ASTM E1885-04(2011)
(Test Method)Standard Test Method for Sensory AnalysisTriangle Test
Standard Test Method for Sensory Analysis<char: emdash>Triangle Test
SIGNIFICANCE AND USE
This test method is effective for the following test objectives:
To determine whether a perceivable difference results or a perceivable difference does not result, for example, when a change is made in ingredients, processing, packaging, handling or storage; or
To select, train and monitor assessors.
This test method itself does not change whether the purpose of the triangle test is to determine that two products are perceivably different versus that the products are not perceivably different. Only the selected values of pd, α, and β change. If the objective of the test is to determine if there is a perceivable difference between two products, then the value selected for α is typically smaller than the value selected for β. If the objective is to determine if the two products are sufficiently similar to be used interchangeably, then the value selected for β is typically smaller than the value selected for α and the value of pd is selected to define “sufficiently similar.”
SCOPE
1.1 This test method covers a procedure for determining whether a perceptible sensory difference exists between samples of two products.
1.2 This test method applies whether a difference may exist in a single sensory attribute or in several.
1.3 This test method is applicable when the nature of the difference between the samples is unknown. It does not determine the size or the direction of the difference. The attribute(s) responsible for the difference are not identified.
1.4 Compared to the duo-trio test, the triangle test can achieve an equivalent level of statistical significance with fewer assessors. For details on how the triangle test compares to other three-sample tests, see Refs (1), (2), (3) and (4).
1.5 This test method is applicable only if the products are homogeneous. If two samples of the same product can often be distinguished, then another method, for example, descriptive analysis, may be more appropriate.
1.6 This test method is applicable only when the products do not cause excessive sensory fatigue, carryover or adaptation.
1.7 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this standard to establish appropriate safety and health practices and determine the applicability of regulatory limitations prior to use.
General Information
Relations
Standards Content (Sample)
NOTICE: This standard has either been superseded and replaced by a new version or withdrawn.
Contact ASTM International (www.astm.org) for the latest information
Designation: E1885 − 04 (Reapproved 2011)
Standard Test Method for
Sensory Analysis—Triangle Test
This standard is issued under the fixed designation E1885; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
1. Scope E456 Terminology Relating to Quality and Statistics
E1871 Guide for Serving Protocol for Sensory Evaluation of
1.1 This test method covers a procedure for determining
Foods and Beverages
whether a perceptible sensory difference exists between
2.2 ISO Standard:
samples of two products.
ISO 4120 Sensory Analysis – Methodology – Triangular
1.2 This test method applies whether a difference may exist 4
Test
in a single sensory attribute or in several.
3. Terminology
1.3 This test method is applicable when the nature of the
difference between the samples is unknown. It does not
3.1 Definitions–For definition of terms relating to sensory
determine the size or the direction of the difference. The
analysis, see Terminology E253, and for terms relating to
attribute(s) responsible for the difference are not identified.
statistics, see Terminology E456.
1.4 Compared to the duo-trio test, the triangle test can
3.2 Definitions of Terms Specific to This Standard:
achieve an equivalent level of statistical significance with
3.2.1 α (alpha) risk—probability of concluding that a per-
fewer assessors. For details on how the triangle test compares
ceptible difference exists when, in reality, one does not. (Also
to other three-sample tests, see Refs (1), (2), (3) and (4).
known as Type I Error or significance level.)
1.5 This test method is applicable only if the products are
3.2.2 β (beta) risk—probability of concluding that no per-
homogeneous. If two samples of the same product can often be
ceptible difference exists when, in reality, one does. (Also
distinguished, then another method, for example, descriptive
known as Type II Error.)
analysis, may be more appropriate.
3.2.3 p —probability of a correct response.
c
1.6 This test method is applicable only when the products
3.2.4 p (proportion of discriminators) —proportion of the
d
do not cause excessive sensory fatigue, carryover or adapta-
population represented by the assessors that can distinguish
tion.
between the two products.
1.7 This standard does not purport to address all of the
3.2.5 product—material to be evaluated.
safety concerns, if any, associated with its use. It is the
3.2.6 sample—unit of product prepared, presented, and
responsibility of the user of this standard to establish appro-
evaluated in the test.
priate safety and health practices and determine the applica-
3.2.7 sensitivity—general term used to summarize the per-
bility of regulatory limitations prior to use.
formance characteristics of the test. The sensitivity of the test
2. Referenced Documents
is rigorously defined, in statistical terms, by the values selected
for α, β, and p .
d
2.1 ASTM Standards:
E253 Terminology Relating to Sensory Evaluation of Mate- 3.3 triad—three uniquely coded samples given to an asses-
rials and Products sor in the triangle test; two samples are alike (that is, of one
product) and one is different (that is, of the other product).
This test method is under the jurisdiction ofASTM CommitteeE18 on Sensory
4. Summary of Test Method
Evaluation and is the direct responsibility of Subcommittee E18.04 on Fundamen-
tals of Sensory.
4.1 Clearly define the test objective in writing.
Current edition approved Aug. 1, 2011. Published August 2011. Originally
approved in 1997. Last previous edition approved in 2004 as E1885 – 04. DOI:
4.2 Choose the number of assessors based on the level of
10.1520/E1885-04R11.
sensitivity desired for the test. The sensitivity of the test is, in
The boldface numbers given in parentheses refer to a list of references at the
part, a function of two competing risks: the risk of declaring
end of the text.
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM
Standards volume information, refer to the standard’s Document Summary page on Available fromAmerican National Standards Institute (ANSI), 25 W. 43rd St.,
the ASTM website. 4th Floor, New York, NY 10036, http://www.ansi.org.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
E1885 − 04 (2011)
the samples different when they are not (that is,α-risk) and the consider the inclusion of samples during training that demon-
risk of not declaring the samples different when they are (that strate its presence and absence. Such demonstration will
is,β-risk).Acceptable values ofα andβ vary depending on the increase the panel’s acuity for the taint but may detract from
test objective and should be determined before the test (see other differences. See STP 758 for details (8). Allow adequate
Appendix X3). time between the exposure to the training samples and the
actual triangle test to avoid carryover.
4.3 Assessors receive a triad and are informed that two of
the samples are alike and that one is different. The assessors 7.4 During the test sessions, avoid giving information about
report which they believe to be the different, or “odd,” sample, product identity, expected treatment effects, or individual
even if the selection is based only on a guess. performance until all testing is complete.
4.4 Results are tallied and significance determined by ref- 7.5 Pooling multiple evaluations by the same assessor is not
erence to a statistical table. recommended because results are less representative of the
population and the risk of incorrect conclusion is greater.
5. Significance and Use
8. Number of Assessors
5.1 This test method is effective for the following test
objectives: 8.1 Choose the number of assessors to yield the level of
5.1.1 To determine whether a perceivable difference results sensitivity called for by the test objectives. The sensitivity of
or a perceivable difference does not result, for example, when the test is a function of three values: the α-risk, and the β-risk,
a change is made in ingredients, processing, packaging, han- and the maximum allowable proportion of distinguishers, p .
d
dling or storage; or
8.2 Priortoconductingthetest,selectvaluesforα,βand p .
d
5.1.2 To select, train and monitor assessors.
The following can be considered as general guidelines.
5.2 This test method itself does not change whether the 8.2.1 For α-risk: A statistically significant result at:
purposeofthetriangletestistodeterminethattwoproductsare 8.2.1.1 10 to 5 % (0.10 to 0.05) indicates “slight” evidence
perceivably different versus that the products are not perceiv- that a difference was apparent;
ably different. Only the selected values of p , α, and β change. 8.2.1.2 5 to 1 % (0.05 to 0.01) indicates “moderate” evi-
d
dence that a difference was apparent;
If the objective of the test is to determine if there is a
perceivable difference between two products, then the value 8.2.1.3 1 to 0.1 % (0.01 to 0.001) indicates “strong” evi-
dence that a difference was apparent; and
selected forα is typically smaller than the value selected forβ.
If the objective is to determine if the two products are 8.2.1.4 Below 0.1 % (<0.001) indicates “very strong” evi-
sufficiently similar to be used interchangeably, then the value dence that a difference was apparent.
selected for β is typically smaller than the value selected for α 8.2.2 For β–risk: The strength of the evidence that a
and the value of p is selected to define “sufficiently similar.” difference was not apparent is assessed using the same criteria
d
as above (substituting “was not apparent” for “was apparent”).
6. Apparatus
8.2.3 For p : the maximum allowable proportion of
d
distinguishers, p , falls into three ranges:
d
6.1 Carry out the test under conditions that prevent contact
8.2.3.1 p < 25 % represent small values;
d
between assessors until the evaluations have been completed
8.2.3.2 25 % < p < 35 % represent medium sized values;
d
for example, booths that comply with STP 913 (5).
and
6.2 Sample preparation and serving sizes should comply
8.2.3.3 p > 35 % represent large values.
d
with Practice E1871. See Refs (6) or (7).
8.3 Having defined the required level of sensitivity for the
test using 8.2, use Table A1.1 to determine the number of
7. Assessors
assessors necessary. Enter Table A1.1 in the section corre-
7.1 All assessors must be familiar with the mechanics of the
sponding to the selected value of p and the column corre-
d
triangle test (the format, the task, and the procedure of
sponding to the selected value of β. The minimum required
evaluation). Experience and familiarity with the product and
number of assessors is found in the row corresponding to the
testmethodmayincreasethesensitivityofanassessorandmay
selected value of α. Alternatively, Table A1.1 can be used to
therefore increase the likelihood of finding a significant differ-
develop a set of values for p , α and β that provide acceptable
d
ence. Monitoring the performance of assessors over time may
sensitivity while maintaining the number of assessors within
be useful for increased sensitivity.
practical limits. The approach is presented in detail in Ref (9).
7.2 Chooseassessorsinaccordancewithtestobjectives.For
8.4 Often in practice, the number of assessors is determined
example, to project results to a general consumer population,
by material conditions (for example, duration of the
assessors with unknown sensitivity might be selected. To
increase protection of product quality, assessors with demon-
strated acuity should be selected.
In this test method, the probability of a correct response, p is modeled as p =
d c
α/β α/β
1 p + (1/3) (1-p ), where p is the proportion of the entire population of
7.3 Thedecisiontousetrainedoruntrainedassessorsshould d d d
assessors who can distinguish between the two products. It is a strictly statistical
be addressed prior to testing. Training may include a prelimi-
“guessing model” of the assessor’s behavior. It is not a psychometric model of the
nary presentation on the nature of the samples and the problem
assessor’s decision process, such as the Thurstone-Ura model that could also be
concerned. If the test concerns the detection of particular taint, applied in discrimination testing.
E1885 − 04 (2011)
experiment, number of available assessors, quantity of prod- acceptance, degree of difference, etc. (see Manual 26) (10).A
uct). However, increasing the number of assessors increases comment section asking why the choice was made may be
the likelihood of detecting small proportions of distinguishers. included for the assessor’s remarks.
Thus, one should expect to use larger numbers of assessors
9.7 The triangle test is a forced-choice procedure; assessors
when trying to demonstrate that products are similar compared
are not allowed the option of reporting “no difference.” An
towhenoneistryingtoprovetheyaredifferent.Often18to36
assessor who detects no difference between the samples should
assessors are used when testing for a difference. For compa-
beinstructedtorandomlyselectoneofthesamplesasbeingthe
rable sensitivity when testing for similarity, 42 to 78 assessors
odd one and can indicate that the selection was only a guess in
are needed.
the comments section of the scoresheet.
9. Procedure
10. Analysis and Interpretation of Results
9.1 Prepareworksheetsandscoresheets(seeAppendixX1–
10.1 Use Table A1.2 to analyze the data obtained from a
Appendix X3) in advance of the test so as to utilize an equal
triangletest.Theactualnumberofassessorscanbegreaterthan
numberofthesixpossiblesequencesoftwoproducts,AandB.
the minimum value given in Table A1.1. If the number of
Distribute these at random in groups of six among the
correct responses is greater than or equal to the number given
panelists. The six sequences are:
in Table A1.2, conclude that a perceptible difference exists
ABB AAB ABA
between the samples. If the number of correct responses is less
BAA BBA BAB
thanthenumbergiveninTableA1.2,concludethatthesamples
9.2 Sometimes the final number of assessors does not end
are sufficiently similar.Again, the conclusions are based on the
up as a multiple of six. For example, if a test was planned for
risks accepted when the level of sensitivity (that is, p , α, and
d
36 assessors and only 34 actually participated, there would be
β) was selected in determining the number of assessors.
five complete series of the six sequences and one incomplete
10.2 If desired, calculate a confidence interval on the
set of four in which two of the six triads were randomly
proportion of the population that can distinguish the samples.
dropped.
This method is described in Appendix X4.
9.3 It is critical to the validity of the test that assessors
cannot identify the samples from the way in which they are
11. Report
presented. For example, in a test evaluating flavor differences,
one should avoid any subtle differences in temperature or 11.1 Report the test objective, the results, and the conclu-
appearance caused by factors such as the time sequence of sions. The following additional information is recommended:
preparation. It may be possible to mask color differences using
11.1.1 The purpose of the test and the nature of the
light filters, subdued illumination, or colored serving contain- treatment studied;
ers. Code the serving containers containing the samples in a
11.1.2 Full Identification of the Samples—Origin, method
uniform manner, preferably using three-digit numbers, chosen
of preparation, quantity, shape, storage prior to testing, serving
at random for each test. Prepare samples out of sight and in an
size, temperature. (Sample information should communicate
identicalmanner:sameapparatus,sameservingcontainers,and
that all storage handling, and preparation was done in such a
same quantities of products (see ASTM Serving Protocols).
way as to yield samples that differ only due to the variable of
interest, if at all);
9.4 Present each triad simultaneously if possible, following
11.1.3 The number of assessors, the number of correct
the same spatial arrangement for each assessor (on a line to be
selections, and the result of the statistical evaluation;
sampled always from left to right, in a triangular array, etc.)
Within the triad, assessors are typically allowed to make 11.1.4 Assessors—Age, gender, experience in sensory
testing, with the product, with the samples in the test;
repeated evaluations o
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.