Designation: E122 − 09s1 An American National Standard

Standard Practice for

Calculating Sample Size to Estimate, With Specified
Precision, the Average for a Characteristic of a Lot or
Process1
This standard is issued under the fixed designation E122; 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 (s) 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.

ε1 NOTE—Editorial corrections were made to 8.4.1.2 in November 2011.

1. Scope
e = E/µ, maximum acceptable difference expressed as a
1.1 This practice covers simple methods for calculating fraction of µ.
how many units to include in a random sample in order to f = degrees of freedom for a standard deviation estimate
estimate with a specified precision, a measure of quality for all (7.5).
the units of a lot of material, or produced by a process. This k = the total number of samples available from the same or
practice will clearly indicate the sample size required to similar lots.
estimate the average value of some property or the fraction of µ = lot or process mean or expected value of X, the result
nonconform- ing items produced by a production process of measuring all the units in the lot or process.
during the time interval covered by the random sample. If the µ0 = an advance estimate of µ.
process is not in a state of statistical control, the result will not N = size of the lot.
have predictive value for immediate (future) production. The n = size of the sample taken from a lot or process.
practice treats the common situation where the sampling units nj = size of sample j.
can be considered to exhibit a single (overall) source of nL = size of the sample from a finite lot (7.4).
variability; it does not treat multi-level sources of variability. p' = fraction of a lot or process whose units have the
nonconforming characteristic under investigation.
2. Referenced Documents p0 = an advance estimate of p'.
p = fraction nonconforming in the sample.
2.1 ASTM Standards:2 R = range of a set of sampling values. The largest minus the
E456 Terminology Relating to Quality and Statistics smallest observation.
Rj = range of sample j.
3. Terminology R¯ = k R /k , average of the range of k samples, all of the
3.1 Definitions—Unless otherwise noted, all statistical
(
j51
j

terms are defined in Terminology E456. σ same size (8.2.2).

3.2 Symbols—Symbols used in all equations are defined as = lot or process standard deviation of X, the result of
measuring all of the units of a finite lot or process.
follows: σ0 = an advance estimate of σ.
E = the maximum acceptable difference between the true
s =
F(n i~ X 2 HX ! 2/ ~ n21 ! 1⁄2 , an estimate of the
i51

standard
average and the sample average. deviation σ from n observation, Xi, i = 1 to n.
k
s¯ Sj/k , average s from k samples all of the same size
(
j51
1
This practice is under the jurisdiction of ASTM Committee E11 on Quality
= (8.2.1).
and
Statistics and is the direct responsibility of Subcommittee E11.10 on Sampling / sp = pooled (weighted average) s from k samples, not all of
Statistics.
the same size (8.2).
Current edition approved Aug. 1, 2009. Published September 2009. Originally
approved in 1958. Last previous edition approved in 2007 as E122 – 07. DOI: sj = standard deviation of sample j.
10.1520/E0122-09E01. Vo = an advance estimate of V, equal to δo /µo.
2
For referenced ASTM standards, visit the ASTM website, www.astm.org, or v = s/X¯ , the coefficient of variation estimated from
contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM the sample.
Standards volume information, refer to the standard’s Document Summary page on
the ASTM website. vp = pooled (weighted average) of v from k samples (8.3).

1
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States

2
 E122 − 09s1
vj = coefficient of variation from sample j. 6. Precision Desired
X = numerical value of the characteristic of an individual 6.1 The approximate precision desired for the estimate must
X unit being measured.
n
¯ must be decided what
( X i /n i average of n
observations, Xi, i
= 1 to n.
i51
4. Significance and Use
4.1 This practice is
intended for use in
determining the sample size
required to estimate, with
specified precision, a
measure of quality of a lot or
process. The practice applies
when quality is expressed as
either the lot average for a
given property, or as the lot
fraction not conforming to
prescribed standards. The
level of a characteristic may
often be taken as an
indication of the quality of a
material. If so, an estimate of
the average value of that
characteristic or of the
fraction of the observed
values that do not conform to
a specification for that
characteristic becomes a
measure of quality with
respect to that characteristic.
This practice is intended for
use in determining the
sample size required to
estimate, with specified
precision, such a measure of
the quality of a lot or process
either as an average value or
as a fraction not conforming
to a specified value.
5. Empirical Knowledge
Needed
5.1 Some empirical
knowledge of the problem is
desirable in advance.
5.1.1 We may have some
idea about the standard
deviation of the
characteristic.
5.1.2 If we have not had
enough experience to give a
precise estimate for the
3
be prescribed. That is, it
maximum deviation, E,
can be tolerated between
the estimate to be made
from the sample and the
result that would be
obtained by
standard deviation, we may be
able to state our belief about
the range or spread of the
characteristic from its lowest
to its highest value and
possibly about the shape of
the distribution of the
characteristic; for instance,
we might be able to say
whether most of the values
lie at one end of the range, or
are mostly in the middle, or
run rather uniformly from one
end to the other (Section 9).
5.2 If the aim is to estimate
the fraction nonconforming,
then each unit can be assigned
a value of 0 or 1 (conforming
or nonconforming), and the
standard deviation as well as
the shape of the distribution
depends only on p', the
fraction nonconforming in the
lot or process. Some rough
idea con- cerning the size of
p' is therefore needed, which
may be derived from
preliminary sampling or from
previous experience.
5.3 More knowledge
permits a smaller sample size.
Seldom will there be
difficulty in acquiring enough
information to compute the
required size of sample. A
sample that is larger than the
equations indicate is used in
actual practice when the
empirical knowledge is
sketchy to start with and when
the desired precision is
critical.
5.4 The precision of the
estimate made from a random
sample may itself be
estimated from the sample.

This estima- tion of the
precision from one sample
makes it possible to fix
more economically the
sample size for the next
sample of a similar material.
In other words, information
concerning the process, and
the material produced
thereby, accumulates and
should be used.
E122 − 09s1
measuring every unit in the 1.96 0.050 or 50 in abnormal material and
1000 (1 in 20)
lot or process. 1.64 0.100 or 100 segregate it for separate
6.2 In some cases, the
in 1000 (1 in 10) treatment.
maximum allowable 7.1.1 If a lot of material 7.2 There are some
sampling error is expressed has a highly asymmetric materials for which σ varies
as a proportion, e, or a distribu- tion in the approxi- mately with µ, in
percentage, 100 e. For characteristic measured, the which case V(= σ ⁄µ) remains
example, one may wish to sample size as calcu- lated approximately constant from
make an estimate of the in Eq 1 may not be large to small values of µ.
sulfur content of coal adequate. There are two 7.2.1 For the situation of
within 1 %, or e = 0.01. things to do when 7.2, the equation for the
asymmetry is suspected. sample size, n, is as
7. Equations for 7.1.1.1 Probe the follows:
Calculating Sample Size material with a view to
n 5 ~3
discovering, for example,
7.1 Based on a normal extra-high values, or Vo/e !
2
distribution for the possibly spotty runs of
characteristic, the equation abnor- mal character, in If the relative error, e, is
for the size, n, of the order to approximate to be the same for all values
sample is as follows: roughly the amount of the of µ, then everything on the
n 5 asymmetry for use with right-hand side of Eq 2 is a
~ 3σo statistical theory and constant; hence n is also a
/E ! adjustment of the sample constant, which means that
2
size if necessary. the same sample size n
The multiplier 3 is a 7.1.1.2 Search the lot for would be required for all
factor corresponding to a values of µ.
low probabil- ity that the 7.3 If the problem is to
TABLE 1 Values of the
difference between the estimate the lot fraction Correctio
sample estimate and the nonconforming, then σ 2 is C4 n Factor
Sample Size3, (nj) d2
result of measuring (by the replaced by2p ( 1 − p ) so that.798 1.13 C4 and d2
same methods) all the units Eq 4 .921 2.06 for
in the lot or process is 5 .940 2.33 Selected
8 .965 2.85 Sample
greater than E. The value 3 10 .973 3.08 Sizes n A
o o o
is recommended for general j
1 becomes:
use. With the multiplier 3,
and with a lot or process n 5
standard deviation equal to ~ 3/E ! 2
the advance estimate, it is po ~ 1 2
practi- cally certain that the po !
A
As nj becomes large, C4 approaches
sampling error will not 7.4 When the average for 1.000.
exceed E. Where a lesser the production process is not
degree of certainty is needed, but rather the
desired a smaller multiplier average of a particular lot is
may be used (Note 1). needed, then the required
sample size is less than Eq 1,
NOTE 1—For example,
multiplying by 2 in place of 3 Eq 2, and Eq 3 indicate. The
gives a probability of about 45 sample size for estimating
parts in 1000 that the sampling the average of the finite lot
error will exceed will be:
E. Although distributions met in
practice may not be normal, the nL 5
following text table (based on
8.2.2 An even simpler,
n/ @ 11 ~
the normal distribution) and slightly less efficient
n/N ! # estimate for σo may be
indicates approximate
probabilities: computed by using the
where n is the value average range (R¯ ) taken
Factor Approximate
computed from Eq 1, Eq 2,
Probability of Exceeding E
or Eq 3. This reduction in from the several previous
3 0.003 or 3 in data sets that have the same
1000 (practical certainty) sample size is usually of
2.56 0.010 or 10 in little importance unless n is group size.
1000 10 % or more of N.
2 0.045 or 45 in
1000 7.5 When the i R¯
4 n
 E122 − 09s1
j
σo 5 2
p
j51
j
d j j
5
Eq 3 may be appropriate. The factor, d2, from
1

When the advance estimate results from use 8.3.1 Example 2—Use
Table 1 is needed to
σ0 is based on f degrees of of Eq 6. of V, the estimated
freedom, the sample size in
Eq 1 may be replaced(by:
F convert the average range
into an unbiased estimate k
~n
( coefficient of variation:
7 8.3.1.1 Problem—To
of(σo.
1
2 / )
k 2 compute the

~ 3σ0/E ! 2 ~ 11 =2/f !
n5 8.2.3 Example 1—Use of
s¯.
s
5
2
1! G sample size needed
to
8.2.3.1 Problem—To 1!
NOTE 2—The standard error s
of a sample variance with f compute the sample size 2/
degrees of freedom, based on the needed to estimate the p j
estimate the
normal distribution, is =2σ 4 /f. average transverse strength
The factor
j j
average abrasion
of a lot of bricks when the j5 j

~ 11 =2/f ! has the effect of value of E is 50 psi, and 0

5
1 resistance (that is,
increasing the preliminary average
practical certainty is
estimate σ 2 by desired. 8.2.1 If each of the number of cycles) of a
one times its standard error.
8.2.3.2 Solution—From previous data sets contains material when the value of e
the data of three previous the same number of is 0.10 or 10 %, and
8. Reduction of measurements, nj, then a practical certainty is
lots, the values of the
Empirical simpler, but slightly less desired.
estimated standard
Knowledge to a efficient estimate for σo may 8.3.1.2 Solution—There
deviation were found to be
Numerical Value of be made by using an average are no data from previous
215, 192, and 202 psi based
σo (Data for (s¯) of the s values obtained samples of this same
on samples of 100 bricks.
Previous Samples from the several previous material, but data for six
The average of these three
Available) samples. The calculated s¯ samples of similar materials
standard deviations is 203
8.1 This section value will in general be a show a wide range of
psi. The c4 value is
illustrates the use of the slightly biased estimate of σo. resistance. However, the
essentially unity when Eq 1
equations in Section 7 An unbiased estimate of σo values of estimated standard
gives the following
when there are data for is computed as follows: deviation are approximately
equation for the required
previous samples. s¯ proportional to the observed
size of sample to give a
8.2 For Eq 1—An maximum sampling error averages, as shown in the
estimate of σo can be of 50 psi: following text table:
σo 5 (8) Coef
obtained from previous sets c n 5 @ ~ 3 3 203! /50# Lot No. 4
Size Cycles Sample Avg
Deviation
of Varia-
S
ficie
of data. The standard 2 t
5 12.2 2 5 149 tion, % nt
a
deviation, s, from any giveni bricks 1 10 90 13 14
n
2 10 190 32 17
sample is computed as: 8.3 For Eq 2—If σ 3 10 350 45
d
13
a
varies approximately 4 10 450 71 16
r
proportionately with µ for 5 10 1000 120 12
d
6 10 3550 680 19
the characteristic of the w Pooled 15.4
material to be measured, h
compute the average, X¯ , e
the standard deviation, s, r
and the e
F~
n 1
/
( coefficient of variation v for
X 2 X¯ (6)
2
s5 i
5
each sample. The pooled V t
1
! 2/ ~ n 2 value h
vp for k samples, not e
1! G necessarily of the same size,
is obtained v
The s value is a sample by a weighted average of a
estimate of σo. A better, the k results. Then use Eq 2.
more stable l
value for σo may be v 5 F k
~2/
u
e
(11) !
computed by pooling the
1/2 n 2 1 v
( (
s values obtained
a ing of the k o
from several samples
v f
from similar lots. The e
pooled s value sp r
for k samples is a t
obtained by a weighted g h

5
 E122 − 09s1
e u
a
c l
o
r d
r a
e t
c a
t
i s
o e
n t
s
f
a (
c n
t j

o )
r
, (
T
c a
4
b
, l
e
d
e 1
3
p
e )
n .

d
s 3

o A
S
n T
M
t
h M
e a
n
u
s a
i l
z
e o
n

o P
f r
e
s
t e
h n
e t
a
t
i i
n o
d n
i
o
v f
i
d D

6
 E122 − 09s1
a the estimated fraction
t
nonconforming, p, for each some other source. Try to
a picture how the other
sample. Then for the
weighted average use the observed values may be
a
n following equation: distributed. A few simple
d observations and questions
total number
nonconforming concerning the past behavior
C in all samples of the process, the usual
o p5
total number procedure of blending,
n
t of units in all mixing, stacking, storing,
r samples etc., and knowledge
o concerning the aging of
l 8.4.1 Example 3—Use of material and the usual
p: practice of withdrawing the
C
h
8.4.1.1 Problem—To material (last in, first out; or
a compute the size of sample last in, last out) will usually
r needed to estimate the elicit sufficient information
t fraction nonconforming in a to distinguish between one
lot of alloy steel track bolts form of distribution and
A
n
and nuts when the value of E another (Fig. 1). In case of
a is 0.04, and practical doubt, or in case the
l certainty is desired. specified precision E is a
y
8.4.1.2 Solution—The critical matter, the
s
i data in the following table rectangular distribution may
s from four previous lots were be used. The price of the
, used for an advance extra protection afforded by
estimate of p: the rectangular distribution
A
S is a larger sample, owing to
T the larger standard
M deviation thereof.
9.2.1 The standard
M deviation estimated from
N
L one of the formulas of Fig. 1
as based on the largest and
7 smallest values, may be
A used as an advance estimate
,
of σo in Eq 1. This method
of advance estimation is
2
0 acceptable and is often
0 preferable to doubtful
2 observed values of s, s¯, or
,
r¯.
9.2.2 Example 4—Use of
P
a σo from Fig. 1.
r 9.2.2.1 Problem (same as
t Example 1)—To compute
the sample size needed to
3 estimate the average
.
transverse strength of a lot
The use of the pooled 8.3.1.3 If a maximum of bricks when the value of
coefficient of variation for allowable error of 5 % were E is 50 psi.
Vo in Eq 2 gives the needed, the required sample 9.2.2.2 Solution—From
following for the required size would be 86 specimens. past experience the spread
size of sample to give a The data supplied by the of values of transverse
maximum sampling error prescribed sample will be strength for a lot of bricks
not more than 10 % of the useful for the study in hand has been about 1200
expected value: and also for the next Numb F
ming

n 5 @ ~ 3 3 15.4! /10# investigation of similar Lot No. er ract

1 75 3 0.040 Size9.2.2.3 TheNonconfor
isosceles
triangle
ion distribution in Fig. 1 appears
2
2
5 21.3→22
100
test material.
10 0.100 ming Non
3 specimens 90 4 0.044 to be most appropriate,conthe advance estimate σo is 1200/
4 125
8.4
4
For Eq 3—Compute
0.032 4.9 = 245 psi. Then for
Total 390 21
7
 E122 − 09s1
psi. we ap in the middle of this band, NOTE 1—What is shown here for the normal distribution is somewhat
The re ed but not necessarily arbitrary, because the normal distribution has no finite endpoints.
F
values he up normally distributed. I
G
.

1
n 5 @ ~ 3 3 245! /50# 2
p = 21/390 = 0.054 5 14.7 2 5 216.1 5
n = (3/0.04)2 (0.054) S
217 bricks (14)
o
(0.946)
= [(9 × 0.0511) ⁄ 0.0016] 9.2.2.4 The difference in m
sample size between 217 e
= 287.4 = 288
and 149 bricks (found in
If the value of E were Example 1) is the cost of
T
0.01 the required sample y
sketchy knowledge. p
size would be 4600. With a e
lot size of 2000, Eq 4 gives 9.3 For Eq 2—In s
nL = 1394 items. Although general, the knowledge that
this value of nL represents the use of Vo instead of σo is o
about 70 % of the lot, the preferable would be f
example illustrates the obtained from the analysis
sample size required to of actual data in which D
achieve the value of E with case the methods of i
practical certainty. Section 8 apply. s
t
9.4 For Eq 3—From r
9. Reduction of past experience, estimate i
Empirical approxi- mately the band b
o
Knowledge to a within which the fraction
u
t
Numerical Value for nonconforming is likely to i
σo (No Data from lie. Turn to Fig. 2 and o
o
Previous Samples of read off the value of σ 2 = n
the Same or Like s
p' (1 − p') for the middle of
Material Available) the possible range of p' and a
9.1 This section use it in Eq 8. In case the n
illustrates the use of the desired precision is a d
equations in Section 7 when critical matter, use the
there are no actual observed largest value of σ 2 within T
values for the computation the possible range of p'. h
e
of σo. i
9.2 For Eq 1—From past 10. Consideration of Cost r
experience, try to discover for Sampling and
what the smallest (a) and Testing S
t
largest (b) values of the 10.1 After the required a
characteristic are likely to size of sample to meet a n
be. If this is not known, prescribed precision is d
a
obtain this information from computed from Eq 1, Eq 2,
r
or Eq 3, the next step is to d
compute the cost of testing
this size of sample. If the D
cost is too great, it may be e
possible to relax the v
i
required precision a
t
i
o
n
s

8

FIG. 2 Values of σ, or (σ)2,
Corresponding to Values
of ρ'
(or the equivalent, which is
to accept an increase in the
probability (Section 7) that
the sampling error may
exceed the maximum error
E ) and to reduce the size
of the sample.
10.2 Eq 1 gives n in
terms of a prescribed
precision, but we may solve
it for E in terms of a given n
and thus discover the
precision possible for a
given cost that is,
E53σ o / = n . The same may be
done for Eq 2 and Eq 3.
10.3 It is necessary to
specify either E or the
allowable cost; otherwise
there is no proper size of
sample.
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
E122 − 09s1
determination of the validity of any such
11. Selection of the Sample patent rights, and the risk of infringement of
such rights, are entirely their own
11.1 In order to make responsibility.
any estimate for a lot or for
This standard is subject to revision at any
a process, on the basis of a time by the responsible technical committee
sample, it is necessary to and must be reviewed every five years and
select the units in the if not revised, either reapproved or
withdrawn. Your comments are invited either
sample at random. An for revision of this standard or for additional
acceptable procedure to standards and should be addressed to
ensure a random selection is ASTM International Headquarters. Your
comments will receive careful consideration
the use random numbers. at a meeting of the responsible technical
Lack of predictability, such committee, which you may attend. If you
as a mechanical arm feel that your comments have not received a
fair hearing you should make your views
sweeping over a conveyor known to the ASTM Committee on
belt, does not yield a Standards, at the address shown below.
random sample.
This standard is copyrighted by ASTM
11.2 In the use of International, 100 Barr Harbor Drive, PO Box
C700, West Conshohocken, PA 19428-2959,
random numbers, the United States. Individual reprints (single or
material must first be multiple copies) of this standard may be
broken up in some manner obtained by contacting ASTM at the above
address or at 610-832-9585 (phone), 610-
into sampling units. 832-9555 (fax), or service@astm.org (e-
Moreover, each sampling mail); or through the ASTM website
unit must be identifiable by (www.astm.org). Permission rights to
photocopy the standard may also be
a serial number, actual, or secured from the Copyright Clearance
by some rule. For packaged Center, 222 Rosewood Drive, Danvers, MA
01923, Tel: (978) 646-2600;
articles, a rule is easy; the
http://www.copyright.com/
package contains a certain
number of articles in
definite layers, arranged in
a particular way, and it is
easy to devise some system
for numbering the articles.
In the case of bulk material
like ore, or coal, or a barrel
of bolts or nuts, the
problem of defining usable
sampling units must take
place at an earlier stage of
manufacture or in the
process of moving the
materials.
11.3 It is not the purpose
of this practice cover the
handling of materials, nor
to find ways by which one
can with surety discover
the way to a satisfactory
type of sampling unit.
Instead, it is assumed that a
suitable sampling unit has
been defined and then the
aim is to answer the
question of how many to
draw.