This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles

for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.

Designation: C670 − 15

Standard Practice for

Preparing Precision and Bias Statements for Test Methods
for Construction Materials1
This standard is issued under the fixed designation C670; 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.
This standard has been approved for use by agencies of the U.S. Department of Defense.

1. Scope* 2. Referenced Documents

1.1 The Form and Style for ASTM Standards requires that 2.1 ASTM Standards:2
all test methods contain statements on precision and bias. C802 Practice for Conducting an Interlaboratory Test Pro-
Further, the precision statement is required to contain a gram to Determine the Precision of Test Methods for
statement on single-operator precision (repeatability) and a Construction Materials
statement on multilaboratory precision (reproducibility). This C1067 Practice for Conducting a Ruggedness Evaluation or
practice provides guidance for preparing precision and bias Screening Program for Test Methods for Construction
Materials
statements that comply with these requirements. Discussion of
D6607 Practice for Inclusion of Precision Statement Varia-
the purpose and significance of precision and bias statements
tion in Specification Limits
for users of test methods is also provided. Examples of
E177 Practice for Use of the Terms Precision and Bias in
precision statements that conform to this practice are included ASTM Test Methods
in Appendix X1. This practice supplements Practice E177 and E456 Terminology Relating to Quality and Statistics
has been developed to meet the needs of ASTM Committees E691 Practice for Conducting an Interlaboratory Study to
dealing with construction materials. Determine the Precision of a Test Method
NOTE 1—Although this practice is under the jurisdiction of Committee
C09, the current version was developed jointly by Committees C01 and 3. Terminology
C09 and has subsequently been adopted for use by other committees 3.1 Definitions:
dealing with construction materials.
3.1.1 For definitions of general statistical terms, refer to
1.2 This practice assumes that an interlaboratory study Terminology E456.
(ILS) has been completed in accordance with Practice C802 or 3.2 Definitions of Terms Specific to This Standard:3
Practice E691. The interlaboratory study provides the neces- 3.2.1 test determination, n—the value of a characteristic of
sary statistical values to write the precision and bias state- a single test specimen obtained by a specified test method.
ments.
3.2.1.1 Discussion—The term "replicate" is often used for a
1.3 The system of units for this practice is not specified. test determination.
Dimensional quantities in the practice are presented only in 3.2.2 test result, n—the value of a characteristic of a
examples of precision and bias statements. material obtained by carrying out a specified test method.
1.4 This standard does not purport to address all of the 3.2.2.1 Discussion—A test result may be a single test
safety concerns, if any, associated with its use. It is the determination or the average of a specified number of test
responsibility of the user of this standard to establish appro- determinations, or replicates (see 4.1 for additional discussion).
priate safety and health practices and determine the applica- 3.2.3 identical test specimens, n—test specimens selected at
bility of regulatory limitations prior to use. random and made from a single quantity or batch of material
that is as homogeneous as possible.

1
This practice is under the jurisdiction of ASTM Committee C09 on Concrete
2
and Concrete Aggregates and is the direct responsibility of Subcommittee C09.94 For referenced ASTM standards, visit the ASTM website, www.astm.org, or
on Evaluation of Data (Joint C09 and C01). contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM
Current edition approved June 15, 2015. Published August 2015. Originally Standards volume information, refer to the standard’s Document Summary page on
approved in 1971. Last previous edition approved in 2013 as C670 – 13. DOI: the ASTM website.
3
10.1520/C0670-15. Terms are listed in order of hierarchy beginning with the basic concept.

*A Summary of Changes section appears at the end of this standard

Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
1
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
3.2.3.1 Discussion—In interlaboratory studies of test meth- probability of about 5 % in the normal and correct operation of
ods for fresh cementitious mixtures, a practicable approach for the test method; used as an index of precision of the test
obtaining identical tests specimens is to assemble technicians method.
from different laboratories at one location and test specimens
are made from the same batch of the fresh mixture. For 3.2.6.1 Discussion—The difference limit has been selected
interlaboratory studies of nondestructive test methods, the as the appropriate index of precision in most precision state-
same test specimens can be circulated among participating ments. A difference limit (d2s) indicates the maximum accept-
laboratories, provided the characteristic of interest does not able difference between two results obtained on identical test
change during the time to complete the study. specimens (see 3.2.3.1) under the applicable system of causes
3.2.4 single-operator standard deviation, sr, (or coeffıcient (single-operator or multilaboratory conditions). The (d2s%)
of variation, CVr), n—the standard deviation (or coefficient of limit is the maximum acceptable difference between two test
variation) of test determinations obtained on identical test results expressed as a percentage of their average. These
specimens by a single operator using the same apparatus in the difference limits are calculated by multiplying the appropriate
same laboratory over a relatively short period of time. standard deviation (sr or sR) or coefficient of variation (CVr or
3.2.4.1 Discussion—The single-operator standard deviation, CVR) by the factor 1.96 =2 , which for the purpose of this
or coefficient of variation, is the fundamental statistic under- Practice is taken to be equal to 2.8. In Practice E177, the terms
lying the single-operator indexes of precision. The single- repeatability limit and reproducibility limit are used for these
operator standard deviation, or coefficient of variation, is an difference limits under single-operator and multilaboratory
indication of the variability of a large group of test determina- conditions, respectively.
tions by the same operator on the same material. This value is
obtained from an interlaboratory study and is equal to the 3.2.7 acceptable range, n—the difference between the larg-
pooled standard deviation of test determinations obtained by est and smallest of three or more test determinations or test
the operators. The coefficient of variation (ratio of standard results that is expected to be exceeded with a probability of
deviation to the average expressed as a percentage) is used if about 5 % in the normal and correct operation of the test
the standard deviation is proportional to the level of the method; used as an index of precision of the test method, if
characteristic being measured. The single-operator standard applicable.
deviation, usually considered a property of the test method,
3.2.7.1 Discussion—This index is usually reported in preci-
will generally be lower than the multilaboratory standard
deviation. In Practice E177, the single-operator standard de- sion statements of test methods that define a test result as the
viation is referred to as the repeatability standard deviation, average of three or more determinations. Otherwise, the
and the subscript r is used. In previous versions of Practice difference limit (d2s or d2s%) is used. See 4.3 for additional
C670, the terms one-sigma limit (1s) or one sigma limit in discussion on how to determine this index.
percent (1s%) were used for the single-operator standard
deviation or single-operator coefficient of variation, respec- 4. General Concepts
tively. In some publications, the term within-test standard 4.1 Test Result—The result of a test method may be a single
deviation (or coeffıcient of variation) has been used. The term test determination or the average of two or more test determi-
within-laboratory standard deviation (or coefficient of varia- nations (or replicates). The precision statement of a test method
tion) should not be used for this statistic (see 4.2.3). applies to a test result as defined in the test method and should
3.2.5 multilaboratory standard deviation, sR (or coeffıcient
state clearly this fact.
of variation, CVR), n—the standard deviation or coefficient of
variation of test results obtained with the same test method on 4.1.1 Number of Test Determinations—The number of test
identical test specimens in different laboratories with different determinations required to obtain a test result by a test method
operators using different equipment. must be taken into account when evaluating testing variations.
The statistic used in evaluating single-operator precision is
3.2.5.1 Discussion—The multilaboratory standard based usually on the standard deviation (or coefficient of
deviation, or coefficient of variation, is the fundamental statis- variation) of single test determinations. The single-operator
tic underlying the indexes of precision under multilaboratory
standard deviation (or coefficient of variation) may be used in
conditions. The multilaboratory standard deviation is an indi-
evaluating the acceptable range of test determinations.
cation of the variability of a group of test results obtained by
different laboratories for identical test specimens. The multi- 4.1.2 Test Result Based on Averages of Determinations—
laboratory standard deviation (or coefficient of variation) is For test methods that define a test result as the average of two
usually greater than the single-operator standard deviation (or or more test determinations (or replicates), the fundamental
coefficient of variation), because different operators and differ- statistic is still the standard deviation (or coefficient of varia-
ent apparatus have been used in different laboratories for which tion) of single test determinations. The report of the analysis of
the environments may have differed. In Practice E177, the the interlaboratory study (see 5.2) must include this statistic.
multilaboratory standard deviation is referred to as the repro- The single-operator standard deviation of test determinations
ducibility standard deviation and the subscript R is used. can be used to calculate the standard deviation of a test result
3.2.6 difference limit (d2s or d2s%), n—the difference be- that is the average of multiple determinations and thereby
tween two test results that is expected to be exceeded with a define the maximum acceptable difference between two test

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
2
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
results obtained by the same operator on identical test speci- 4.2.2 Multilaboratory Precision—The multilaboratory stan-
mens. The precision statement may also include the maximum dard deviation (or coefficient of variation) obtained from the
acceptable range of individual determinations that comprise the interlaboratory study provides a measure of the greatest differ-
test result (see 4.3). ence between two test determinations that would be considered
4.1.3 Standard Deviation of an Average—The standard acceptable when properly conducted tests are made by two
deviation of the average of n test determinations obtained from different operators in different laboratories on portions of a
identical specimens taken from the same population is equal to material that are intended to be identical, or as nearly identical
the standard deviation of the individual determinations divided as possible. If results differ by more than the difference limit
by the square root of n. This relationship is valid, however, (d2s) or (d2s%), there is a high probability that one or both
only if the determinations are obtained using identical speci- laboratories are in error or that a difference exists in the
mens. It is not applicable to averages obtained on specimens characteristics of the test specimens used for the tests. In such
made from different batches of cementitious mixtures as cases, retests should be made. If possible, newly drawn test
discussed in 4.2.3. specimens should be used for such retests.
4.2.2.1 If the test method calls for reporting the average of
4.2 Types of Precision—A precision statement meeting the more than one test determination, multilaboratory precision is
requirements of this practice normally contains two main expressed as a maximum allowable difference between aver-
elements: (1) single-operator precision, and (2) multilaboratory ages of such groups obtained by two laboratories (Note 3). In
precision. For test methods that require test results on speci- this case, the multilaboratory standard deviation derived from
mens made from more than one batch, the single-operator, the interlaboratory study is based on the number of replications
multi-batch precision is also included. required to obtain a test result as defined by the test method.
4.2.1 Single-Operator Precision—The pooled, single-
operator standard deviation (or coefficient of variation) of test NOTE 3—Example 5 in Appendix X1 shows an example of this
situation. If a test result is based on tests of specimens made from different
determinations obtained from the interlaboratory study is the batches of the cementitious mixture, the consideration in 4.2.3 apply, and
underlying statistic of the test method. This is used to calculate Example 6 provides an example of this situation.
the greatest difference between two or more determinations
that would be considered acceptable when properly conducted 4.2.3 Single-Operator, Multi-Batch Precision—Some test
repetitive determinations are made on the same material by a methods require reporting the averages of two or more deter-
competent operator. As discussed in 4.1.2, the single-operator minations obtained on specimens from two or more batches
standard deviation (or coefficient of variation) of test determi- made using the same materials. The single-operator, multi-
batch standard deviation is a measure of the variation of the
nations is also used to calculate the greatest acceptable
averages among the batches. This standard deviation will
difference between test results defined as the average of two or
usually be greater than the value obtained by dividing the
more determinations. The single-operator precision provides a
single-operator standard deviation by the square root of the
quantitative guide to acceptable performance by an operator. If
number of determinations used to obtain the average test result
two determinations or test results by the same operator differ
for each batch. This is because the single-operator, multi-batch
by more than the difference limit, (d2s) or (d2s%), or if the
standard deviation includes the batch-to-batch variability. The
range of more than two determinations or test results exceeds
precision statement for this type of test method will include
the values defined in 4.3, there is a high probability that an
three indexes of precision: (1) the single-operator precision, (2)
error has occurred and retests should be made.
the single-operator, multi-batch precision, and (3) the multi-
NOTE 2—It is beyond the scope of this practice to describe in detail laboratory precision. In some test methods, the term within-
what action should be taken in all cases if two test results differ by more laboratory precision has been used. The preferred term,
than the (d2s) or (d2s%) limits or the range of more than two determina- however, is single-operator, multi-batch precision because this
tions exceeds the maximum expected range. Such an occurrence is a is more descriptive of the conditions under which results are
warning that there may have been some error in the test procedure, or
some departure from the prescribed conditions of the test on which the obtained. The single-operator, multi-batch precision statement
limits appearing in the test method are based; for example, faulty or would indicate the acceptable range (or difference limit, if only
misadjusted apparatus or improper conditions in the laboratory. In judging two batches are involved) among batch averages. The advice of
whether or not results are in error, information other than the difference a statistical consultant should be sought in planning the
between two test results is needed. Often a review of the circumstances
interlaboratory study for this type of test method so that the
under which the test results in question were obtained will reveal some
reason for a departure. In this case, the data should be discarded and new necessary statistics can be determined.
test results obtained and evaluated separately. If no physical reason for a 4.2.4 Other Measures of Precision—The elements described
departure is found, retests should still be made, but the original tests in 4.2.1, 4.2.2, and 4.2.3 involve the main systems of causes
should not be ignored. If the second set of results also differs by more than
the applicable limit, the evidence is very strong that something is wrong
that are of interest to users of test methods involving construc-
or that a real difference exists between the specimens tested. If the second tion materials. In cases where other systems of causes apply
set produces a result within the limit, it may be taken as a valid test, but (for example, single-operator-apparatus, multi-day precision;
the operator or laboratory may then be suspected of producing erratic or multi-operator, single-day-apparatus precision), the appro-
results, and a closer examination of the procedures would be in order. If priate statistics for those systems of causes need to be
knowledge about the test method in question indicates that certain actions
may be appropriate in cases where deviant results occur, then such developed and the appropriate combination of modifiers given
information should be included in the test method, but details of how this in Practice E177 should be used to describe those statistics.
should be done will depend upon the particular test method. These should not, however, be taken as the fundamental

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
3
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
precision parameters for the test method. The advice of a may have a minor affect on multilaboratory variability if the
statistical consultant should be sought in planning the inter- between-laboratory component of variance is greater than the
laboratory study so that the correct statistics can be determined. single-operator variance. Because specification limits should
4.3 Acceptable Range Among Results—If the test method be established with consideration of testing variability, it is not
requires more than two test results, as so defined in the method, appropriate to consider the indexes of precision of the test
the difference between highest and lowest test results in the method as tolerances to be added to statistically-derived
group must be compared to the maximum acceptable range for specification limits for the purpose of judging acceptance or
the applicable system of causes. The range among different rejection of materials.
numbers of test results in the group, including two, that would 4.4.2 For Qualifying an Operator—As discussed in 4.2.1,
be expected to be exceeded with no more than about 5 % indexes of single-operator precision are sometimes used as a
probability is obtained by multiplying the appropriate standard basis for qualifying an operator. The assumption is that results
deviation or coefficient of variation by the corresponding factor that do not differ by more than the stated index are indicative
from the second column of Table 1. If more than two test of proper performance of the test. This assumption, however, is
results are obtained, the index of precision for the difference not necessarily correct. Uniform misunderstanding of
between two results cannot be used as a criterion for judging instructions, incorrect specimen preparation, or maladjust-
acceptability of the differences between pairs of results se- ments of equipment may produce consistent but erroneous test
lected from the group. results. Whenever possible, tests conducted for the purpose of
qualifying an operator should be made on materials for which
4.4 Uses of Indexes of Precision the measured characteristic is known, so that bias as well as
4.4.1 In Setting Specification Limits—The indexes of preci- precision can be evaluated. Participation in proficiency sample
sion described in this practice are applicable to test results programs is an effective way to evaluate operator performance
obtained on identical test specimens and provide information among peers.
on the inherent variability of the test method. In routine quality
control or acceptance testing for a project, the variation of the 5. Basis for Precision Statement
test results will be affected by the inherent variability of the test 5.1 In order to be valid, the indexes of precision to be
method, the variability of the materials, and the variability included in the precision statement must be based on estimates
associated with the sampling method. Specifiers need to of the precision of the test method obtained from a statistically
consider these sources of variability in setting specification designed interlaboratory study. Before proceeding with the
limits so as to control the producer’s risk of rejection of a lot interlaboratory study, the ruggedness of the test method should
of acceptable material and the purchaser’s risk of accepting a be investigated in accordance with Practice C1067. A rugged-
lot of deficient material.4 Practice D6607 provides a method- ness evaluation requires the involvement of a few laboratories
ology for setting specification limits that accounts for the and the use of several materials that encompass the range of the
inherent variability of the test method along with the material level of the characteristics to be measured by the test method.
variability. The variability associated with the test method may This evaluation will provide a preliminary estimate of single-
be reduced by requiring a test result to be the average of two operator precision and indicate whether tighter tolerances may
or more test determinations. A balance, however, needs to be be needed for key aspects of the test method. The interlabora-
achieved between the incremental cost of additional testing and tory study, on the other hand, must involve a sufficient number
the corresponding incremental reduction in uncertainty. Also, of laboratories, materials, and replicate measurements so that
increasing the number of determinations to obtain a test result the results obtained provide reliable estimates of the precision
of the test method (Note 4). The procedures described in this
practice are based on the assumption that proper estimates of
4
Philleo, R. E., "Establishing Specification Limits for Materials," Cement, precision have been obtained. Practice C802 is a companion
Concrete, and Aggregates, CCAGDP, Vol. 1, No. 2, 1979, pp. 83-87. document that describes how to organize and conduct a
suitable interlaboratory study and how to analyze the data to
TABLE 1 Maximum Acceptable Range of Test ResultsA obtain the relevant estimates of precision.
Multiplier of Standard
Number of NOTE 4—The requirement of “reliable estimates of the precision”
Deviation or Coefficient
Test Results presupposes an estimate obtained from a properly designed and executed
of VaritationB
2 2.8
interlaboratory series of tests involving at least 30 degrees of freedom for
3 3.3 single-operator standard deviation and at least 10 laboratories. See
4 3.6 Practice C802.
5 3.9 5.2 The Form and Style for ASTM Standards requires that
6 4.0
7 4.2 data and details of the interlaboratory study used to determine
8 4.3 precision and bias be filed as a research report at ASTM
9 4.4 International Headquarters.
10 4.5
A
A test result can be a single determination or the average of two or more 5.3 The ASTM International Interlaboratory Study Program
determinations as defined in the test method. (ILS) can support subcommittees in the development of
B
Values were obtained from Table A7 of “Order Statistics and Their Use in Testing precision statements by assisting in the design of an interlabo-
and Estimation,” Vol 1, by Leon Harter, Aerospace Research Laboratories, United
States Air Force. ratory study, distribution of materials or test specimens, data
analysis, and preparation of a draft research report.

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
4
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
5.4 A subcommittee may wish to postpone the organization 6.1.2 Information on Units—Many precision indexes for
of the interlaboratory study until a new test method has been test methods of construction materials are based on data
approved. In such cases, the precision statement of the new test obtained using the inch-pound version of a combined standard
method must include as a minimum the single-operator and these indexes have been converted to SI units. The
standard-deviation (or coefficient of variation) obtained in at following examples provide recommended wording for a note
least one laboratory. Preferably, the standard deviation (or to the precision statement, if applicable, and how the conver-
coefficient of variation) should be obtained by using materials sion should be performed.
with different levels of the characteristic being measured. A 6.1.2.1 Case 1—Precision is stated in terms of a coefficient
ruggedness evaluation in accordance with Practice C1067 can of variation. The precision indexes are non-dimensional and
be a source of data to develop a temporary precision statement. there would be no need for dual presentations. In this case, it
The temporary statement addresses only the single-operator is only necessary to state that the data were obtained in the
standard deviation (or coefficient of variation). This temporary inch-pound system.
precision statement is permitted for five years at which time it Example 1:
needs to be replaced with a complete statement based on an The data used to develop the precision statement were
interlaboratory study. See Example 9 in Appendix X1. obtained using an earlier inch-pound version of this test
5.5 When an approved test method is being revised, the method.
responsible subcommittee should determine whether the pro- 6.1.2.2 Case 2—For a combined standard in which both
posed change(s) to the test method will affect the validity of the systems of units are to be used separately:
precision statement in the existing standard. If the subcommit- Example 2:
tee believes the precision of the method may be affected by the The data used to develop the precision statement were
revision, a new interlaboratory study should be conducted to obtained using the inch-pound version of this test method. The
provide data for updating the precision statement. precision indexes shown in SI units are exact conversions of the
values in inch-pound units.
5.6 For some tests under the jurisdiction of Committees
6.1.2.3 Case 3—For a inch-pound standard that has been
C01, C09, D04, and D18 there may be an extensive database of
converted to an SI standard and the inch-pound units have been
interlaboratory test data obtained from various proficiency
dropped:
sample programs. If such data are available for a particular test
Example 3:
method, a precision statement can be prepared by carrying out
the data analysis described in Practice C802 based upon a The data used to develop the precision statement were
much larger population of data than can normally be assembled obtained using the previous inch-pound version of this test
in an interlaboratory study. Care is needed, however, in method. The indicated precision indexes in SI units are exact
evaluating the data because the requirement for identical test conversions of the previous values obtained originally in
specimens may not be met by data from some proficiency inch-pound units.
sample programs. For example, participating laboratories may 6.2 Manner of Expression
be shipped the dry ingredients to prepare specimens of cemen- 6.2.1 If the interlaboratory study data, which are the basis
titious mixtures for testing. The resulting specimens among the for the precision statement, indicate that the standard deviation
laboratories are not identical test specimens and the resulting is essentially the same for all levels of the characteristic in
multilaboratory precision includes an additional source of question, the precision statement shall be expressed in the units
variation associated with making the test specimens. This of the measured characteristic.
needs to be mentioned in the precision statement. 6.2.2 If the standard deviation is essentially proportional to
the average for different levels of the characteristic in question,
6. Form of Precision Statement that is, the coefficient of variation is essentially constant, the
6.1 Background Information coefficient of variation and difference limit in percent (d2s%)
6.1.1 Description of the Interlaboratory Study—The Form shall be given. The coefficient of variation is determined as the
and Style for ASTM Standards requires that the precision ratio of the standard deviation to the average value of the
statement include a summary of the interlaboratory study that results multiplied by100 %.
will permit the user of the test method to judge the reliability 6.2.3 If neither of these conditions is met, the applicable
of the precision statement. This summary should include the precision limits for specific ranges of the measured character-
number of laboratories, number of materials, range of material istic shall be stated. See Example 3 in Appendix X1.
characteristics measured, and number of test determinations 6.2.4 The phrase “are not expected to differ by more than”
(replicate tests) for each material. The research report (see 5.2) is used to introduce the applicable difference limits (d2s or
should be referenced for the details of the interlaboratory study d2s%). The intent of this wording is to recognize that the
and the data analysis leading to the precision statement. This difference limits are expected to be exceeded in the long run
summary information should be provided in a note. with a probability of about 5 %. For introducing the acceptable
range of more than two results, the corresponding phrase is
NOTE 5—Example 1 in Appendix X1 illustrates the wording that may “the range is not expected to exceed.”
be used in a note to summarize the interlaboratory study. The subcom-
mittee should exercise its discretion in choosing the exact wording for the 6.2.5 The abbreviations (d2s) and (d2s%) are given in a
note as this will depend on the nature of the test method and the actual footnote to the precision statement and reference is made to
interlaboratory study. Practice C670 as shown in the examples in Appendix X1.

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
5
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
6.2.6 If the standard deviation varies erratically or the 7.4 Procedure to Estimate Bias—If it is possible to obtain
coefficient of variation is not constant over the range of the data to determine if bias exists, a two-tailed t-test can be used
characteristic tested, the maximum value of the index of in accordance with 7.4.1 – 7.4.4.
precision shall be used. The word “maximum” shall be used in 7.4.1 Obtain at least 30 results from separate specimens of
the first sentence of the precision statement and the abbrevia- the reference material or the material compounded to a known
tion “max” shall be added as a subscript to the abbreviation for value of the characteristic in question. Calculate the quotient
the difference limit in the footnote, that is, use (d2s)max, or using Eq 1.
(d2s%)max. This form should be used rarely, and then only as a
X̄ 2 X r
last resort. For additional discussion of this situation, see the t5 (1)
s
section on Irregular or Nonlinear Relationship Between Stan-
dard Deviation, Coefficient of Variation and Average Level in =N
Practice C802. Example 4 in Appendix X1 gives an example of where:
this type of precision statement.
Xr = reference value,
X̄ = mean of the measured values,
7. Bias Statement s = standard deviation of the measured values, and
7.1 Introduction—Bias is a systematic error inherent in the N = number of measured values on separate specimens.
test method that contributes to the difference between the mean 7.4.2 The quotient obtained using Eq 1 has a t-distribution
of the test results and an accepted reference or true value. In with N-1 degrees of freedom. Usually, the level of significance
any test method, tolerances are placed on the accuracy of for the t-test, α, is taken to be 0.05; and because a two-tailed
measuring equipment. Tests made with a given set equipment t-test is used, α/2 = 0.025 is used to determine the critical
that has an error within the permitted tolerance will produce t-values. The null hypothesis that no bias exists is rejected if
results with a small consistent bias, but that bias is not inherent the value of t calculated by Eq 1 is less than -tα/2 or greater
to the test method and is not included in the bias statement for than tα/2. For α = 0.05 and N-1 = 29 degrees of freedom, the
the test method. critical t-values for a two-tailed test are 62.045, and the
7.2 Estimating Bias—There are two conditions that permit inequalities for rejecting the null hypothesis are: t< -2.045 or t>
the bias of a test method to be estimated: (1) a standard 2.045. Thus if the calculated value of t for 30 measurements
reference sample with a known value of the characteristic in falls between -2.045 and 2.045, there is no strong evidence to
question has been tested by the test method, and (2) the test reject the null hypothesis and it may be concluded that there is
method has been applied to a sample that has been com- no bias.
pounded in such a manner that the true value of the character- 7.4.3 If the calculated value of t falls in the rejection region,
istic being measured is known, such as may be the case, for it may be concluded that there is a bias in the test method and
example, in a test for cement content of concrete. Judgment is the 95 % confidence limits for the bias are:
required to determine whether a potential reference sample is 2 S
suitable for the purpose. For example, a metal bar of accurately ~ X̄ 2 X r ! 1 t α⁄2 (2)
known physical properties might not be suitable for establish-
=N
ing the bias of a test for the corresponding concrete properties 7.4.4 In some cases, the bias may be a function of level of
because the level of the values may differ by an order of the characteristic being measured. If the differences X̄2X r for
magnitude. If it is possible to examine bias, it is necessary to different levels of Xr are statistically different from each other,
determine whether there are enough data to determine statisti- the above procedure may be applied to each such level. A
cally that the mean of the test results is significantly different different bias may be applicable for different levels.
from the true value. If there is a difference, an absolute measure
of bias cannot be made, but confidence limits can be placed on 7.5 Form of Bias Statement
the bias. 7.5.1 If a study for bias has been made, a bias statement
based on one of the following examples may be used:
7.3 If Bias Cannot be Estimated—For most test methods Example 1—No bias:
there is no reference value available or the characteristic can be Bias—If measured results are compared with accepted
measured only by using that test method. In those cases, a bias reference values (or known values from accurately com-
statement based on one of the following may be used: pounded specimens), the test method is found to have no bias.
Example 1: Example 2—Bias exists:
Bias—This test method has no bias because the values Bias—If measured results are compared with accepted
determined can be defined only in terms of the test method. reference values (or known values from accurately com-
Example 2: pounded specimens), the bias of the test method is found with
Bias—There is no accepted reference material suitable for 95 % confidence to lie between 0.0062 and 0.0071.
determining the bias in this test method, therefore, no statement Example 3—Bias depends on level:
on bias is made. Bias—If measured results are compared with accepted
Example 3: reference values (or known values from accurately com-
Bias—No justifiable statement can be made on the bias of pounded specimens), the bias of the test method is found with
this test method because (insert here the reason). 95 % confidence to lie between -0.0004 and -0.0001 for results

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
6
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
in the range of 6 to 10 and between -0.0006 and -0.0002 for
results in the range of 10 to 15.
8. Keywords
8.1 acceptable range; bias; coefficient of variation; differ-
ence limit; multilaboratory precision; single-operator preci-
sion; standard deviation; test determination; test result

APPENDIX

(Nonmandatory Information)

X1. EXAMPLES OF RECOMMENDED FORMS OF PRECISION STATEMENTS

X1.1 Introduction: Example 2:

Single-Operator Precision—The single-operator coefficient of variation has been
X1.1.1 If proper estimates of precision are available, the found to be 2.5 %. Therefore, results of two properly conducted tests by the
precision statement should be written in the form of the same operator on the same material using the same viscometer are not ex-
appropriate example as given below for the applicable estimate pected to differ from each other by more than 7.0 %A of their average.
Multilaboratory Precision—The multilaboratory coefficient of variation has been
of precision (standard deviation or coefficient of variation) and found to be 5.0 %. Therefore, results of two properly conducted tests by two
corresponding system of causes (single-operator; multilabora- different laboratories on specimens of the same material are not expected to
differ from each other by more than 14 %A of their average.
tory; and if applicable, single-operator multi-batch).
X1.1.2 Some of the following examples are taken from test A
These numbers represent the difference limits in percent (d2s%) as
methods current at the time this practice was first adopted and described in Practice C670.
others are hypothetical.
X1.3 Statements for Which the Precision, Measured by
X1.2 Statements for Which One Estimate of Precision for Either the Standard Deviations, or the Coeffıcient of Variation,
Each System of Causes Applies: is not Constant over the Range of Values of the Characteristic
X1.2.1 The following examples illustrate the simplest and in Question:
most common form for precision statements. The form of X1.3.1 If the precision value applies only over a certain
Example 1 is applicable if the standard deviation is constant, range of the material characteristic being measured, this shall
and Example 2 is applicable if the standard deviation is be indicated by inserting the words “over the range from __ to
proportional to the average value of the measured __,” or “below,” or “above” a certain limit after the words
characteristic, that is, if the coefficient of variation is constant. “standard deviation” or “coefficient of variation” in the first
In both cases, the difference limits apply to single determina- sentence of the statement. If precision limits have been
tions. Example 1 also illustrates the use of a note for summa- obtained for more than one range of the characteristic, separate
rizing the interlaboratory study that forms the basis for the statements shall be written for each range. The applicable range
precision indexes. The exact wording for such a note depends may be indicated in subparagraph headings if separate subpara-
on the nature of the test method and the interlaboratory study. graphs are used. This example also shows a note to explain the
All precision statements should have a note to summarize the precision values for the two systems of units in a combined
source of the precision information, but only Example 1 will standard.
show such a note. Example 3:
Example 1: PrecisionA
Single-Operator Precision—The single-operator standard deviation has been Single-Operator Precision—The single-operator standard deviation has been
found to be 0.45 %. Therefore, results of two properly conducted tests by the found to be 1.4 °F [0.8 °C] for flash points below 220 °F [104 °C] and 7.1 °F
same operator on the same material are not expected to differ by more than [3.9 °C] for flash points above 220 °F [104 °C]. Therefore, results of two
1.3 %.A properly conducted tests by the same operator on the same material are not
Multilaboratory Precision—The multilaboratory standard deviation has been expected to differ from each other by more than 4 °F [2.2 °C]A for flash
points below 220 °F [104 °C] or by more than 20 °F [11.1 °C]B for flash
found to be 0.75 %. Therefore, results of two properly conducted tests by points above 220 °F [104 °C].
two different laboratories on specimens of the same material are not ex- Multilaboratory Precision—The multilaboratory standard deviation has been
pected to differ by more than 2.1 %.A found to be 2.1 °F [1.2 °C] for flash points below 220 °F [104 °C] and 8.8 °F
Note X—These precision statements are based on an interlaboratory study that [4.9 °C] for flash points above 220 °F [104 °C]. Therefore, results of two
involved 10 laboratories, three materials with average mass loss of approx- properly conducted tests on the same material in two different laboratories
imately 10 %, 15 %, and 20 %, and four replicate tests per operator. Sup- are not expected to differ from each other by more than 6 °F [3.3 °C]B for
porting data have been filed at ASTM International Headquarters and may flash points below 220 °F [104 °C] or by more than 25 °F [13.9 °C]B for flash
be obtained by requesting Research Report RR:C09-XXXX. points above 220 °F [104 °C].

A A
These numbers represent the difference limits (d2s) as described in Practice The data used to develop the precision statements were obtained using the
C670. inch-pound version of this test method. The precision indexes shown in SI units are
exact conversions of the values in inch-pound units.

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
7
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
B
These numbers represent the difference limits (d2s) as described in Practice sion defines the maximum expected range of the batch
C670.
averages. The multilaboratory precision defines the maximum
X1.3.2 The following example is for the case where the expected difference in test results (nine determinations) be-
coefficient of variation is not constant or the standard deviation tween two laboratories.
varies erratically over the range of the characteristic tested. The Example 6:
maximum value of the index of precision, in this case the Single-Operator Precision—The single-operator standard deviation of single test
determinations has been found to be 35 psi [0.24 MPa]. The test method re-
coefficient of variation, is used. quires three test determinations per batch. Therefore, the range (difference
Example 4: between highest and lowest) of three test determinations obtained by the
Single-Operator Precision—The maximum single-operator coefficient of variation same operator on specimens made from the same batch is not expected to
has been found to be 4.25 %. Therefore, results of two properly conducted exceed 116 psi [0.80 MPa].A
tests by the same operator on specimens of the same material are not ex- Single-Operator, Multi-Batch Precision—The single-operator, multi-batch stan-
pected to differ from each other by more than 12 %A of their average. dard deviation of batch averages has been found to be 64 psi [0.44 MPa].
The test method requires testing specimens from three batches. Therefore,
A
This number represents the difference limit (d2s%)max as described in the range (difference between highest and lowest) of the batch averages ob-
Practice C670. tained by the same operator from three batches of the same material made
on the same day is not expected to exceed 211 psi [1.45 MPa].A
X1.3.3 See Examples 7 and 8 for alternative tabular forms Multilaboratory Precision—The multilaboratory standard deviation of a test result
has been found to be 78 psi [0.53 MPa]. A test result is defined as the aver-
of precision statements if the standard deviation or coefficient age of the results from three batches with three test determinations per
of variation is not constant for the range of values used in the batch. Therefore, results of two properly-conducted tests in different laborato-
interlaboratory study. ries on the same material are not expected to differ by more than 218 psi
[1.50 MPa].B

X1.4 Statement for Which a Test Result is Defined as the A

These numbers are obtained by multiplying the standard deviation by the value
Average of a Specified Number Determinations: 3.3 taken from Table 1 of Practice C670.
B
This number represents the difference limit (d2s) as described in Practice
X1.4.1 In this example, a test result is the average of three C670.
determinations. The single-operator standard deviation is di-
vided by the square root of three to obtain the standard X1.6 Alternative Form of the Precision Statement—In cases
deviation of a test result. The difference limit (d2s) of two test where separate statements for a number of different materials
results is obtained by multiplying this standard deviation by or a number of different levels of the characteristic are
2.8. Thus the value of d2s is 2.8/√3 = 1.62 times the involved, the precision indexes may be presented in table form
single-operator standard deviation. The acceptable range of the as shown in the following examples:
three test determinations used to obtain a test result is obtained Example 7:
Precision—Table X1.1 provides the precision information for this test method.
by multiplying the single-operator standard deviation by the Column 2 shows the single-operator and multilaboratory standard deviations
applicable factor (3.3) from Table 1. that have been found for the materials described in Column 1. The upper
part of the table is for single-operator conditions and the lower part is for mul-
Example 5:
tilaboratory conditions. The results of two properly conducted tests on the
Single-Operator Precision—The single-operator standard deviation for single
same material are not expected to differ by more than the values given in
test determinations has been found to be 3.5 %. A test result is defined as
Column 3.
the average of three test determinations. Therefore, the results of two prop-
erly conducted tests (each consisting of the average of three single determi-
nations) by the same operator on the same material are not expected to dif- Example 8:
fer by more than 5.7 %.A The range (difference between highest and lowest) Precision—Table X1.2 provides the precision information for this test method.
of the three single determinations used in calculating the average is not ex- Column 2 shows the single-operator and multilaboratory coefficients of varia-
pected to exceed 11.6 %.B tion that have been found for the materials described in Column 1. The upper
Multilaboratory Precision—The multilaboratory standard deviation of a test result part of the table is for single-operator conditions and the lower part is for mul-
has been found to be 2.8 %. A test result is defined as the average of three tilaboratory conditions. The results of two properly conducted tests on the
test determinations. Therefore, results of two properly-conducted tests in dif- same material are not expected to differ by more than the values given in
ferent laboratories on the same material are not expected to differ by more Column 3 expressed as a percentage of their average.
than 7.8 %.A

A
This number represents the difference limit (d2s) as described in Practice
C670. TABLE X1.1 Indexes of Precision
B
Obtained by multiplying the single-operator standard deviation of single test Material Standard Acceptable
determinations by the value 3.3 taken from Table 1 of Practice C670. Deviation Difference
Between
X1.5 Statements for Which a Test Result is Based on Two
ResultsA
Specimens from More Than One Batch: Single-Operator Precision:
X1.5.1 The following example is for a test method that Asphalts, solubility more than 99 %B 0.035 0.10
C
Tars, liquid grades 0.11 0.31
requires replicate determinations on specimens made from Tars, simi-solidC 0.17 0.48
more than one batch of the material. The precision statement, Multilaboratory Precision:
therefore, includes an additional statement on single-operator, Asphalts, solubility more than 99 %B 0.090 0.26
Tars, liquid gradesC 0.22 0.61
multi-batch precision. In the example, a test result is defined as Tars, semi-solidC 0.83 2.34
the average of the results from three batches and three test A
These numbers represent the difference limits (d2s) as described in Practice
determinations are required for each batch. Thus a test result C670.
B
involves nine test determinations. The single-operator preci- Applicable if carbon disulfide, carbon tetrachloride, trichloroethylene, or
benzene is used.
sion defines the maximum expected range of the test determi- C
Applicable if carbon disulfide is used.
nations within a batch. The single-operator, multi-batch preci-

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
8
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.
 C670 − 15
TABLE X1.2 Indexes of Precision X1.7 Temporary Precision Statement—In accordance with
Material Coefficient of Acceptable the Form and Style of ASTM Standards, if the subcommittee
Variation Difference
(Percent of Mean) Between
decides to delay an interlaboratory study, the test method must
Two include a temporary precision statement. The temporary state-
Results ment addresses only the single-operator precision. The differ-
(Percent
of Their ence limit (d2s or d2s%) as defined in 3.2.6 is not included. The
Average)A statement should include information on the average property
Single-Operator Precision:
Asphalt Cements at 275 °F [135 °C] 0.64 % 1.8 %
values of the materials used.
Liquid Asphalts at 140 °F [60 °C]: Example 9:
Below 3000 cSt 0.53 % 1.5 % PrecisionA —The single-operator standard deviation for percent mass loss from
3000 cSt and above 0.71 % 2.0 % a single laboratory has been determined to be 1.3 % for materials with aver-
Multilaboratory Precision: age mass loss ranging from 10 to 25 %.
Asphalt Cements at 275 °F [135 °C] 3.10 % 8.7 %
Liquid asphalts at 140 °F [60 °C]: A
An interlaboratory study of this test method is being conducted and a complete
Below 3000 cSt 1.06 % 3.0 % precision statement is expected to be available by (insert year).
3000 cSt and above 3.11 % 8.7 %
A
These numbers represent the difference limits in percent (d2s%) as described
in Practice C670.

SUMMARY OF CHANGES

Committee C09 has identified the location of selected changes to this standard since the last issue (C670 – 13)
that may impact the use of this standard. (Approved June 15, 2015.)

(1) Revised 5.4 and added X1.7.

ASTM International takes no position respecting the validity of any patent rights asserted in connection with any item mentioned
in this standard. Users of this standard are expressly advised that determination of the validity of any such patent rights, and the risk
of infringement of such rights, are entirely their own responsibility.

This standard is subject to revision at any time by the responsible technical committee and must be reviewed every five years and
if not revised, either reapproved or withdrawn. Your comments are invited either for revision of this standard or for additional standards
and should be addressed to ASTM International Headquarters. Your comments will receive careful consideration at a meeting of the
responsible technical committee, which you may attend. If you feel that your comments have not received a fair hearing you should
make your views known to the ASTM Committee on Standards, at the address shown below.

This standard is copyrighted by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959,
United States. Individual reprints (single or multiple copies) of this standard may be obtained by contacting ASTM at the above
address or at 610-832-9585 (phone), 610-832-9555 (fax), or service@astm.org (e-mail); or through the ASTM website
(www.astm.org). Permission rights to photocopy the standard may also be secured from the Copyright Clearance Center, 222
Rosewood Drive, Danvers, MA 01923, Tel: (978) 646-2600; http://www.copyright.com/

Copyright by ASTM Int'l (all rights reserved); Sat Jul 13 13:30:02 EDT 2019
9
Downloaded/printed by
(UFSC) Universidade Federal de Santa Catarina ((UFSC) Universidade Federal de Santa Catarina) pursuant to License Agreement. No further reproductions authorized.