United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
P.O. Box 93478
Las Vegas NV 89193-3478
EPA/600/4-90/013
May 1990
Research and Development
A Rationale for the
Assessment of Errors
in the Sampling of
Soils
-------
A RATIONALE FOR THE ASSESSMENT OF ERRORS IN THE SAMPLING OF SOILS
by
J. Jeffrey van Ee and Louis J. Blume
Exposure Assessment Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89193
and
Thomas H. Starks
Environmental Research Center
University of Nevada - Las Vegas
Las Vegas, Nevada 89154
U.S. Environment! Portion Ap°nn;
Region 5, Librr--" """' ~
77 West JacH •-'. .,
Chicago, it COL- ' ' '--
Environmental Monitoring Systems Laboratory
Office of Research and Development
Las Vegas, Nevada 89193-3478
-------
NOTICE
The information in this document has been funded wholly or in
part by the United States Environmental Protection Agency under
Cooperative Agreement CR814701-01 . It has been subject to the
Agency's peer and administrative review, and it has been approved
for publication as an EPA document.
-------
TABLE OF CONTENTS
Page
LIST OF FIGURES iv
LIST OF TABLES V
ACKNOWLEDGMENT vi
INTRODUCTION 1
Purpose l
BACKGROUND 3
Sources of Error 3
Representative Sampling 5
Data Quality Levels for Analytical Measurements in
Superfund 7
NUMBER OF QUALITY ASSESSMENT SAMPLES 9
Background 9
Confidence Levels for the Assessment of Measurement
Variability 10
Quantitative Assessments of Bias 11
Recommendations for the Assessment and Control of Bias
and Measurement Variability 12
DEFINITIONS AND TERMS 15
Quality Assessment Samples 15
Split Samples 19
Spiked Samples 19
Blanks 19
Batch 19
AVAILABILITY OF QUALITY ASSESSMENT SAMPLES 21
A RATIONALE FOR ASSESSING ERRORS 22
Quality Assessment Sample Design 22
Field Duplicates 25
Preparation Splits 26
Field Evaluation Samples 27
External Laboratory Evaluation Samples 28
Preparation Rinsate Blanks 29
Field Rinsate Blanks 30
Equations for Estimating Bias and Precision 31
AN ALTERNATIVE QA DESIGN THAT DOES NOT EMPLOY FES AND ELES 34
EXAMPLE OF QUALITY ASSESSMENT 38
QA/QC from a Pilot Study 38
Post Pilot Study Data Analysis 39
Development of the QA Plan for the Primary Study 44
ii
-------
RECOMMENDATIONS FOR FUTURE WORK 45
REFERENCES 46
APPENDIX A 48
KEY COMPONENTS 48
Training 48
Pilot Study 48
Audits 49
Documentation 50
APPENDIX B 51
DEFINITIONS 51
Quality Assurance 51
Quality Control 51
Quality Assessment 51
Soil 51
Standard Additions 52
APPENDIX C 53
Soil Sampling Methods Table 53
APPENDIX D 54
The Lognormal Distribution and Logarithmic
Transformations 54
APPENDIX E 55
Non-blind Quality Assessment Samples in the Contract
Laboratory Program 55
APPENDIX F
Upper Confidence Limits for the Variance, a2, as a
Function of the Number of Degrees of Freedom for
the Variance Estimate ,s3. 57
iii
-------
LIST OF FIGURES
Figure Page
1. Quality Assessment Samples 24
2. Alternative Quality Assessment Sample Design 37
3. Scatter Plot of QA Data from ASSESS 43
4. Error Plot from ASSESS 44
iv
-------
LIST OF TABLES
Table Page
1. Sources of Bias in Soil Sampling Studies 4
2. Sources of Variability in Soil Sampling Studies 5
3. Some 95 Percent Confidence Intervals for Variance 10
4. Types of Quality Assessment Samples or Procedures 17-18
5. Equations for Determining Precision and Bias 32-33
6. Equations for Determining Precision without FES and ELES 35
7. Quality Assessment Data 41
8. Transformed Quality Assessment Data 42
-------
ACKNOWLEDGMENT
During the initial development of this document, Kenneth W. Brown
and Daniel T. Heggem of the Exposure Assessment Division
contributed their knowledge, words, and constructive suggestions
to the development of the first drafts. Later, Ann M. Pitchford
of the Exposure Assessment Division made further constructive
changes to the developing document. The document was then
reviewed by a variety of people with a variety of backgrounds
since the guidance provided in this document is meant to apply to
an entire soils study which involves a diverse group of people.
The resulting document is meant to serve as an initial "roadmap"
for people to use in assessing the errors and uncertainties in a
soils study. With time the importance of the guidance provided
by this document is expected to become greater as more experience
is gained in applying the rationale to greater numbers of
hazardous waste site investigations. Further revisions are
anticipated, but those revisions should not detract from the
significant contributions that were made in the development of
this document to the assessment of total quality of data from a
soils investigation.
VI
-------
INTRODUCTION
Considerable guidance has been provided on the importance of
quality assurance (QA), quality control (QC), and quality
assessment procedures for assessing and minimizing errors in
environmental studies. QA/QC terms, such as quality assurance
project plans and program plans, are becoming a part of the
vocabulary for remedial project managers (RPM's). Establishment
of data quality objectives (DQOs) early in the process of a site
investigation has been stressed in EPA QA/QC guidance documents.
Quality assessment practices, such as the use of duplicate,
split, spiked, and reference samples, are becoming widely
accepted as important means for assessing errors in measurement
processes. Despite the existence of numerous and various forms
of guidance for hazardous waste site investigations, there have
been no clear, concise, well-defined strategies for precisely how
recommended QA/QC practices can be utilized.
Purpose
The purpose of this document is to provide a foundation for
answering two basic questions:
How many, and what type, of samples are required to
assess the quality of data in a field sampling effort?
(quality assessment samples)
How can the information from the quality assessment
samples be used to identify and control sources of
error and uncertainties in the measurement process?
This document expands upon the guidance for quality control
samples for field sampling as contained in Appendix C of EPA's
Data Quality Objectives for Remedial Response Activities -
Development Process (9). This report outlines, in greater
detail, strategies for how errors may be assessed and minimized
in the sampling of soils with emphasis on inorganic contaminants.
Basic guidance for soil sampling, which includes a
discussion of basic principles, may be found in EPA's Revised
Soil Sampling Quality Assurance Users Guide (15). The Users
Guide is intended to be revised on a periodic basis. It is
anticipated that some of the guidance provided in this document
will eventually be incorporated into the Users Guide.
The primary audience for this document is assumed to be
RPM's who have a concern about the quality of the data being
collected at Superfund sites but have little time to understand
the complexities of the processes used to assess the quality of
data from the total measurement process. The approach outlined
in this document for assessing errors in the field sampling of
-------
inorganics in soils may be transferrable, with modification, to
other contaminants in other media.
The example offered at the end of this document illustrates
the planning process for determining a reasonable number of
quality assessment samples. The example also demonstrates how
the information from the process may be used to document the
quality of the measurement data, and how this data may be used to
make adjustments to the monitoring program.
-------
BACKGROUND
Superfund and RCRA site investigations are complicated by:
the variety of media being investigated, an assortment of
methods, the diversity of people, the variety of contaminants,
and the numerous risks and effects to human health and the
environment. Many phases exist in Superfund site investigations.
An initial phase, generally described as a "preliminary
investigation," consists of collecting and reviewing existing
data and data from limited measurements using practically any
available method. The next phase, generally described as "site
characterization," uses selected methods and prescribed
procedures to characterize the magnitude and areal extent of the
contamination. Final phases include an examination of remedial
actions, which involve an examination of treatment technologies,
and monitoring to assess the degree of cleanup at a site. A
final phase may require long-term monitoring to substantiate that
no new or additional threats occur to human health and the
environment. Throughout Superfund site investigations QA/QC
procedures change as data quality objectives vary and different
phases occur.
Sources of Error
In many of the phases of Superfund and RCRA site
investigations, errors and uncertainties occur. During the
measurement process, random errors will be induced from:
sampling; handling, transportation and preparation of the samples
for shipment to the laboratory; taking a subsample from the field
sample and preparing the subsample for analysis at the
laboratory, and analysis of the sample at the laboratory
(including data handling errors). The magnitude of these errors
can be expected to vary during the. measurement process and make
it more difficult to determine the natural variability of
contaminants in the environment. Errors introduced in the
interpretation and analysis of data are not considered in this
document.
Typically, errors in the taking of field samples are much
greater than preparation, handling, analytical, and data analysis
errors; yet, most of the resources in sampling studies have been
devoted to assessing and mitigating laboratory errors. It may be
that those errors have traditionally been the easiest to
identify, assess and control. This document adopts the
approaches used in the laboratory, e.g. the use of duplicate,
split, spiked, evaluation and calibration samples, to identify,
assess and control the errors in the sampling of soils.
Systematic errors, termed bias (B), can accumulate during a
measurement process. Bias may result from: faults in sampling
design, sampling procedure, analytical procedure, contamination,
losses, interactions with containers, deteriorations,
-------
displacement of phase or chemical equilibria, and inaccurate
instrument calibrations. (Table 1) Bias causes the mean value of
the sample data to be either consistently higher or consistently
lower than the "true" mean value. Laboratories usually introduce
various quality control samples into their sample load to detect
possible bias. Bias in soil sampling is difficult to detect.
Components of bias can be discovered by the technique described
as standard additions or by using evaluation samples. On the
other hand, it is difficult to demonstrate that bias is not
present because an apparent lack of bias may be the result of an
inability to measure it rather than its actual absence.
Table 1. Sources of Bias in Soil Sampling Studies
B. = Measurement bias introduced in sample collection not caused
by contamination
B,c = Measurement bias introduced in sample collection caused by
contamination
Bh = Measurement bias introduced in handling and preparation not
caused by contamination
Bhe = Measurement bias introduced in handling and preparation
caused by contamination
B.. = Measurement bias introduced in subsampling not caused by
contamination
B.M =Measurement bias introduced in subsampling caused by
contamination
B. = Measurement bias introduced in the laboratory analytical
process not caused by contamination
B.e = Measurement bias introduced in the laboratory analytical
process caused by contamination
B, = Total measurement bias
NOTE: It is necessary to realize that biases, other than
contamination biases in the measurement of a sample, will often
be dependent on the original concentration of the contaminant
being measured and on the sample matrix. Biases caused by
contamination are listed separately because some QA samples, such
as rinsate samples, detect only contamination bias.
Also, variability occurs in the measurement process from the
heterogeneity of the soil and random errors throughout the
measurement process. The variability caused by any type of
random error is frequently described quantitatively by the
variance, a3, of the random error, or by the positive square
root, the standard deviation, a, of the random error. Variances
of independent random errors are additive in that the variance of
the sum of errors is the sum of the variances of the individual
errors (Table 2). Other quantifications of variability do not
have this useful, additive property.
-------
Table 2. Sources of Variability in Soil Sampling Studies
a^ = a.'+a,2
where at = total variability
a, = measurement variability
ap = population variability
a.2 = a.2+oh2+a..2+a.2+ab2
where a. = sampling variability (standard deviation)
ah = handling, transportation and preparation
variability
-------
taken in association with the other samples, be
representative of that population.
Deposition of airborne contaminants, especially those
recently deposited, is often evident in the surface layer of the
soil. Contaminants that have been deposited by liquid spills or
by long-term disposition of water soluble materials may be found
at depths up to several meters. Plumes emanating from hazardous
waste dumps or leaking storage tanks may be found at considerable
depths. The methods of sampling each of these may be different;
but all make use of one of the following three basic sampling
tools: (1) scoops, (2) coring, or (3) augering devices.
Two major considerations must be addressed when selecting a
specific sampling tool. These two considerations include soil
conditions and the contaminant(s) that are to be analyzed from
the collected material. Soil condition can be extremely variable
from location to location. For example, soils can be wet or dry,
stony, cohesive (e.g., clay) or cohesionless (e.g., sand).
Similarly, contaminants are extremely diverse, varying between
metals which in most cases are relatively immobile, to highly
mobile water soluble substances, to contaminants that are
volatile.
Improper use and selection of sampling tools may result in
data that are not representative of the soil environment being
sampled (See Appendix C). Measurement errors can result from a
tool being either inappropriate for the particular task, or
improperly used. Results based on previous experience, or from
an equivalency test, may be used to evaluate and select the
proper tool for a specific sampling objective. Operational
measurement errors are identified and assessed by implementing
and utilizing a number of field QA samples. The optimal number
and timing of QA samples depend in part on the proper soil
sampling method being utilized.
A variety of sampling methods may be used to obtain a
measurement of inorganics in soil. EPA's Soil Sampling Quality
Assurance Users Guide notes that the concentrations measured in
an heterogenous medium such as soil are related to the volume of
soil sampled and the orientation of the sample within the volume
of earth that is being studied. (The term "support" is used to
describe this concept.) A RPM not only wants to know the
concentration of contaminants, but their location. Frequently,
an average concentration of contaminants in the soil is sought
and compared against some standard. Depending on the sampling
method used, the location of the samples collected, the number of
samples taken, and the time the samples were collected, the
reported concentrations can vary considerably even when
relatively stable contaminants such as inorganics in soil are
measured.
-------
The processes involved in collecting representative samples
of inorganics in soil can be complicated. (The Soil Sampling
Quality Assurance Users Guide should be consulted for further
information on these processes.) The problem of measuring the
natural variability of contaminants, such as inorganics, in soil
and adequately representing the site to be studied is also a
problem for traditional QA/QC programs which have emphasized the
assessment and minimization of errors and variabilities in the
analytical process.
A major problem in obtaining a "representative" sample is
the spatial scale chosen for the study. Geostatistical
techniques, such as kriging, may also be used to estimate the
natural variability of contaminants in soil. A measure of the
spread of the distribution of contaminant concentrations about
the mean concentration is the population variance, apa.
Data Quality Levels for Analytical Measurements in Superfund
As many as five different levels of quality have been
assigned by EPA in the Superfund program to analytical data.
These levels have been generally associated with different phases
of a site investigation (9); however, it may prove to be
necessary to have all five levels of data quality in any one
phase of a site investigation. Levels III and IV involve off-
site analytical laboratory measurements with Level IV in the
Contract Laboratory Program (CLP) having the most rigorous QA/QC
protocols and documentation. Levels III and IV are assumed in
this document in the development of a strategy for assessing
field sampling errors.
Since the assumption is made in this document that CLP
laboratories are involved in the analysis of samples from a soil
sampling study, errors and biases from those laboratory
measurements are presumed to be small and known. These
assumptions allow greater emphasis to be given to the
identification of errors and biases in the field sampling rather
than in the laboratory analysis.
In a pilot study, within a particular phase of a Superfund
site investigation, it may be necessary to utilize Level III and
IV analytical levels to identify, assess and reduce errors in
field sampling even though these analytical levels might not be
needed for every sample and measurement. For example, a field
portable x-ray fluorescence instrument, which measures inorganics
in soil, is frequently used to identify sampling locations for
samples to be sent to the CLP. The data quality from the
portable x-ray fluorescence instrument is not classified as being
Level III or IV; however, data from the instrument is used to
screen samples for subsequent analysis by Level III and IV
methods. It may be advantagous to compare the performance of the
field-screening, portable x-ray fluorescence instrument against
-------
more rigorous, well characterized laboratory analytical methods
even though the level of data quality desired from each method is
different.
The assessment of errors from non-conventional "sampling"
and analytical methods, such as the portable x-ray fluorescence
instrument, are not specifically addressed in this document.
Two important factors must be considered in the collection
of environmental data. These are the probability that the
collected data will yield the correct assessment or solution for
an environmental problem and the costs. The strategy developed
in this document recognizes that these important factors must be
considered in the implementation of QA/QC measures in the
sampling of soils for inorganics.
8
-------
NUMBER OF QUALITY ASSESSMENT SAMPLES
Background
A key question for a RPM is: how many samples must be
collected to adequately characterize the site? A question for a
QA/QC officer is: how many quality assessment samples must be
taken to adequately characterize the errors and uncertainties in
the measurement process? The timing and type of those quality
assessment samples in the measurement process determines the type
of information that is obtained. The number of quality
assessment samples is determined by the available resources and
the degree to which investigators need to be sure that they have
adequately characterized the measurement process. The simplest
case is when one method, one sampling crew, and one laboratory
are used to analyze the soil samples. A more difficult, and
probably more typical, case is when more than one batch of
samples are either collected or analyzed at various times or by
various laboratories.
The percentage of the-total monitoring effort allocated to
QA/QC activities will depend on many factors including the size
of the project, the available knowledge concerning sampling and
analytical procedures, the relationship of risk to human health
and the environment at various pollutant concentrations, the
nearness of action levels to method detection limits, and the
natural variability and distribution of the contaminants.
Typically the smaller the project, the larger will be the
proportion of cost allocated to QA/QC. New, untried procedures
will typically require pilot-study runs and additional training
for personnel. If the action level is near the method detection
limit, there will be little room for error in the measurements,
and the QA/QC effort may have to be large to assure that
measurement errors are kept small. If the natural variability of
the contaminants is relatively large, it may be necessary to
collect more samples rather than collect more quality assessment
samples. One should not specify a certain percentage of a
project's costs be allocated to QA/QC without considering the
above factors.
Previous EPA guidance for the number of quality assessment
samples has been one for every 20 field samples (9). However,
such rules of thumb are oversimplifications and should be treated
with great caution. A better approach is to determine how each
type of QA sample is to be employed and then determine the number
for that type based on the use. For example, field duplicates,
i.e., duplicate samples at the same location, are used to
estimate the combined variance contribution of several sources of
variation. Hence, the number of field duplicates to be obtained
in the study should be dictated by how precise one wants that
estimate of the total measurement variance to be.
-------
Confidence Levels for the Assessment of Measurement Variability
The precision of an estimate, s2, of the "true" variance,
a1, depends on the number of degrees of freedom for the estimate
which is directly related to the number of quality assessment
samples. Table 3 gives the 95% confidence intervals for various
numbers of degrees of freedom, based on an assumption that the
data are, or have been transformed to, normally distributed data.
Methods for obtaining such confidence intervals for any number of
degrees of freedom are given in most statistics texts.
Table 3. Some 95 Percent Confidence Intervals for Variance
Decrrees of Freedom
Confidence Interval
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
30
40
50
100
0.27s2 < a3 <
0.32s2 < a3 <
0.36s2 < a3 <
0.39s2 < 02 <
0.42s2 < a2 <
0.44S2 < O2 <
0.46s2 < a2 <
0.47s2 < a2 <
0.49s2 < a3 <
0.50s2 < a2 <
0.52s2 < a2 <
0.53s2 < <72 <
0.54s2 < a2 <
0.54s2 < a2 <
0.56s2 < a1 <
0.56s2 < a2 <
0.57s2 < a2 <
0.58s2 < a2 <
0.58s2 < a2 <
0.59s2 < a2 <
0.60s2 < a2 <
0.60s2 < a2 <
0.61s2 < a2 <
0.62s2 < a2 <
0.64s2 < a2 <
0.67s2 < a2 <
0.70s2 < a2 <
0.77s2 < a2 <
39.21s2
13.89s2
8.26S2
6.02s2
4.84s2
4.14s2
3.67s2
3.33S2
3.08s2
2.88s2
2.73S2
2.59S2
2.49S2
2.40S2
2.32S2
2.25s2
2.19S2
2.13s2
2.08S2
2.04S2
2.00S2
1.97S2
1.94S2
1.91S2
1.78S2
1.64S2
1.61S2
1.35S2
If it is decided that 20 degrees of freedom gives
satisfactory precision for the estimate of the total measurement
variance, one might equally space 20 field-duplicate samples
among the routinely collected field samples so as to have 20
duplicates by the end of the study. (Note: Each pair, field
duplicate sample and associated routine sample, provides one
10
-------
degree of freedom in the estimation of the between-collocated-
samples variance.) Alternatively, one might take duplicate
samples at a fairly high frequency at the start of the study
until 10 duplicate -airs are obtained and then obtain the
remaining ten dupl^jate pairs at a reduced rate over the
remainder of the study. This second procedure would allow an
early estimate of the variance based on 10 degrees of freedom to
determine whether the QA plan is resulting in error variances in
the range expected, and the remaining ten pairs would allow the
after-study variance estimate to take the entire study into
account.
The confidence intervals in Table 3 are called two-sided
confidence intervals because they put both upper and lower limits
on the value of a2. However, in practice, one is particularly
interested in the upper limit; that is, one is interested in how
much larger than the estimated variance might the true variance
be. If the true measurement error variance is seriously
underestimated in a pilot study, it may cause one to proceed to
an expensive final study with an inadequate protocol. If the
measurement error variance is underestimated in a final study it
may cause the RPM to put more reliance on the study results than
are warranted and may also cause an inadequate protocol to be
copied in a following study. For these reasons, some may be more
interested in one-sided confidence intervals that provide only an
upper limit on the true variance (i.e., since a variance cannot
be negative, a one-sided confidence interval is between zero and
an upper limit). Such upper limits for one-sided confidence
limits are provided in Appendix F for confidence levels of 90, 95
and 99 percent. (Intervals between zero and the upper limits in
Table 3 would be 97.5 percent one-sided confidence intervals;
that is, for 2 degrees of freedom, one would be 97.5 percent
confident that a2 is between zero and 39.21s2.)
Quantitative Assessments of Bias
Quantitative assessments of bias are complicated by the
different types of bias that may be present in a study and the
different times when those biases may occur. There may be
consistent additive-constant biasing error (e.g., the data
handling algorithm might add a constant to all measurements
entered into the database). There may be consistent
multiplicative biasing error (e.g., a recovery error in the
analytical system). There are also random biasing errors of the
additive and multiplicative types (e.g., sample taking, recovery,
contamination, and calibration errors). In sample taking, the
field crew may occasionally have the bottom portion of a soil
core drop out of the sampler prior to bagging the core; if
concentration decreases with depth, this would be a random
biasing error that would increase the expected concentration
above "true" value. The rate of recovery of a chemical from
samples may depend on the individual matrix properties of the
11
-------
samples. The random biasing errors are most difficult to detect
and to quantify. If contamination occurs in one of 20 samples on
average, the use of 20 contamination blanks would have a 36
(=100[1-0.05]2°) percent chance of not encountering a
contamination incident. Random biasing errors contribute to
measurement variance as well as the bias of the measurement
system. The number of samples required to detect random bias
will depend on the distribution of the biasing errors, and this
distribution will generally be unknown. A major problem in data
analysis is the separation of random biasing errors from random
nonbiasing errors. Estimation of the magnitude of bias and its
effect on the estimates and decisions, would require many more
quality assessment samples than are required for the detection of
bias. Bias estimates that are reported in the literature are
often only estimates of the analytical bias obtained either as
the difference in recovery rate from 100% obtained by the method
of standard additions, or the average difference between reported
and reference values of performance evaluation samples.
With these considerations in mind, it would seem that the
best one can do is to include some bias-detecting quality
assessment samples in each batch of routine samples and hope that
they will detect bias if it is present. If bias is detected, an
effort should be made to eliminate the source of bias, rather
than attempt to correct routine-sample measurements for bias.
Non-blind samples, such as the calibration check standards
at the analytical laboratory (Appendix E), are used to assess
bias in the laboratory and provide a quality control function.
That is, if measurements of these check standards differ by too
much from their reference values, the instrument is declared "out
of control" and is re-calibrated. Then, samples between the last
in-control reading and the out-of-control reading may be
reanalyzed. The frequency of use of samples of this quality
control type should be based on the consequences associated with
out-of-control data and the costs of the analyses of these
samples versus the costs of re-analyzing routine field samples in
out-of-control situations. This frequency of use is usually
related to the probability of obtaining an out-of-control
situation in the laboratory with the objective being to minimize
expenditures of both time and money while obtaining data of
quality suitable for the intended end-use of the data.
Recommendations for the Assessment and Control of Bias and
Measurement Variability
A two-phased approach is suggested to answer the questions
posed in the introduction:
How many, and what type, of samples are required to
assess the quality of data in a field sampling effort?
(quality assessment samples)
12
-------
How can the information from the quality assessment
samples be used to identify and control sources of
error and uncertainties in the measurement process?
The first phase involves the acquisition of data to estimate
total measurement variability and bias. The second phase
involves identification of the sources of the bias and
variability.
The required number of samples will vary depending on the
data quality objectives and available resources. The more
quality assessment samples that are used, the better the
assessment of measurement errors. Five field-duplicate samples,
as demonstrated in Table 3, will yield an estimate of the
measurement variability that may be high by a factor of 6 or low
by a factor of 2.5 with a confidence level of 95%. As noted
earlier, accurate assessments of measurement bias are more
complicated. A recommendation of one quality assessment sample
for bias per batch would allow for the plotting of bias on a
control chart to determine if the measured bias is within
acceptable limits. (Bias may be random, constant, or varying.)
In either the measurement of bias, or measurement variability,
the accumulation of historical data is extremely important in
judging the appropriate number of quality assessment samples and
the relative importance of that data to the overall objectives of
the study.
It must be emphasized that estimates of measurement-error
variance components are of little value if they are based on so
few degrees of freedom that they may differ from the true
variances by large factors (Table 3). Hence, even in pilot
studies with few routine samples, it is important to obtain
measurement-error component variance estimates that are based on
a sufficient number of quality assessment samples (i.e., based on
a sufficient number of degrees of freedom) that the user can have
some confidence that the large estimates actually represent large
variances and small estimates represent small variances.
Otherwise, corrective actions taken to improve precision may be a
waste of money, and failure to take corrective action may result
in the failure of the subsequent study.
Experience indicates that an effective quality assurance
program is negated if certain key elements of a sampling effort
are not adequately addressed. Those elements are: training,
pilot studies, audits and documentation. More detailed
discussion of those key components is provided in Appendix A;
however, the importance of pilot studies to the overall
monitoring effort cannot be stressed enough. The experience and
data gained during the pilot study can be extremely important to
the success of the larger monitoring effort.
13
-------
Frequently, time and financial constraints lead to a minimum
number of samples being collected in the initial measurement
phase of a hazardous waste study. If sufficient historical
quality assessment data has not been collected for the selected
sampling crews, sampling methods, and analytical laboratories
involved in the initial measurement phase, an accurate assessment
of performance during the "pilot study" will require a relatively
large number of quality assessment samples. However, even with
an experienced sampling crew, tested sampling methods, and well-
characterized analytical laboratories, unique characteristics of
a particular site may require an increased number of quality
assessment samples to measure performance against stated data
quality objectives.
14
-------
DEFINITIONS AND TERMS
Problems can often occur when important terms in
environmental sampling are not defined, poorly defined, or not
well understood. Commonly accepted terms such as soil can have
many different definitions depending upon the interests and
background of an investigator. Major errors can occur if
everyone involved in an investigation has a different
understanding of a term.
While the definitions contained in this document may not be
universal, it is important in an understanding of the overall
process of assessing errors and identifying their sources that
the terms be defined early. Important steps and samples that are
required for the assessment of variability and bias are defined
in the main document. Basic definitions that are less critical
to an understanding of the quality assurance process are in
Appendix B.
Quality Assessment Samples
Many terms have been used to describe quality assessment
samples. (Quality evaluation is also used.) Quality assessment
samples are defined as those samples that allow statements to be
made concerning the quality of the measurement system. These
samples have two primary reasons for utilization. They allow
assessment of the quality of the data, and most importantly, they
allow for control of data quality to assure that it meets the
original objectives.
The objective of this section is to identify the various
sample types, define what they are used for, and how they have
been previously used in characterizing the measurement process.
Quality assessment samples include samples from three
groups, based upon whether they are double-blind, single-blind,
or non-blind to the analytical laboratory. Double-blind samples
are samples that cannot be distinguished from routine samples by
the analytical laboratory. Single-blind samples are samples that
can be distinguished from routine samples but are of unknown
concentration. Non-blind samples are samples that have a
concentration and origin that are known to the analytical
laboratory. Some references state that quality control samples
are only those that are blind to the analytical laboratory (9).
Other documents refer to quality control samples as non-blind to
the analytical laboratory (3,13). The intent of categorizing
quality assessment samples into these categories, i.e., double-
blind, single-blind, and non-blind, is to avoid confusion due to
terminology. The key point is that all of the samples discussed
here refer to those samples that make some statement about the
quality of the measurement system. This discussion will not
include samples such as background samples or critical samples
15
-------
(9) because they are required for characterization of the
contamination at a site and not for characterization of the
errors in the measurement system.
Table 4 and Appendix E list typical quality assessment
samples and describe how measurements of these samples are used
in the control of the measurement process and in the evaluation
of the quality assurance procedures. To obtain an unbiased
measure of the internal consistency of the samples and their
analyses, the individual quality assessment samples must be
double-blind and should be labeled as routine samples so that the
analyst (and preferably also the laboratory) does not know the
relationship between the samples. This reduces the chances of
conscious or unconscious efforts to improve the apparent
consistency of the analyses.
16
-------
Table 4. Types of Quality Assessment Samples or Procedures
Double-blind Samples
1. Field Evaluation Samples TFES1
These samples are of known concentration, subjected to the same
manipulations as routine samples and introduced in the field at
the earliest stage possible. They can be used to detect
measurement bias and to estimate precision.
2. Low Level Field Evaluation Samples fLLFES1
These samples are essentially the same as field evaluation
samples, but they have very low or non-existent concentrations of
the contaminant. They are used for determination of
contamination in the sample collection, transport, and analysis
processes. They can also be used for determination of the system
detection limit (13).
3. External Laboratory Evaluation Samples fELES)
This sample is similar to the field evaluation sample except it
is sent directly to the analytical laboratory without undergoing
any field manipulations. It can be used to determine laboratory
bias and precision if used in duplicate. We recommend using the
same sample as the FES to allow isolation of the potential
sources of error. Spiked soil samples have been used as external
laboratory evaluation samples in past studies for dioxin,
pesticides, and organics (1,6), and natural evaluation samples
have been used for metals analysis in soil and liquid samples
(3,13).
4. Low Level External Laboratory Evaluation Sample CLLELES1
This sample is similar to the LLFES except it is sent directly to
the analytical laboratory without undergoing any field
manipulations. It is used to determine method detection limit,
and the presence or absence of laboratory contamination. We
recommend using the same sample source as for the LLFES to allow
isolation and identification of the source of contamination.
5. Field Matrix Spike rFMSI
This is a routine sample spiked with the contaminant of interest
in the field. Because of the inherent problems associated with
the spiking procedure and recovery it is not recommended for use
in field studies (9).
6. Field Duplicate (Fm
An additional sample taken near the routine field sample to
determine total within-batch measurement variability. The
17
-------
differences in the measurements of duplicate and associated
samples are in part caused by the short-range spatial variability
(heterogeneity) in the soil and are associated with the
measurement error in the field crew's selection of the soil
volume to be the physical sample (i.e., two crews sent to the
same sampling site, or the same crew sent at different times,
would be unlikely to choose exactly the same spot to sample) .
7. Preparation Split (PS)
After a routine field sample is homogenized, a subsample is taken
for use as the routine laboratory sample. If an additional
subsample is taken from the routine field sample in the same way
as the routine laboratory sample, this additional sample is
called a preparation split. The preparation split allows
estimation of error variability arising from the subsampling
process and from all sources of error following subsampling.
This sample might also be sent to a reference laboratory to check
for laboratory bias or to estimate inter-laboratory variability.
These have also been called replicates (5).
Single-Blind Samples
1. Field Rinsate Blanks
These samples, also called field blanks (9), decontamination
blanks (14,15), equipment blanks (5), and dynamic blanks (5), are
obtained by running distilled, deionized (DDI) water through the
sampling equipment after decontamination to test for any residual
contamination.
2. Preparation Rinsate Blank (PRBt
These samples, also called sample bank blanks (12,14,15), are
obtained by passing DDI water through the sample preparation
apparatus after cleaning in order to check for residual
contamination .
3. Trip Blank
-------
Split Samples
Samples can be split to provide:
samples for both parties in a litigation or potential
litigation situation;
a measure of the within-sample variability;
materials for spiking in order to test recovery; and
a measure of the analytical and extraction errors.
The location of the sample splitting determines the components of
variance that are measured by the split. A split made in the
sample bank (i.e., at the sample preparation facility to which
samples are sent from the field) measures error introduced from
that level onward. A split made in the field includes errors
associated with field handling. A split or series of subsamples
made in the laboratory for extraction purposes measures the
extraction error and subsequent analytical errors.
Spiked Samples
Spiked samples are prepared by adding a known amount of
reference chemical to one of a pair of split samples. Comparing
the results of the analysis of a spiked member to that of the
non-spiked member of the split measures spike recovery and
provides a measure of the analytical bias. Spiked samples are
difficult to prepare with soil material itself. Frequently the
spike solution is added to the extract of the soil sample. This
avoids the problem of mixing, but does not provide a measure of
the interaction of the chemicals in the soil with the spike;
neither does it provide an evaluation of the extraction
efficiency. A predigest spike, as utilized in the CLP (9) would
allow a check of the extraction or digestion efficiency. In
addition, if the laboratory does the spiking, the spiking is non-
blind to the laboratory.
Blanks
Blanks provide a measure of various cross-contamination
sources, background levels in the reagents, decontamination
efficiency, and other potential error that can be introduced from
sources other than the sample. For example, a blank introduced
at the earliest point in the field can measure input from
contaminated dust or air into the sample. A rinsate blank, i.e.,
decontamination sample, measures any chemical that may have been
on the sampling and sample preparation tools after the
decontamination process is completed.
Batch
A batch is defined as a group of samples which are sampled,
shipped and analyzed under similar conditions. The total number
19
-------
of routine and quality assessment samples in a batch is dependent
on the desired frequency of quality assessment sampling that a
budget will allow. The term batch is synonymous with the term
"sample delivery group" as used in the CLP (9).
20
-------
AVAILABILITY OF QUALITY ASSESSMENT SAMPLES
Presently, perforraance-evaluation-materials (PEMs) for
inorganics in soils are not readily available. (PEMs, closely
resemble routine samples, are well characterized, and are
provided as unknown samples to a laboratory to demonstrate that
the laboratory can produce analytical results within specified
limits of performance.) Soil performance-evaluation-materials
are available for routine soil characterization for acid
deposition purposes (13), and performance materials have been
developed for dioxin analysis (1,6). These materials are
available as quarterly blind samples; however, adequate PEMs do
not exist for analysis of inorganics in hazardous waste soils.
To meet the growing needs for PEMs the Environmental
Monitoring Systems Laboratory - Las Vegas (EMSL-LV) in
conjunction with EPA's Office of Emergency and Remedial Response
(OERR) has begun a project 1 to develop, test, and produce "case-
specific" or "site-specific" PEMs. Samples taken at Superfund
sites are organized into groups called "Superfund Cases". Each
-ase of samples is sent to a specific CLP laboratory for
analysis. The objective is to provide a multi-matrix, multi-
analyte, multi-level library or shopping list of PEMs which the
Regional site-managers could order by telephone or from a
catalog. Each PEM would be included as just another sample
within the Case. Ultimately the PEMs would be double-blind to
the laboratories. The PEMs would be tailored-made for each
Superfund Case of analytical samples to enable more reliable,
accurate decision making about Superfund sites.
PEMs are an important component of the rationale that is
used to assess variability and bias throughout the measurement
process; however, variability and biases may also be assessed
without the use of these materials since present availability is
limited. An alternative QA design that does not rely on the use
of FES and ELES is provided after the discussion of the rationale
that is based on the use of PEMs.
Butler, L., 1989. Personal communication.
Environmental Monitoring Systems Laboratory. Las Vegas,
NV.
21
-------
A RATIONALE FOR ASSESSING ERRORS
Quality Assessment Sample Design
An effective quality assurance program should ensure that
the uncertainty associated with the measurement system will be
insignificant when compared to the uncertainty allowed for the
population of interest. As stated by J.K. Taylor (7):
"When the uncertainty of a measured value is one-third
or less of the permissible tolerance for its use, it
can be considered as essentially errorless for that
use."
Therefore it is critical that a quality assurance system provides
for quantification of total measurement error. Measurement error
consists of three major components, i.e. sample collection,
preparation, and analysis. Each of these phases can then be
divided into smaller components depending on the specific design
of the operation.
It is important to realize that if the error associated with
the sample collection or preparation phase is large, then the
best laboratory quality assurance program is inadequate. Thus a
manager should apply the greatest amount of emphasis to the phase
that contributes the largest component of error; this will not be
possible if the quality assurance design does not provide for
error evaluation of the major measurement phases.
The following sample design (Figure 1) is proposed as a
complete quality assurance design that will allow determination
and control of the various components of measurement bias and
precision. It is assumed that only one analytical laboratory is
utilized; nevertheless, the design can be applied to multiple
laboratories. A multiple-laboratory approach is not discussed
here for simplicity. The samples, discussed in the design, were
defined previously in Table 4.
Figure 1 depicts how quality assessment samples of several
types are treated at the sample collection, preparation, and
analysis stages. Starting at the left of the diagram, the
collection of a field duplicate is shown. At the location
selected for the duplicate, two collocated samples are collected.
One is designated as the routine sample (RS), the other as the
field duplicate (FD). During the preparation phase, after a
routine field sample has been homogenized, a subsample is taken
to be used as the routine laboratory sample. If an additional
subsample is taken from the routine field sample, this additional
sample is called a preparation split. In a similar manner, a
subsample is also obtained from a field duplicate to provide the
laboratory sample from which a concentration measurement for the
22
-------
FD will be obtained. These samples are then forwarded to the
analytical laboratory for analysis.
Moving to the middle of the diagram, paired evaluation
samples are shown entering the process at two stages. In
general, field evaluation samples (FES) are introduced as early
in the collection and packaging process as possible. However, in
the case of soil sampling, it is normally not possible to pass
them over the sampling tools, so they enter the process
immediately after the collection step. These samples are then
subjected to the same handling and analysis procedures as the
other samples. The external laboratory evaluation samples (ELES)
are introduced after the preparation stage in such a way that
they cannot be identified as QA samples by the laboratory and
thereby serve as double-blind samples. These samples are then
subjected to the same analytical procedures as routine samples.
On the right side of the diagram, two types of rinsate
blanks are shown. The field rinsate blank (FRB) is used to check
for sample-collection equipment contamination, and the
preparation rinsate blank (PRB) is used to check for preparation
equipment contamination.
Some consideration must be given to how quality assessment
samples should be assigned to batches of routine samples. Each
batch should contain either one pair of field evaluation samples
or none. Typically, external evaluation samples will only be
assigned to batches of samples containing field evaluation
samples, and, in such cases, only one pair will be assigned to a
batch. Any particular batch may contain zero, one, or several
field duplicates and their associated routine samples. However,
some attempt should be made to distribute field duplicates
throughout the batches from the beginning to the end of the
study. Rules for the assignment of preparation splits, and
associated routine samples, to batches are the same as for field
duplicates. As a general rule, each batch should contain one
field rinsate blank and one preparation blank.
23
-------
Figure 1
QUALITY ASSESSMENT SAMPLES
Dunlk-.ilcs and Snlits
F.valvr.ition Samples
Klnnks
Sample Taking
Preparation
Phase
Analysis
Routine
Sample
Field
Duplicate
Routine
Subsample
Prep. Split
Subsamole
CM:
FD
Subsample
FES"] ["FES"!
No Preparation
[ELESl - JELES]
IFES | [FES i [ELES] IELES] |FRB| \PRQ
24
-------
Field Duplicates (FD)
Sample Taking
Preparation
Phase
Analysis
Routine _ Field
Sample Duplicate
Routine FD
Subsample Subsample
RS FD
Function of Field Duplicates:
To provide data required to
estimate total measurement
error variance minus between-
batch error variance (a*-cr£).
In other words, field
duplicates can be used to
estimate the sum of all
measurement-error variance
components except the
between-batch error variance
component. To assess
between-batch errors, field
evaluation samples or
external lab evaluation
samples may be used.
NUMBER REQUIRED:
Since field duplicates are employed in the estimation of total
measurement error variance and since an estimate of this variance
is required, at least 20 pairs (i.e., routine sample and field
duplicate (co-located) sample) or 10 triples (i.e., routine
sample and two field-duplicate samples) must be obtained to meet
the minimal 20 degrees of freedom objective.
25
-------
Preparation Splits (PS)
Sample Taking
Preparation
Phase
Analysis
Function of Preparation
Splits:
To provide data required to
estimate the sum of
subsampling and analytical
variances (aj«+aj) To
accomplish this, the split
must be performed before the
sample arrives at the
analytical laboratory.
NUMBER REQUIRED:
If the estimation of the components of variance is an important
objective of the project, then one should follow the 20 degrees
of freedom rule and run at least 20 preparation pairs (i.e.,
routine subsample and preparation-split subsample). However,
unlike estimation of the total measurement error variance,
estimation of variance components may be unnecessary in the
quality evaluation of some projects. If the estimation of
variance components is unnecessary, then the only reason for
preparation splits might be for quality control purposes and the
number would be determined in terms of the quality control
requirements.
26
-------
Field Evaluation Samples (FES)
Sample Talcing
Preparation
Phase
Analysis
Function of Field Evaluation
Samples (in pairs)
To provide data which when taken
in conjunction with the data
obtained from field duplicate
samples and their associated
routine samples allows one to
obtain unbiased estimates of the
total measurement error variance,
a*, the between-batch error
variance, al, and the sample-
collection error variance, a*.
The data from the FES also allow
estimation of the handling error
variance, al, and of the total
measurement bias minus the sample
collection biases (B.-B.-B.J .
NUMBER REQUIRED:
Since field evaluation samples are employed in the estimation of
the total measurement error variance and since an estimate of
this variance is required, at least 21 field-evaluation pairs
must be obtained to meet the minimal 20 degrees of freedom for
all variance estimates. (The number of required pairs is 21
rather than 20 because one of the variances being calculated from
the data is the variance of the paired sample averages;
therefore, 21 averages are required to obtain a variance estimate
with 20 degrees of freedom.)
27
-------
External Laboratory Evaluation Samples (ELES)
Sample Taking
Preparation
Phase
No Preparation
Analysis
Function of ELES:
To provide data required to
estimate the sum of the biases
due to analysis and to data
handling (B.+B«.), and analytical
error variances (a*).
NUMBER REQUIRED:
If the estimation of the components of variance is an important
objective of the project, then one should follow the 20 degrees
of freedom rule and run at least 20 laboratory evaluation pairs.
However, unlike estimation of the total measurement error
variance, estimation of variance components and bias components
may be unnecessary in the quality evaluation of some projects.
If the estimation of variance and bias components is unnecessary,
then the only reason for laboratory evaluation samples might be
for quality control purposes and the number would be determined
in terms of the quality control requirements.
28
-------
Preparation Rinsate Blanks (PRB)
Sample Taking
Preparation
Phase
Analysis
PRB
PRB
Function of PRB:
To provide data required to
estimate the sum of bias
caused by contamination,
analysis and data handling
29
-------
Field Rinsate Blanks (FRB)
Sample taking
FRB
Preparation
Phase
Analysis
FRB
Function of FRB:
To provide data required to
estimate the sum of the bias caused
by contamination at the time of
sampling and at the laboratory and
by analysis and data handling
(8,+B.e+B..).
30
-------
Equations for Estimating Bias and Precision
Once the analytical results are received, computation of
bias and precision values is the next step. Bias may be
expressed as the difference between the measured concentration of
the evaluation samples and the reference or known value for the
evaluation sample. A reference value and an expected range of
values are usually available for evaluation samples. If the
measured values are within this range, then one can say that bias
has not been detected. If the measured values are outside this
range, then bias may be present and its amount may be estimated
from the differences between the measured and the reference
values.
Precision is usually described by variance, although
standard deviations are sometimes used. However, standard
deviations are not additive, while variances are. Table 5
provides equations for the variance estimates. Most of these
equations are based on the statistical definitions of variance
for the difference between paired values. Subscripts 'Vjs" and
"BWS" refer to within and between-batch variances, respectively,
which are computed from paired field-evaluation samples.
Subscripts nin and "," refer to individual samples in a pair. In
developing these equations, it is assumed that splits and field
duplicates were assigned to sample locations such that no
location had both a field-duplicate and a preparation split. If
duplicates and splits are assigned to the same locations, some of
the above variance formulas must be modified. However, all the
above variance components can be estimated in either case. The
symbol "n" always represents the number of pairs involved. It
is also assumed, as will typically be case, that the field
evaluation samples are of the same size (weight or volume) as the
routine laboratory samples forwarded from the preparation phase:
this means that there is no subsampling of the FES in the
preparation phase. Triples are not considered here.
Details for computing variance estimates for total
measurement error, sample collection, sample handling,
subsampling, and analytical error are provided in Table 5. If
the estimates of variance components, involving differences of
variance estimates, s% yield negative values, the reported
estimate is zero.
31
-------
Table 5
EQUATIONS FOR DETERMINING PRECISION AND BIAS
PRECISION
Field Evaluation Samples sWFES = Z [FESii-FES2i]2/(2n)
LI
Field Evaluation Samples 8^5 = 2% [FESi-FES]2/(n-l)
where FES{ = (FESii+FES2i)/2,
n
_
and FES = (FESli+FES2i)/(2n)
i=l
n
External Lab Eval. Samp, s^s = J) [ELESu-ELES2i]2/(2n)
|)ata Source Estimate Parameter Estimated
n
Field Duplicates sfe = £ [RSi-FDi]2/(2n) (<&-<%) =
Prep. Splits s^s = £ [RSi-PSi]2/(2n) (ai+ais)
S£D ~ SWFES -
(S!FES -
<*ss
SWFES- SWLES
SWLES
32
-------
Table 5 (continued)
SUGGESTED BIAS FORMULAE
Field Evaluation Sample (FES): Bias = 100(F-R)/R %
where R is the reference value for the FES, and F is the
reported measurement of the FES.
2. Field Rinsate Blank (FRB): Bias - 100 X/CRDL %
where X is the measured value of the FRB, and CRDL is the
Contract Required Detection Limit.
3. Preparation Rinsate Blank (PRB): Bias = 100 P/CRDL %
where P is the measured value of the PRB, and CRDL is as
defined above.
4. Pre-Digest Spike: Bias = 100 (SSR-SR-SA)/SA %
where SSR is the spiked sample result, SR is the sample
result, and SA is the spike amount.
5. Post-digest Spike: (same formula as for the pre-digest spike)
33
-------
AN ALTERNATIVE QA DESIGN THAT DOES NOT EMPLOY FES AND ELES
The quality assurance design given in the preceding section
employed field evaluation samples (FES) and external laboratory
evaluation samples (ELES). It was assumed that these samples
would be double blind samples of homogeneous soil and that the
soil would be very similar (i.e., similar in soil type, in
concentrations of the pollutants of concern, and in
concentrations of other possible chemical interferents) to that
to be sampled in the study. This usually implies that a fairly
large quantity (or quantities) of soil should be collected from
the study site, sent to a laboratory to be dried, mixed, sieved,
and split into homogeneous subsamples to be used as FES and ELES.
It also requires analysis of a sufficient number of samples by
the laboratory to establish the homogeneity of the samples, and
the sending of samples to a number of other laboratories to
establish a reference value for the FES and ELES. This is a
time-consuming process, and the time required for the process may
not be available to the RPM prior to the start of the soil
sampling study. This section addresses how one may plan the
study without FES and ELES so as to still be able to search out
bias sources, to estimate some error variance components, and to
estimate total measurement error variance.
The basic use of the FES in the preceding section was to
estimate between batch variance. As an alternative, it is
suggested that additional field duplicates may be employed for
this purpose. One may go back to a particular sampling location
(e.g., a point at which one sample of soil is taken), and take a
fresh (collocated) sample to include with each batch (or with at
least 21 randomly selected batches if there are a larger number
of batches). If it is difficult to take so many collocated
samples from one sampling location, one might use two or three
such locations and take collocated samples to include in the
batches, alternating between locations (e.g., for two sampling
locations A and B, batch 1 has a collocated sample from location
A, batch 2 from location B, batch 3 from location A, ...). By
comparing the variability between collocated samples that are
collected and analyzed in different batches, with the variability
within the field-duplicate-and-associated-routine-sample pairs,
one can estimate the variability contributed by changes in the
measurement process between batches. These collocated samples
are actually field duplicates, but because they are used in a
different way than the field duplicates encountered in the
previous section, they will be identified as batch field
duplicates (BFD). Equations for determining the variance
estimates using this procedure are given in Table 6. The
assumption stated for Table 5 that field duplicates and
preparation splits are not associated with the same sampling
location is again applied in Table 6. It is shown in Table 6
that the variance component associated handling cannot be
separated from that associated with sample collection, and that
34
-------
Table 6
EQUATIONS FOR DETERMINING PRECISION WITHOUT FES AND ELES
Data Source Estimate Parameter Estimated
n
Field Duplicates s2^ = £ [RSi-FDi]2/(2n)
n
Prep. Splits s^s = £ [RSi-PSi]2/(2n)
i»l
m
Batch Field Duplicates* s^ = £ [BFDi-BFD]2/(m-l) ^
or
L °* L
Batch Field Duplicates11 s&ro = Z £ fBFDij - BFDj]2/£ (mj-1) ^
4i
S2)
Sfq
sS« ci •+• oi«
fO « ^*5a
a This equation is appropriate when the m batch field duplicate samples are all taken from
one sampling location. BFD is the sample mean of the m samples.
b This equation is appropriate when the batch field duplicate samples are from L locations
with mj (>1) BFDs coming from sampling location j. BFDj is the sample mean of the
m.- samples taken for location j.
35
-------
the variance component associated with subsampling cannot be
separated from that associated with analysis. This loss of
information is a consequence of not using FES and ELES in the
study.
The bias detection allowed by use of FES and ELES may be
again obtained at least in part by the introduction of well-
characterized single-blind samples, containing the contaminants
of interest, that are already available from previous studies or
from EMSL-LV. These are single-blind samples, since the
laboratory analyst will probably be able to distinguish them from
the routine samples. These samples will be denoted here by FES1
and ELES1. It will not be necessary to run these samples in
pairs as they will not be used in variance estimation. Figure 2
is a diagram of this alternative study that plays the same role
as Figure 1 did for the procedure involving FES and ELES.
36
-------
Figure 2
Alternative Quality Assessment Sample Design
Duplicates and Snlits
Evaluation Samples Rlank.s
Sample Taking
Routine
Sample
Field
Duplicate
Preparation
Phase
Analysis
Prep. Split
Subsample
FD
Subsample
No Preparation
ELESI
IFESH 1 ELES1 I iFRBl (PRSl
37
-------
EXAMPLE OF QUALITY ASSESSMENT
The purpose of this example is to show how the guidance in
this document can be implemented. Data used in this example were
obtained from an actual Superfund site which was contaminated by
lead deposition from a smelter; however, arrangement of the data
into batches and data from field evaluation samples are fictional
and are included for illustrative purposes.
QA/QC from a Pilot Study
A pilot study was conducted over a representative area to
determine spatial variability and extent of the lead
contamination in order to develop an efficient sampling network
for obtaining representative measurements of contamination over a
large area. Since measurement variability is known to contribute
to the overall variability of data from a sampling effort, a
quality assessment program was implemented to assess the
variability from the collection, handling, and analysis of the
samples. Data from the quality assessment program were intended
for use in determining if the measurement variability was so high
as to prevent accurate assessments of the spatial variability
from being made and whether corrective actions would be required
to reduce the measurement variability. One sampling crew and one
laboratory were selected for collection and analysis of the
samples.
The quality assessment samples are identified in Figure 1
and defined in Table 4. A laboratory control sample was
recommended at a rate of one per batch. This sample was obtained
from the EMSL-LV EPA laboratory (9), and the acceptable
concentration range was provided to the laboratory. This sample
was used by the analytical laboratory as a quality control
sample; thus, it could not be used to estimate analytical
laboratory bias because it was a non-blind sample.
Field evaluation samples were made by sampling 50 kg of a
soil type, which was the same as that in the contaminated zone,
but was located 5 miles away in an area known from past studies
to have background concentrations of lead. The bulk material was
then processed as follows:
- the material was air-dried for a one week period
- the material was then sieved, and all material that passed a
2-mm sieve was saved (40 kg of air dried material)
- 16 grams of lead were added to the 40 kg of soil (400 ppm)
- the sample was homogenized by rolling the material in a
Teflon-coated drum for 48 hours
- 100 subsamples were made by using a closed-bin riffle
splitter
- 10 subsamples, chosen at random, were then shipped to a
referee laboratory to check the lead concentrations and to
38
-------
verify that the lead was equally distributed in each
subsample
The total number of samples, by type, utilized in this study were
as follows:
routine samples: 180
field evaluation (FES) samples: 6 (3 pairs)
field duplicates (FD): 10
field rinsate blanks (FRB): 10
preparation-split (PS) samples: 10
laboratory control samples (LCS): 10
total samples analyzed: 226
total quality assessment samples: 36
total quality control samples: 10
percentage of QA/QC: 22%
High concentrations of lead were encountered in the field
rinsate blank (FRB) from the second batch of samples sent to the
analytical laboratory. This problem was not detected until after
the 4th batch of samples was sent to the analytical laboratory.
Fortunately, this problem was not observed with later batches.
Nevertheless, the sampling crew was advised of this problem and
told to be more careful. In addition, all samples associated
with that batch were resampled and reanalyzed. This problem was
not evident with the field evaluation samples (FES) because they
could not be used with a split-spoon sampling device. The field
evaluation samples were introduced after the sample was taken out
of the ground. It was also evident that the contamination did
not come from the preparation phase because the preparation
blank was acceptable.
Post Pilot Study Data Analysis
After all data were received from the analytical laboratory
the equations defined in Table 5 were utilized to calculate
estimates of total measurement variance, the sum of sample-
collection and sample-handing variances, and between-batch
variances.
A computer program, entitled "ASSESS"*, was developed from
the equations in Tables 5 and 6, and data were entered into the
program to estimate measurement-error variance components. Data
listed in Table 7 were entered into the program to facilitate the
calculation of the terms described in Tables 5 and 6. The
measured lead concentrations in soil (in mg/kg) are given for 10
3 This is a public-domain program written in Fortran for
use on an IBM PC. It may be obtained by writing to the
Exposure Assessment Division, Environmental Monitoring
Systems Laboratory, P.O. Box 93478, Las Vegas, NV 89193.
39
-------
preparation-split pairs and for 10 field-duplicate pairs. The
amount of data used has been kept small to make it easier to read
and to illustrate the use of the computer program to calculate
variances.
The first step is to determine whether a transformation
(e.g., taking the natural log (In) of the values) is needed to
stabilize the variance. The estimation of variance components
implies that there are unique variances to be estimated that
describe measurement-error variance for all measurements. This
is not the case if measurement error variances change with sample
concentrations. The dependence of measurement error variances on
sample concentrations is frequently encountered. Fortunately,
this problem can be overcome through the appropriate selection
and use of such transformations as are discussed by Hoaglin,
et.al. (1983) and Box and Cox (1964). A typical rule of thumb
used by statisticians is that if the ratio of the maximum
observation to the minimum observation is less than 20, no
variance stabilizing transformation is needed; otherwise, the
need for a variance stabilizing transformation should be
investigated. The information provided by the field-duplicate
pairs and the preparation splits is useful in deciding whether a
transformation is required to stabilize variance with respect to
sample concentration.
For each field-duplicate pair and each preparation-split
pair in Table 7, the sum and absolute difference of the two
measurements is calculated. One compares how the pair absolute
differences change as the pair sums change, which is equivalent
to comparing how the sample standard deviations change as the
associated sample means change. For the field-duplicate pairs,
one observes that the differences associated with the larger sums
tend to be larger than those associated with smaller sums. For
example, the median difference associated with the five largest
sums is 96, while the median difference associated with the five
smallest sums is only 28. Similarly, for the preparation-split
pairs, one finds the median difference associated with the five
largest sums is 28, while the median difference associated with
the five smallest sums is only 1.3. This is reasonably clear
evidence measurement error variances are changing with sample
concentrations and that a transformation is required. There is
insufficient information available in the table to choose an
appropriate variance stabilizing transformation. However, lead
concentration data from other soil sampling studies indicate that
the simple logarithmic transformation (Y=ln(lead concentration))
satisfactorily stabilizes the variances. The log-transformation
was performed on the data in Table 7, and the results are given
in Table 8 along with the variance component calculations and
estimates from the ASSESS program.
40
-------
Table 7. Quality Assessment Data
QUALITY EVALUATION DATA3 Transformed? ~" ~~
Batch
1
4
8
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
its
389.0
246.0
33.4
960.0
221.0
180.0
60.0
87.0
275.0
349.0
474.0
478.0
33.5
33.0
1,360.0
104.0
313.0
201.0
67.0
275.0
FD
410.0
780.0
208.0
221.0
400.0
382.0
33.3
128.0
161.0
199.0
No
PS FES Pairs
448.0 505.0
475.0 488.0
423.0 424.0
430.0
32.1
244.0
72.0
233.0
446.0
32.7
1,340.0
294.0
67.0
ELES Pairs (BS+FD)/24
.0
.0
.0
.0
328.0
.0
870.0
.0
194.0
.0
154.0
.0
374.5
.0
430.0
.0
33.2
.0
116.0
.0
181.0
.0
237.0
|RS-FDJS
.0
.0
.0
.0
164.0
.0
180.0
.0
28.0
.0
134.0
.0
51.0
.0
96.0
.0
.3
.0
24.0
.0
40.0
.0
76.0
(RS+PS)/2
.0
.0
.0
409.5
.0
32.8
.0
232.5
.0
66.0
.0
254.0
.0
460.0
.0
33.1
.0
1,350.0
.0
303.5
.0
67.0
.0
IKS-PS j
.0
.0
.0
41.0
.0
1.3
.0
23.0
.0
12.0
.0
42.0
.0
28.0
.0
.8
.0
20.0
.0
19.0
.0
.0
.0
3 Concentrations in rag/kg
4 The average concentration of the routine sample and field
duplicate is computed for the purpose of determining
whether a transformation of the data is required.
* This is computed to determine whether a transformation of
the data is required and is equal to the standard
deviation times the square root of 2.
41
-------
Table 8. Transforned Quality Assessment Data
QDALITY EVALUATION DATA6 Transformed?
Batch
1
4
8
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
US
5.964
5.505
3.509
6.867
5.398
5.193
4.094
4.466
5.617
5.855
6.161
6.170
3.512
3.497
7.215
4.644
5.746
5.303
4.205
5.617
FD
6.
6.
5.
5.
5.
5.
3.
4.
5.
5.
Total measurement error
016
659
338
398
991
945
506
852
081
293
In
PS FES Pairs
6.105 6.225
6.163 6.190
6.047 6.050
6.064
3.469
5.497
4.277
5.451
6.100
3.487
7.200
5.684
4.205
ELES Pairs (RS+FD)/27
.000
5
6
5
4
5
6
3
4
5
5
.000
.000
.000
.761
.000
.763
.000
.265
.000
.932
.000
.923
.000
.058
.000
.501
.000
.748
.000
.192
.000
.455
IRS-FD!"
.000
.000
.000
.000
.511
.000
.208
.000
.145
.000
.932
.000
.136
.000
.224
.000
.009
.000
.208
.000
.222
.000
.323
(RS+PS)/2
.000
6
3
5
4
5
6
3
7
5
4
.000
.000
.014
.000
.489
.000
.448
.000
.186
.000
.534
.000
.131
.000
.499
.000
.208
.000
.715
.000
.205
.000
ISS-PSJ
.000
.000
.000
.100
.000
.040
.000
.099
.000
.182
.000
.166
.000
.061
.000
.024
.000
.015
.000
.063
.000
.000
.000
variance .077
Sample collection variance
Between batch variance
Subsampling variance
Handling variance
Insufficient samples
.004
Insufficient samples
Insufficient samples
for the computation
for the computation
for the computation
to be
to be
to be
made
made
made
Concentrations in mg/kg
The average concentration of the routine sample and field
duplicate is computed for the purpose of determining
whether a transformation of the data is required.
This is computed to determine whether a transformation of
the data is required and is equal to the standard
deviation times the square root of 2.
42
-------
Figure 3 from the ASSESS
program further illustrates
the need for a transform of
the original data. The
standard deviation of the data
from routine samples and field
duplicates increases with the
average concentration. A
logarithmic transform of the
data will stabilize the
standard deviation of the data
over the measured
concentration range. After
this has occurred, the
variances may be computed for
the purpose of assessing
variability throughout the
measurement process. The
ASSESS program allows these
calculations to be performed
easily.
1
1
1
•
*
1
*
*
12A
i
I
<*
0
1
• A
I •
t-
I
t
139. 3t
+
*
4
i. 43B. «•
j.
•. 73i. M
•.
few*** can*«ntMtU» at n « F»
Figure 3. Scatter Plot of QA
Data from ASSESS
The variances reported in Table 8 indicate that the sum of
the variances arising from sample collection and from sample
handling amount to about 7/8ths of the total measurement-error
variance. The estimate of the sum of the variances caused by
sample collection and sample handling is given by (see Table 5)
Sro2 - Sp.2 = 0.0730 - 0.0045 = 0.0685,
while the estimate of total measurement-error variance is given
by
n
- SmaI)/2 = 0.0730 + (0.0100 - 0.0025)/2 = 0.0768
However, the estimate sBraaJ, based on only 2 degrees of freedom (3
pairs -1), may underestimate the true variance a^,3 by a factor
of 30 or more (Table 3). The other estimates, s2, of variance
are based on 10 or fewer degrees of freedom and may underestimate
the true variances by factors of 3 or more. If all of the
estimates of variance had been based on at least 20 degrees of
freedom each, one would have much more confidence in the
estimates and would certainly be justified in instituting a more
rigorous training program for the sample-taking crews and in
considering an increase in the volume of soil in each sample. In
point of fact, the sum of the variances caused by sample
collection and sample handling was such a large portion
(approximately 1/3) of the total (measurement plus spatial)
variation that action was taken to reduce its contribution in the
primary study. While more data from the use of field evaluation
samples and external laboratory evaluation samples would have
43
-------
been useful in implementing the rationale for assessing errors
and variability in all phases of the pilot study, sufficient
information was provided from the existing quality assessment
samples to begin making some changes. Tighter adherence to the
rationale during the main part of the study would ensure that
sufficient data were available to accurately assess the
significance and sources of variability during the study of the
entire site.
Estimates at UartMtt* f
236M.
158W.
i
r*» tt San»l«s
sM svs safes shfrs
Figure 4.
Error Plot from
ASSESS
Figure 4 from the ASSESS
program illustrates the range
in which the estimates of the
various variance components
can be expected to occur
within a 95% confidence
interval. It is clear from
the length of the line for sBFES2
that greater use of field
evaluation samples would have
improved the assessment of
between-batch variability as
well as of the total
measurement error and sample
collection variances.
Development of the QA Plan for
the Primary Study
The DQOs for the primary
study were developed with the
background data from the pilot
study. Goals were established
for accuracy, precision, completeness, comparability and
representativeness of the data to be collected during the
expanded study. Historical data on variability in major portions
of the measurement process were input into the ASSESS program for
reference* Based upon the limited number of FES, lack of ELES in
the pilot study and the importance of having reliable assessments
of data quality throughout the measurement process, Table 3 was
used to determine the added number of QA samples that were
required to better estimate variability during sample collection,
handling, transportation, subsampling and analysis. Even though
the personnel, procedures, and analytical equipment would be
identical in the primary study to that in the pilot study, the
decision-makers felt that 20 degrees of freedom were needed for
all quality assessment samples to permit assessments of
variability to be within a factor of roughly two to the actual
value, at least to the 95% confidence interval. These
assessments would confirm that the changes made during the pilot
study were effective in reducing sampling and measurement
variability to acceptable levels, i.e., to permit spatial
variability to be accurately assessed.
44
-------
RECOMMENDATIONS FOR FUTURE WORK
1. Greater attempts to define and standardize QA/QC terms need
to be made in conjunction with the Quality Assurance
Management Staff (QAMS) within the Office of Research and
Development (ORD) and the program offices.
2. Protocols and materials for the preparation of QA/QC samples
for the field need to be reviewed further and described in
greater detail. In some cases, new materials and protocols
will need to be developed and standards established.
3. The rationale presented in this document needs to be
developed further to integrate it with the work of QAMS.
Data quality objectives (DQOs) are important in determining
the level of QA/QC for a study, and QAMS effort to develop a
standardized approach to the development of DQOs through the
use of computer software could incorporate the rationale
presented in this document. It appears that this rationale
could be translated into a spreadsheet or expert systems
computer program.
4. Greater characterization of commonly used sampling methods
needs to be made. The choice of a sampling method
determines to some extent the amount of QA/QC involved in a
study. For methods such as the portable x-ray fluorescence
instrument, the volume of earth sampled, minimum detection
level, interferences, and range of contaminants detected are
some of the characteristics that need to be defined, when
practicable, on a scale common to the other sampling
methods.
5. The rationale presented in this document needs to be
evaluated at several actual Superfund site investigations.
If the rationale proves to be workable and worthwhile, the
rationale needs to be adopted for use at all Superfund site
investigations to try to achieve a uniform measure of data
quality from all of the investigations of inorganics in
soil.
6. Training may be conducted prior to a study or during a
study. The optimum approach is to complete the training and
evaluation of sampling crews prior to the initiation of a
major study. The feasibility of establishing a national
training/certification program for sampling crews should be
considered further.
7. Specific sub-sampling techniques need to be defined and
developed for utilization in the field and the laboratory.
45
-------
REFERENCES
(1) Bennet, Tom. 1989. standard Operating Procedures. Region
IV Analytical Services Branch, U.S. Environmental Protection
Agency.
(2) Box, G.E.P., and D.R. Cox. 1964. The Analysis of
Transformations. Journal of the Royal Statistical Society,
Series B. Vol. 26(2):211-243.
(3) Drouse, S. K., D.C. Hillman, L.W. Creelman, and S. J. Simon,
1986. National Surface Water Survey, Eastern Lake Survey
(Phase 1 - Synoptic Chemistry) Quality Assurance Plan.
EPA/600/4-86/008. U.S. Environmental Protection Agency,
Environmental Monitoring Systems Laboratory, Las Vegas,
Nevada.
(4) Hoaglin, D.C., F. Hosteller, and J.W. Tukey. 1983.
Understanding Robust and Exploratory Data Analysis. John
Wiley & Sons. New York, NY
(5) La Barge, R. 1989. "A Programmatic Approach to Achieving
Data Quality Objectives." presented at 2nd. Annual Symposium
for Quality Assurance for Ecological Monitoring, Feb. 1989.
U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory, Las Vegas, Nevada.
(6) Morey, Debra. 1989. "Standard Operating Procedures. U.S.
Environmental Protection Agency, Region VIII.
(7) Taylor, J.K. 1987. "Quality Assurance of Chemical
Measurements"
(8) U.S. DOE. 1987. The Environmental Survey Manual. DOE/EH-
0053.
(9) U.S. EPA. 1987. Data Quality Objectives for Remedial
Response Activities - Development Process. EPA/540/G-
87/003.
(10) U.S. EPA. 1988. USEPA Contract Laboratory Program -
Statement of Work for Inorganics Analysis - Multi-Media
Multi Concentration. SOW No. 788.
(11) U.S. EPA. 1988. Sampling for Hazardous Materials.
Environmental Response Team, Office of Emergency and
Remedial Response, Hazardous Response Support Division,
Washington, D.C.
(12) U.S. EPA. 1980. Quality Assurance Plan Love Canal Study.
LC1-619-026, EPA Contract 68-02-3168.
46
-------
(13) U.S. EPA. 1989. Direct/Delayed Response Project: Quality
Assurance Plan for the Mid-Appalachian Region of the United
States (Draft). Environmental Monitoring Systems
Laboratory, Las Vegas, Nevada.
(14) U.S. EPA. 1984. Soil Sampling Quality Assurance Users Guide
(1st. Edition). Environmental Monitoring Systems Laboratory,
Las Vegas, Nevada. EPA/600/4-84-043
(15) U.S. EPA. 1989. Soil Sampling Quality Assurance Users Guide
(2nd. Edition). Environmental Monitoring Systems Laboratory,
Las Vegas, Nevada. EPA 600/8/89/046
47
-------
APPENDIX A
KEY COMPONENTS
A rigorous program to assess the quality of data cannot be
developed and implemented if key components for a field study are
neglected. Those components include training, pilot studies, a
variety of audits to assess the effectiveness of the QA/QC
program, and documentation.
Training
Training is an integral part of an effective quality
assurance program. It should furnish the essential knowledge
needed by all study participants to assure that plans, methods,
and procedures are all accomplished as designed. The training
program should review key principles and point out changes in
protocols to experienced workers. At the same time it should
orient the new employee to all the study methods. Training
should always occur at the earliest possible time before a study
in order to give personnel, time to make adjustments. However,
training typically occurs, in varying degrees, throughout multi-
phased studies as personnel change and more is learned about the
site, methods and procedures. Training may be formal, or
informal, in the classroom, or on the job. Training should
include lectures on sampling principles, a demonstration of
procedures, question and answer sessions, and hands-on sampling
practice.
A practical way to implement a training program is to integrate
it into a "preliminary" study. However, additional training may
also occur during a "pilot" study as personnel are evaluated on
the methods and procedures that are expected to be used during
the main study.
Pilot Study
Pilot studies may serve as the impetus for further training
before the full-scale study. The purpose of a pilot study is to
evaluate the logistics, equipment, sampling plans and analytical
protocols prior to implementation of the full study. It should
also provide a test of the study design, quality assurance
design, and data interpretation plan. Pilot studies are
recommended for all programs so that state-of-the-art measurement
methods and study designs can be fully tested before the full
study begins. In order to be useful, the pilot study should
employ all of the plans for the full-scale project, including the
same personnel, management structure, equipment, and procedures.
After the pilot study is complete, the study methods can be
carefully assessed to see whether or not changes need to be made
before the full-scale study starts. The data must be interpreted
so that the study methods may be evaluated in relation to how tha,
48
-------
study objectives are being met. It is critical that managers
provide time and resources to evaluate the pilot study and
incorporate changes before the start of the next phase of a site
investigation.
The guidance in this document, to assess errors in field
sampling, may be used to assess the proficiency of a training
effort by allowing the measured errors to be compared against
stated data quality objectives.
Audits
An adequate QA program ensures that the quality of the final
product meets the DQOs. Audits are an integral part of the QA
process and are vital for assuring that program procedures are
being implemented.
Audits are performed to document the implementation of the
quality assurance program plan, quality assurance project plan
and/or associated operational protocols. Four specific kinds of
audits can be used to determine the status of the measurement
systems, the adequacy of the data collection systems, the
completeness of documentation of data collection activities and
the abilities of the program management to meet the mandated data
collection and data quality objectives. These four audit types
are respectively. Performance Audits, Technical System Audits,
Data Quality Audits and Management System Audits.
* Performance Audits (PA) are generally based on Quality
Assessment or Evaluation (QE) samples. Samples having
known concentrations may be tested as unknowns in the
laboratory or a sample may be analyzed for the presence
of certain compounds. Performance audits are used to
determine objectively whether an analytical measurement
system is operating within established control limits at
the time of the audit.
* Technical System Audits fTSAl are qualitative on-site
audits that evaluate the technical aspects of field
operations against the requirements of the approved
protocols and QA plans. TSA reports will note any
problems, allowing corrective action to be taken to
protect the validity of future data.
* Data Quality Audits (DOAT are evaluations of the
documentation associated with data quality indicators of
measurement data to verify that the generated data are of
known and documented quality. This is an important part
of the validation of data packages showing that the
methods and SOPs designated in the QA plans were
followed, and that the resulting data set is a functional
49
-------
part of satisfying the established DQOs. The results are
vital to decisions regarding the legal defensibility of
the data should it be challenged in litigation.
* A Management System Audit (MSA1 is a formal review of an
entire program, e.g., a review of a state's QA program,
or a review of a state-contracted Laboratory. In a MSA,
key elements in the program, e.g., lab certification
program, QC in field operations, and QC in the certified
lab, are evaluated to see if QA is being implemented. If
deficiencies are detected, corrective actions are
suggested and implementation monitored.
The guidance in this document is particularly useful for the
performance audit.
Documentation
Documented procedures should be developed prior to a study
and followed. Errors can increase and blunders can occur in any
measurement program through inadequately prepared and reviewed
documentation. Data transcribed onto paper or recorded on
magnetic media must be checked for accuracy on a timely basis by
qualified personnel. Measures to assess and minimize errors, as
described in this document, are not going to be effective if
adequate documentation is not developed and reviewed on a timely
basis.
It is important that field personnel follow specified,
documented procedures. If changes in program execution and
design (e.g., sample site selection, number of samples to be
collected, sampling intervals, and tools) are required, these
changes, with appropriate rationale, must also be documented.
50
-------
APPENDIX B
DEFINITIOHS
Quality Assurance
A system of activities whose purpose is to provide to the
producer or user of a product or service the assurance that it
meets defined standards of quality. It consists of two separate,
but relate activities, quality control and quality assessment.
Quality Control
The overall system of activities whose purpose is to control the
quality of the measurement data so that they meet the needs of
the user.
Quality Assessment
The overall system of activities that provide an objective
measure of the quality of data produced.
Soil
The soil referred to in this document encompasses the mass
(surface and subsurface) of unconsolidated mantle of weathered
rock and loose material lying above solid rock. Further, a
distinction must be made as to what fraction of the
unconsolidated material is soil and what fraction is not. The
soil component here is defined as all mineral and naturally
occurring organic material that is 2 mm or less in size. This is
the size normally used to distinguish between soils (consisting
of sands, silts, and clays) and gravels. In addition, the 2-mm
size is generally compatible with analytical laboratory methods,
capabilities, and requirements.
The non-soil fraction (e.g., automobile fluff, wood chips,
various adsorbents and mineral/organic material greater than 2-mm
in size) must also be addressed in sampling and monitoring. This
fraction may contribute and/or contain a greater amount of
contaminant(s) than the associated soil fraction. At sites in
which this occurs, reporting contaminant levels only in the soil
fraction will ultimately lead to inappropriate and incorrect
decision making. Decision makers must realize that a number of
problems are normally encountered in obtaining and using data
from the non-soil components. For example, questions arise
concerning the validity of data obtained from the analysis of
materials that do not meet the size and volume requirements for
which the analytical processes were validated. Also, standard
reference and audit materials are not available to substantiate
and validate the analytical results.
51
-------
The current recommended procedures are to identify and
record the type and volume of non-soil material for each sample
collected with a minimum of 10 percent (%) of these non-soil
samples submitted for analysis. Data from the non-soil material
are important to the assessment of the representativeness of the
soil sampling/monitoring program. The behavior of contaminants
in the soil environment is a function of the contaminant's and
soil's physical and chemical properties. Soil sorption (the
retention of substances by adsorption or absorption) is related
to properties of the contaminant (e.g., solubilities, heats of
solution, viscosity, and vapor pressure) and to properties of
soils (e.g., clay content, organic content, texture,
permeability, pH, particle size, specific surface area, ion
exchange capacity, water content, and temperature). The soil
components that are most associated with sorption are clay
content and organic matter. The soil particle surface
characteristics thought to be most important in adsorption are
surface area and cation exchange capacity (CEC).
Standard Additions
A procedure called standard additions is commonly used to
detect bias in chemical analysis. In this procedure, known
amounts of standard solutions are added to aliguots of soil
samples. It is recommended that this be done in the field or in
a field laboratory. The main problem encountered is that mixing
soils to obtain homogeneity is difficult in a laboratory, and
even more so in the field. Several known quantities of the
standard are added to the aliguots of the soil samples. The
analytical results should follow a straight line:
y = a + bx,
where x is the increase in concentration caused by the addition
and y is the value obtained by the laboratory. Bias is indicated
if the data do not follow a straight line, or if a < 0. If the
units of x and y are the same, the value of b should be near one,
and a significant deviation from one would indicate a
proportional bias.
52
-------
APPENDIX C
Soil Sampling Methods Table
Type at Sampler
Most Suitable
Core Types
Access to
Sampling
Most Suit- Sites
able Soil During
Moisture Poor Soil Relative
Conditions Conditions Saaple Sue
D»pth
Cho'lessb Cither r«v onfav wet Dry Cither
No Ssull
Shallow Deep Either 1 2/Ker> Sat
I. Mind-H«ld
Spoons
Scoops
5cr»w-eyp« Auq«ci
3arr«l Au9*rs
post-noi« Augvr
Dutch Aug«c
R«gul«r 3«rr«l
Auq«r
S*nd Aua*rs
Kuo, Aug*rs
Soil ?rob»»
w*t Tips
Dry Tips
Finn-nail«d Tub*
Sasiplsrs
n. Power Driven
Auger
Hand-Held Screw Type
Power Auger
Truck Mounted Auger
Tripod Mounted Drive s
Sampler
Split Spoons x
Cobecive Soils (e.g., city)
Coheciveleis soils (e.g., dry send)
53
-------
APPENDIX D
THE LOGNORMAL DISTRIBUTION AND LOGARITHMIC TRANSFORMATIONS
If the random variable W = Ln(X) has a normal distribution
with mean nw and variance aw2 (i.e., N(nw,aw2) ) , then the random
variable X has a lognormal distribution (Johnson and Kotz, 1970,
Chapter 14') with mean
E(X) = n, =
with variance
V(X) = a,2 = iix2(exp(av2) - 1)
and with standard deviation
-------
APPENDIX E
NON-BLIND
QUALITY ASSESSMENT SAMPLES IN THE CONTRACT LABORATORY PROGRAM
1. Laboratory Control Sample CLCS1 - A sample of well-
characterized soil, whose analyte concentrations are known
to the laboratory, is used for internal laboratory control
(10,15). This sample is also called a quality control audit
sample (13).
2. Pre-digest Spike Sample - A routine sample in which a known
quantity of analyte is added to an aliquot of the sample.
It is used to determine bias from the digestion and analysis
of components.
3. Post-digest Spike Sample - A routine sample in which a known
quantity of analyte is added to an aliquot after the
digestion process is completed. It is used to determine
bias from the analytical or detection phase. When used in
combination with a pre-digest spike sample, the bias from
the digestion phase may be determined by difference.
4. Analytical Laboratory Duplicate (ALP) - This sample is a
subsample of a routine sample which is analyzed by the same
method. It is used to determine method precision, but
because it is a non-blind sample, or known to the analyst,
it can only be used by the analyst as an internal control
tool and not as an unbiased estimate of analytical
precision.
5. Initial Calibration Verification (ICV) and Continuing
Calibration Verification (CCV) Solutions - These are
prepared solutions containing known concentrations of
analytes that originated from a different source as the
calibration standards. They are used as an independent
check of the instrument calibration accuracy. The CCV
samples are normally run in an ordered fashion after a
specified number of routine samples.
6. initial Calibration Blank (ICB) and Continuing Calibration
Blank (CCB1 Solution. These are blank samples run at the
same frequency as the ICV and CCV, and they are used to
check for instrument baseline drift.
7. CRDL Standard for ICP and AA - This is a solution standard
at a concentration of two times the CRDL, or two times the
IDL, whichever is greater. It is used during each run in
place of a formal instrumental detection limit determination
to assure the instrument is running properly.
8. Linear Range Verification Check Standard - This standard is
55
-------
a solution of known analyte at concentrations within the
upper limit of the linear range. Above this range, the
samples must be diluted.
9. ICP Interference Check Sample - This sample contains two
parts. Part A contains potential interfering analytes, and
Part B contains both the analytes of interest and the target
analytes. Part A and B are analyzed separately to determine
the potential for interferences.
56
-------
APPENDIX F
Upper Confidence Limits for the Variance, a2, as a Function of
the Number of Decrees of Freedom for the Variance Estimate. s*.
Levels of Confidence (%}
Degrees
Freedom
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
30
40
50
100
of
2_fl
9.49s2
5.13s2
3.76s2
3.10S2
2.72s2
2.47s2
2.29s2
2.ies2
2.05s3
1.97s2
1.90s2
1.85s2
1.80s2
1.76s2
1.72s2
1.69s2
1.66s2
1.63S2
1.61s2
1.59s2
1.57s2
1.55S2
1.53s2
1.52s2
1.46s2
1.38S2
1.33s2
1.21s2
as.
19.49s2
8.52s2
5.63S2
6.01S2
3.67s2
3.23S2
2.92S2
2.71s2
2.54S2
2.40s2
2.29s2
2.21S2
2.13s2
2.07s2
2.01s2
1.96S2
1.92s2
1.87S2
1.84s2
1.81s2
1.78s2
1.76S2
1.73S2
1.71s2
1.62S2
1.51s2
1 . 44s2
1 . 28s2
9_9
99.50s2
26.13s2
13.46s2
9.02s2
6.88s2
5.65s2
4.86s2
4.31s2
3.91s2
3.60s2
3.36s2
3.17s2
3.00s2
2.87s2
2.75s2
2.65s2
2.57s2
2.49S2
2.42s2
2.36s2
2.31s2
2.26S2
2.21s2
2.17S2
2.01s2
1.80S2
1.68s2
1.43S2
57
------- |