mean 123 Deviation? Panel mean 951 Deviation? Panel mean 632 Deviation? Panel mean 701 Deviation? Panel mean Judge Shape 4 4 3 4 4 4 3 3 6 Yes 4 4 4 No. cheeses: No. parameters: No. deviations: % Deviations: No. deviations: % Deviations: Rind 3 4 2 3 3 3 3 3 2 3 3 2 6 48 3 6.25 4 8.3 Paste colour 4 4 2 2 3 3 3 3 2 3 2 1

tied in previous routine sessions. If the results of repetition show fewer than ve members of the panel to identify the defect, then it will be openly discussed to achieve consensus as to whether or not the specic defect is found in the cheese. To back this up, by means of harmonization tasting, cases where panelists were not unanimous (34 panelists noting down the defect) are investigated in order to reduce the frequency of these situations as much as possible. 3.8.2. Follow-up and monitoring the panelists To ensure the individual qualication of the panelists and the panel as a whole, it is essential to follow up and monitor each one of the expert judges and the panel. To do so, the SAL has standardized procedures: Follow-up and routine monitoring of the panelists. An individual report is drawn up comparing the results given by each the judges in each session with the average results from the panel as a whole. In those cases where the standard deviation of the panel is P1, an analysis of the data from panelists responsible of this dispersion is conducted. Panelists who stand at a distance from the rounded o average obtained by the panel as a whole by one unit are noted as a deviation for these parameters. This report is drawn up for each session, and is to be shown to the judges so that they can see where there were deviations. Where a panelist, in a given session,

Eyes 3 3 3 3 3 3 3 3 3 3 2 2

Texture 3 3 3 3 3 3 3 3 4 3 3 3

Odour 5 Yes 3 4 4 3 3 4 4 4 4 3 3

Flavour 4 3 4 4 4 Yes 2 3 3 3 3 3 3

Aftertaste 4 3 5 4 3 2 4 3 4 3 4 3 28 5 23 OK

Panelist Panel

Defects Defects identied by the panel: Defects not identied by the panelist: Defects identied by the panelists: % Identied: 82.14

438

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

presents more than 15% deviations for all the parameters evaluated, one of the following measures is taken: the panelist is interviewed to evaluate any possible causes, re-training on the characteristics aected, temporary withdrawal and/or personal re-qualication. As an example, Table 12 shows the individual (each panelist) report appearance after a session where six cheeses were analysed. Follow-up and routine monitoring of the panel. If during a given session the panel as a whole presents 15% deviations for all the attributes evaluated, then the following measures are to be taken: re-training of the panel and/or temporary withdrawal from participating. In order to ensure maintaining the qualication of defects, a follow-up and monitoring of the rate of defects identied by each panelist at each session is carried out. If for the total of cheeses in a session, the rate of defects noted by a panelist is less than 66.6% of the total number of existing defects (noted by more than ve out of the seven

panelists), then he/she should be subject to: re-training and/or temporary withdrawal. 3.8.3. Panel re-training: specicity of the PDO Idiazabal cheese Idiazabal cheese is a seasonal product, made from January to July. Sensory evaluations are also seasonal (FebruaryOctober). During the period from November to January, panelists are called to carry out some of the specic training activities mentioned above. 3.9. Management system and improving activities All technical activities should be developed in line with a quality management system. The management requirements under the ISO 17025 standard should be reected in a manual and/or in quality procedures (Table 13). This system of documenting and recording makes it possible to demonstrate to the accreditation body that the standards are complied with in the external audits (or third party audits). The system for dealing with internal nonconformities and claims by correctional and/or preventive actions, internal and external audits and by the management reviewing the system, are the most dynamic activities of the quality system management and they make it possible to improve the activities at SAL. 4. Conclusions Ensuring the food quality on a sensory basis is an important corporate goal in a competitive environment. Sensory evaluation is a powerful resource with a great variety of applications in the commercial sector. The standardization and accreditation of sensory evaluation methods for the quality certication of food products is a pressing need, particularly with regard to those with protected designation of origin (PDO). This study proposes a general approach for the development of an accredited sensory method for the quality evaluation of food and beverages, taking into account specic characteristics and possible defects in the products. The work shows the application with a specic product (Idiazabal cheese) for example. Also, it shows the selection, training and qualication required for expert panelists for implementation. An important activity is the systematic control quality to assure the maintenance of expert panelist qualication (individually) and of the previously validated method. In terms of PDO products, this manuscript would make possible to respond to an existing need: to show a work methodology so that the members of expert panels (Sensory Committees) can carry out an objective and independent task. The proposed method allows the application of good sensory practices and ensures consistent data. The approach proposed may be transferred to any type of food as a reference for the accreditation of sensory quality evaluation methods according to ISO 17025, especially

Table 13 Cross-reference table showing ISO 17025 Standard items and the quality manual (QM) and operative quality procedures (OQP) ISO 17025:1999 standard section 1. Object and eld of application 2. Standards for consulting 3. Terms and denitions 4. Management requirements 4.1. Organization 4.2. Quality system management 4.3. Monitoring documents QM 01 02 03 04 OQP

4.4. Reviewing applications oers and contracts 4.5. Subcontracting tests and calibrations 4.6. Out pooling services and supplies 4.7. Customer service 4.8. Claims 4.9. Non-conformities control 4.10. Corrective actions 4.11. Preventive actions 4.12. Records control 4.13. Internal audits

OQP-01: Drafting OQP OQP-02: Drafting NWP* OQP-03: Monitoring documents OQP-08: Acquisition and inventory of supplies and evaluation of suppliers

OQP-12: Customer service and claims management OQP-10: Treatment of nonconformities, correctional and preventive actions

OQP-06: Control of internal records and test reports OQP-11: Conducting internal audits and management reviewing the system

4.14. Reviews by the management


*

NWP: normalized work procedure (technical procedure).

rez Elortondo et al. / Food Quality and Preference 18 (2007) 425439 F.J. Pe

439

for sensory certication activities. Each product should dene its quality standard and, on the basis of this standard, the sensory specialist should carry out the sensory evaluation using a panel specically trained for this purpose. The approach is well suited to standard commodities where minimum levels of quality could be ensured, but excellence is rarely the issue. This proposal does not exclude applying other sensory analysis methods for other purposes (hedonic and descriptive tests) to complement it. Quality evaluation methods should take the data derived from the application of_hedonic and descriptive tests. Acknowledgements Financial support of the work was provided by Consejo n de Origen Queso Idiazabal, Regulador de la Denominacio the Departamento de Agricultura y Pesca del Gobierno s Vasco/Euskal Herriko Vasco and the Universidad del Pa Unibertsitatea (project 9/UPV 00042.125-15317/2003 and Quality Chair nancement). Expert Members of the Tasting Committee of the Regulating Council of the Designation of Origin are thanked for their enthusiastic participation in this study. References
Beckley, J. P., & Kroll, D. R. (1996). Searching for sensory research excellence. Food Technology, 50, 6163. Bertozzi, L., & Panari, G. (1993). Cheeses with appellation dOrigine e (AOC): Factors that aect quality. International Dairy le Contro Journal, 3, 297312. DOCE (1992). Reglamento 92/2081/CEE del Consejo relativo a la n de las indicaciones geogra cas y de las denominaciones proteccio colas y alimenticios, Diario Ocial de de origen de los productos agr las Comunidades Europeas de 14 de julio de 1992. a para la acreditacio n de laboratorios de ana lisis ENAC (2003). Gu sensorial. G-ENAC-02, Rev. 1. Octubre 2003. Entidad Nacional de n, Madrid, Espan Acreditacio a. I.S.O. 5495 (1983). Sensory analysis. Methodology. Paired comparison ` ve, test. International Organization for Standardization, Gene Switzerland.

I.S.O. 4120 (1983). Sensory analysis. Methodology. Triangular test. ` ve, Switzerland. International Organization for Standardization, Gene I.S.O. 6564 (1985). Sensory analysis. Methodology. Flavour prole ` ve, methods. International Organization for Standardization, Gene Switzerland. I.S.O. 6658 (1985). Sensory analysis. Methodology. General guidance. ` ve, Switzerland. International Organization for Standardization, Gene I.S.O. 10399 (1991). Sensory analysis. Methodology. Duo-trio test. ` ve, Switzerland. International Organization for Standardization, Gene I.S.O. 8586 (1993). Sensory analysis. General guidance for the selection, training and monitoring of assessors. Part 1: Selected assessors. ` ve, Switzerland. International Organization for Standardization, Gene I.S.O. 1036 (1994). Sensory analysis. Methodology. Texture prole. ` ve, Switzerland. International Organization for Standardization, Gene I.S.O. 17025 (1999). General requirements for the competence of testing and calibration. International Organization for Standardization, ` ve, Switzerland. Gene Issanchou, S., Schlich, P., & Lesschaeve, I. (1997). Sensory analysis: methodological aspects relevant to the study of cheese. Lait, 77, 512. rodier, F., Zannoni, M., Noel, Y., Adamo, C., Squella, Lavanchy, P., Be valuation sensorielle de la texture des J., & Herrero, L. (1993). Le ` pa te dure ou semi-dure. Etude interlaboratoires. Lebenfromages a smittel Wissenschaft und Technology, 26, 5968. Lawless, H. T., & Heymann, H. (1998). Sensory evaluation of food: principles and practices. New York: Chapman and Hall. Lyon, D.H. 2001. International guidelines for prociency testing in sensory analysis. Guideline No. 35. CCFRA (Campden & Chorleywood Food Research Association), Chipping Campden, GL55 6LD, United Kingdom. McEwan, J. A., Heinio, R., Hunter, E. A., & Lea, P. (2002a). Prociency testing for sensory ranking panels: measuring panel performance. Food Quality and Preference, 14, 247256. McEwan, J. A., Hunter, E. A., Van Germert, L. J., & Lea, P. (2002b). Prociency testing for sensory prole panels: measuring panel performance. Food Quality and Preference, 13, 181190. Moskowitz, H. R., Mun oz, A. M., & Gacula, M. C. (2003). In viewpoints and controversies in sensory science and consumer product testing. Trumbull, Connecticut: Food and Nutrition Press, Inc. Mun oz, A. M., Civille, G. V., & Carr, B. T. (1992). Sensory evaluation in quality control. New York: Van Nostrand Reinhold. rez Elortondo, F. J. (1993). Actividades te cnicas y de control de la Pe n de Origen Idiazabal. Sustrai, 28, 4651. Denominacio Perez Elortondo, F. J., Barcenas, P., Casas, C., Salmeron, J., & Albisu, M. (1999). Standardization of sensory methodologies: some applications to protected designation of origin cheeses. Sciences des Aliments, 19, 543558.

You might also like