Scribd
Upload
18 views

Recommended Criteria For Single Samples: Table 1 Table 1

Formula for the outliers.

Uploaded by

Elmo Charles
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
18 views

Recommended Criteria For Single Samples: Table 1 Table 1

Formula for the outliers.

Uploaded by

Elmo Charles
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

E178 − 08

4.3.1.1 Reject observation(s). criteria presented may also be used to test the hypothesis of
4.3.1.2 Correct observation(s) on physical grounds. normality or that the random sample taken did come from a
4.3.1.3 Reject it (them) and possibly take additional obser- normal or Gaussian population. The end result is for all
vation(s). practical purposes the same, that is, we really wish to know
4.3.2 Physical Reason Unknown—Use Statistical Test: whether we ought to proceed as if we have in hand a sample of
4.3.2.1 Reject observation(s). homogeneous normal observations.
4.3.2.2 Correct observation(s) statistically.
4.3.2.3 Reject it (them) and possibly take additional obser- 6. Recommended Criteria for Single Samples
vation(s). 6.1 Let the sample of n observations be denoted in order of
4.3.2.4 Employ truncated-sample theory for censored obser- increasing magnitude by x1 ≤ x2 ≤ x3 ≤ ... ≤ x n. Let xn be the
vations. doubtful value, that is the largest value. The test criterion, Tn,
4.4 The statistical test may always be used to support a recommended here for a single outlier is as follows:
judgment that a physical reason does actually exist for an T n 5 ~ x n 2 x̄ ! /s (1)
outlier, or the statistical criterion may be used routinely as a
basis to initiate action to find a physical cause. where:
x̄ = arithmetic average of all n values, and
5. Basis of Statistical Criteria for Outliers s = estimate of the population standard deviation based on
5.1 There are a number of criteria for testing outliers. In all the sample data, calculated as follows:
of these, the doubtful observation is included in the calculation
of the numerical value of a sample criterion (or statistic), which s = n n

! ( ~ x i 2x̄ ! 2
! (x 2 2
i 2n·x̄
is then compared with a critical value based on the theory of i51
5
i51

random sampling to determine whether the doubtful observa- n21 n21


tion is to be retained or rejected. The critical value is that value n

S( D
n 2

of the sample criterion which would be exceeded by chance


with some specified (small) probability on the assumption that
5!( i51
x i 22
n21
i51
xi /n

all the observations did indeed constitute a random sample


from a common system of causes, a single parent population, If x1 rather than xn is the doubtful value, the criterion is as
distribution or universe. The specified small probability is follows:
called the “significance level” or “percentage point” and can be T 1 5 ~ x̄ 2 x 1 ! /s (2)
thought of as the risk of erroneously rejecting a good obser-
vation. It becomes clear, therefore, that if there exists a real The critical values for either case, for the 1 and 5 % levels of
shift or change in the value of an observation that arises from significance, are given in Table 1. Table 1 and the following
nonrandom causes (human error, loss of calibration of tables give the “one-sided” significance levels. In the previous
instrument, change of measuring instrument, or even change of tentative recommended practice (1961), the tables listed values
time of measurements, etc.), then the observed value of the of significance levels double those in the present practice, since
sample criterion used would exceed the “critical value” based it was considered that the experimenter would test either the
on random-sampling theory. Tables of critical values are lowest or the highest observation (or both) for statistical
usually given for several different significance levels, for significance. However, to be consistent with actual practice and
example, 5 %, 1 %. For statistical tests of outlying in an attempt to avoid further misunderstanding, single-sided
observations, it is generally recommended that a low signifi- significance levels are tabulated here so that both viewpoints
cance level, such as 1 %, be used and that significance levels can be represented.
greater than 5 % should not be common practice. 6.2 The hypothesis that we are testing in every case is that
NOTE 1—In this practice, we will usually illustrate the use of the 5 % all observations in the sample come from the same normal
significance level. Proper choice of level in probability depends on the population. Let us adopt, for example, a significance level of
particular problem and just what may be involved, along with the risk that 0.05. If we are interested only in outliers that occur on the high
one is willing to take in rejecting a good observation, that is, if the side, we should always use the statistic Tn = (xn − x̄)/s and take
null-hypothesis stating “all observations in the sample come from the
same normal population” may be assumed correct.
as critical value the 0.05 point of Table 1. On the other hand,
if we are interested only in outliers occurring on the low side,
5.2 It should be pointed out that almost all criteria for we would always use the statistic T1 = (x̄ − x1)/s and again take
outliers are based on an assumed underlying normal (Gaussian) as a critical value the 0.05 point of Table 1. Suppose, however,
population or distribution. When the data are not normally or that we are interested in outliers occurring on either side, but
approximately normally distributed, the probabilities associ- do not believe that outliers can occur on both sides simultane-
ated with these tests will be different. Until such time as criteria ously. We might, for example, believe that at some time during
not sensitive to the normality assumption are developed, the the experiment something possibly happened to cause an
experimenter is cautioned against interpreting the probabilities extraneous variation on the high side or on the low side, but
too literally. that it was very unlikely that two or more such events could
5.3 Although our primary interest here is that of detecting have occurred, one being an extraneous variation on the high
outlying observations, we remark that some of the statistical side and the other an extraneous variation on the low side. With

Copyright by ASTM Int'l (all rights reserved); Wed Jul 22 11:50:37 EDT 2015
2
Downloaded/printed by
Universidad Nacional De Colombia (Universidad Nacional De Colombia) pursuant to License Agreement. No further reproductions authorized.

You might also like