1、ICS 67.240Sensorische Prfverfahren ,A“ ,nicht-A“ PrfungIn keeping with current practice in standards published by the International Organization for Standardization(ISO), a comma has been used throughout as the decimal marker.Ref. No. DIN 10972 : 2003-08English price group 10 Sales No. 011009.04DEUT
2、SCHE NORM August 2003 No part of this translation may be reproduced without the prior permission ofDIN Deutsches Institut fr Normung e.V., Berlin. Beuth Verlag GmbH, 10772 Berlin, Germany,has the exclusive right of sale for German Standards (DIN-Normen).Translation by DIN-Sprachendienst.In case of d
3、oubt, the German-language original should be consulted as the authoritative text.Sensory analysisA or not A testContinued on pages 2 to 8.10972ForewordThis standard has been prepared by Technical Committee Sensorik of the Normenausschuss Lebensmittelund landwirtschaftliche Produkte (Foodstuffs and A
4、gricultural Products Standards Committee). It is equiva-lent to International Standard ISO 8588, but differs in thata) the two different not A test samples provided in the ISO 8588 procedure for exceptional cases havenot been included because the test situations are seldom encountered in practice an
5、d the statisticalevaluation was considered to be too cumbersome,b) it contains a note that the method is unsuitable for equivalence tests, which is important, in particular,if incorrect use of the present standard is to be avoided.ContentsPageForeword . . . . . . . . . . . . . . . . . . . . . . . .
6、. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Normative references . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Conc
7、epts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8、. . . . . . . . . . . . . . . . . . . . . . . . . 26 Assessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Test room and test equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Preparation of test portions .
9、 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10、 . . . 411 Test report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Appendix A Critical values for test statistics . . . . . . . . . . . . . . . . . . . . . . . . . 5Appendix B Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11、. . . . . . . . . . . . . . . 5Appendix C Specimen response forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Page 2DIN 10972 : 2003-081 ScopeThis standard specifies a method of detecting a difference between two test samples and, in particular, maybe used as:a) a difference test fo
12、r test samples whose attributes are subject to variation;b) a recognition test for determining whether assessors are able to detect the deviation of an attribute froma known attribute in a test sample;c) a test for recognizing new attributes in addition to those already known;d) a perception test fo
13、r determining the sensitivity of an assessor to a particular attribute.This method is not suitable for equivalence tests.2 Normative referencesThis standard incorporates, by dated or undated reference, provisions from other publications. These norma-tive references are cited at the appropriate place
14、s in the text, and the titles of the publications are listed below.For dated references, subsequent amendments to or revisions of any of these publications apply to thisstandard only when incorporated in it by amendment or revision. For undated references, the latest edition ofthe publication referr
15、ed to applies.DIN 10950-1 Sensory analysis ConceptsDIN 10956 Equipment for sensory analysis Universal testing glasses and lids Requirements andinstructions for useDIN 10960 Equipment for sensory analysis Wine-tasting glassesDIN 10962 Areas for sensory analysis Specification of test roomsISO 5492 : 1
16、992 Sensory analysis VocabularyISO 8588 : 1987 Sensory analysis Methodology “A” “not A” test1 Fisher, R. A., Statistical methods for research workers, Edinburgh: Oliver b) test portion size;c) number of test portions;d) temperature of test portions;e) masking of individual attributes.The test portio
17、ns shall be served in the form in which they are normally consumed. The assessor shall not beable to draw any conclusions about the origin of the sample from the manner of presentation and the testportions shall be served in an identical way.8.3 Test portion codingThe test portions shall preferably
18、be coded using three-figure random numbers and shall be re-coded beforeevery test.9 ProcedureThe A or not A test assumes an explicit familiarization with test sample A and, after testing has started, theassessor will no longer have access to test sample A.The order in which the A and not A test port
19、ions are served shall be random and different for every assessor,each of whom shall receive the same number of A and not A test portions, but it is not necessary for thenumber of A and not A test portions to be the same.The actual number of A and not A test portions shall not be disclosed to the ass
20、essors.The assessors shall taste the test portions one after the other in the order served and decide whether each testportion is A or not A.Depending on the objective of the test, the method may be varied as follows:a) the initial familiarization may cover not only A test portions, but also not A t
21、est portions;b) the A test portion may be freely available throughout the test.The time interval between serving the test portions will depend on their nature, but shall be constant to avoidinterfering effects due to sensory adaptation.Page 4DIN 10972 : 2003-0810 EvaluationThe results shall be evalu
22、ated by summarizing the correct and incorrect assignments of A and not A testportions in a table (see table 1).Table 1: Summary of resultsTest portion servedTotalA not AAssessorsA n11n12n1.identificationnot A n21n22n2.Total n.1n.2nwherea) n11and n22are the numbers of correct A and not A responses, r
23、espectively;b) n21and n12are the numbers of incorrect A and not A responses, respectively;c) n1.and n2.are the sums of the correct responses in the A and not A rows, respectively;d) n.1and n.2are the sums of the correct A column and not A column responses, respectively;e) nis the sum of all the resp
24、onses.The total number of (correct or incorrect) A and not A responses shall be calculated as follows:n1.= n11+ n12and n2.= n21+ n22(1) (2)The total number of A test portions or not A test portions shall furthermore be calculated as follows:n.1= n11+ n21(3)n.2= n12+ n22(4)and the sum of all the resp
25、onses shall be calculated using:n= n1.+ n2.(5)To interpret the results, the number n11of correctly classified A test portions shall be compared with theexpected number, n1.n.1/n, that would result if A and not A were equal, this being done using a normaldistribution test. The test parameter, U, shal
26、l be calculated using the following equation:n11 n1.n.1/nU = (6)1n1.n2.n.2n.1/(n)3This parameter shall be compared with the critical value, za, of the standard normal distribution associated witha given error probability. If U is greater than za, there is a sensory difference between A and not A. Va
27、luesof zawill be found in Appendix A.Evaluation examples will be found in Appendix B.The use of parameter U requires a sufficiently large number of responses. If one of the numbers n11, n21, n12and n22is less than 10, a finite population correction shall be made, test parameter UTbeing substituted f
28、or U,calculated as follows:n11 n1.n.1/n 0,5UT= (7)1n1.n2.n.2n.1/(n)3This shall be compared with the critical values of the normal distribution.If the number of responses is small, it is advisable to use Fishers exact probability test 1 instead of parameterU or UTand the normal distribution.NOTE: Acc
29、ording to ISO 8588, the evaluation is performed using the one-sided x2test. The x2value is thesquare of the U value for four-fold tables such as tables B.1, B.2 or B.3. Table B.3 yields the same x2valueas table B.1, but this is not significant. The sign of the parameter n11n22 n12n21, which is negat
30、ive in thiscase, is to be taken into account. The one-sided test rejects only if n11n22 n12n21is positive. Using the Uvalue yields the same results as the one-sided x2-test, but it appears more transparent. See, for example,2 for information on one-sided hypotheses relating to four-fold tables.Page
31、5DIN 10972 : 2003-0811 Test reportThe test report shall refer to this standard and include the following details:a) relevant standards;b) object of test;c) type of test sample;d) influencing variables to be specified or agreed as required by this standard (e.g. sample treatment,temperature);e) numbe
32、r of test portions;f) number of reference samples (if any);g) particular test instructions;h) qualification of assessors, and their number;i) any deviation from the specifications of this standard;j) test results and conclusions for a chosen error probability;k) name of supervisor;l) date of testing
33、;m) signature.Appendix ACritical values for test statisticsTable A1 gives the critical values zathat indicate a significant difference for a chosen error probability if the valueof U or UT, or x2, exceeds the value given in table A.1.Table A.1: zavaluesLevel of significance10 % 5% 1 %zavalues 1,28 1
34、,65 2,33Appendix BExamples of application of the A or not A testB.1 Example 1Distinguishing a sweetener (test sample not A) from sugar (test sample A).The two substances are tasted in an aqueous solution whose sweetness is equal to that of a 40 g/l sugarsolution.Number of assessors: 20.Number of tes
35、t portions per assessor: 5 A and 5 not A test portions.Table B.1 shows the results for all the assessors.Table B.1: Results for example 1Test portion servedTotalA not AAssessorsA 60 35 95identificationnot A 40 65 105Total 100 100 200Page 6DIN 10972 : 2003-08This yields the following value for U:60 9
36、5 100/200U = = 3,54 (B.1)1100 100 95 105/(200)3For a chosen error probability, a, of 5%, the calculated value of U is greater than the zavalue (1,65) and thereis consequently a significant difference between the two sweet-tasting solutions.If the U value were less than the critical value, there woul
37、d be no significant difference between the twosolutions, as shown in table B.2.Table B.2: Example of an insignificant differenceTest portion servedTotalA not AAssessorsA 60 50 110identificationnot A 40 50 90Total 100 100 200Test portion servedTotalA not AAssessorsA 35 60 95identificationnot A 65 40
38、105Total 100 100 200Test portion servedTotalA not AAssessorsA 50 55 105identificationnot A 30 65 95Total 80 120 200In this case, the value of U is 1,42 and therefore less than 1,65. The difference is consequently not significantat the 5% error probability.Table B.3 shows another example of a differe
39、nce that is not significant.Table B.3: Further example of an insignificant differenceIn this case, the value of U is 3,54, which is less than 1,65. Consequently, the difference is not significant atthe 5% error probability.B.2 Example 2Identical to example 1, but each assessor assesses A four times
40、and not A six times. The results are shownin table B.4.Table B.4: Results for example 2The value of U is 2,31 and the difference is significant at the 5% error probability.B.3 Example 3Identical to example 1, but with different numbers of A and not A portions and individual evaluation ofassessors.Ea
41、ch assessor assesses 32 test portions, 13 being A and 19 being not A. In addition, the evaluation isperformed not as a total result but individually for each assessor. This would be useful, for example, in trainingassessors.Table B.5 shows the result for assessor 1.Page 7DIN 10972 : 2003-08Table B.5
42、: Result for assessor 1Test portion servedTotalA not AAssessorsA 8 6 14identificationnot A 5 13 18Total 13 19 32In this case, the number of responses is too low, the finite population correction has to be applied (see notein clause 10). The corrected value of U, UT, is therefore:8 14 13/32 0,5UT= =
43、1,31 (B.2)113 19 18 14/(32)3This is less than the critical value of 1,65 and, at the 5% error probability, the difference between A and notA is not significant.The uncorrected U value would be 1,68 and would therefore be significant at the 5% error probability.Appendix CSpecimen response formsForms
44、C.1 and C.2 may be duplicated for further use.C.1 Form for when only test sample A is served in advanceTest sampleDateAssessor1) Taste a test portion of test sample A and return the container to the supervisor. Take the coded testportions.2) The coded test portions consist of A and not A in a random
45、 order. All the not A test portions areidentical.3) Taste the coded test portions in the order presented and record your verdicts below.Remarks:.Test portion code.The test portion isA not APage 8DIN 10972 : 2003-08C.2: Form for when both test sample A and test sample not A are served in advanceTest
46、sampleDateAssessor1) Taste a test portion of A and not A test samples and return the containers to the supervisor. Takethe coded test portions.2) The coded test portions consist of A and not A in a random order. All the not A test portions areidentical.3) Taste the coded test portions in the order presented and record your verdicts below.Remarks:.Test portion code.The test portion isA not A
