FRANK C.

CHIKANI
SES/PGH/25/11
ANALYTICAL DATA HANDLING
There are certain basic concepts in analytical chemistry that are helpIul to the analyst when
treating analytical data. Some oI these concepts include; accuracy, precision, mean, and standard
deviation as related to chemical measurements in the general Iield oI analytical chemistry.
Accuracy
In analytical chemistry, the term 'accuracy' is used in relation to a chemical measurement. The
International Vocabulary oI Basic and General Terms in Metrology (VIM) deIines accuracy of
measurement as... "closeness oI the agreement between the result oI a measurement and a true
value." The VIM thereIore shows that accuracy is a "qualitative concept" and that a true value is
indeterminate by nature. In theory, a true value is that value that would be obtained by a perIect
measurement. Since there is no perIect measurement in analytical chemistry, we can never know
the true value.
Our inability to perIorm perIect measurements and thereby determine true values does not mean
that we have to give up the concept oI accuracy. However, we must add the reality oI error to
our understanding. For example, lets call a measurement we make X
I
and give the symbol Ior
the true value. We can then deIine the error in relation to the true value and the measured value
according to the Iollowing equation:
error X
I
-
Accuracy is oIten spoken in qualitative terms such a "good," "expected," "poor," and so on.
However, we have the ability to make quantitative measurements. We thereIore have the ability
to make quantitative estimates oI the error oI a given measurement. Since we can estimate the
error, we can also estimate the accuracy oI a measurement. In addition, we can deIine error as the
diIIerence between the measured result and the true value as shown in equation above. However,
we cannot use the equation above to calculate the exact error because we can never determine
the true value. We can, however, estimate the error with the introduction oI the conventional
true value which is more appropriately called either the assigned value, the best estimate oI a
true value, the conventional value, or the reIerence value. ThereIore, the error can be estimated
using the equation above and the conventional true value.
Errors in analytical chemistry are classiIied as systematic (determinate) and random
(indeterminate). The deIinitions oI error, systematic error, and random error Iollow:
O Error - the result oI a measurement minus a true value oI the measurand.
O Systematic Error - the mean that would result Irom an inIinite number oI measurements
oI the same measurand carried out under repeatability conditions, minus a true value oI
the measurand.

O Random Error - the result oI a measurement minus the mean that would result Irom an
inIinite number oI measurements oI the same measurand carried out under repeatability
conditions.
A systematic error is caused by a deIect in the analytical method or by an improperly
Iunctioning instrument or analyst. A procedure that suIIers Irom a systematic error is always
gives a mean value that is diIIerent Irom the true value. The term 'bias' is sometimes used when
deIining and describing a systematic error. The measured value is described as being biased high
or low when a systematic error is present and the calculated uncertainty oI the measured value is
suIIiciently small to see a deIinite diIIerence when a comparison oI the measured value to the
conventional true value is made.
Systematic error may also be reIerred to as determinate` because it is more descriptive in stating
that this type oI error can be determined. A systematic error can be estimated, but it cannot be
known with certainty because the true value cannot be known. Systematic errors can thereIore be
avoided, i.e., they are determinate. Sources oI systematic errors include spectral interIerences,
chemical standards, volumetric ware, and analytical balances where an improper calibration or
use will result in a systematic error, i.e., a dirty glass pipette will always deliver less than the
intended volume oI liquid and a chemical standard that has an assigned value that is diIIerent
Irom the true value will always bias the measurements either high or low and so on. The
possibilities seem to be endless.
Random errors are unavoidable. They are unavoidable due to the Iact that every physical
measurement has limitation, i.e., some uncertainty. Using the utmost oI care, the analyst can only
obtain a weight to the uncertainty oI the balance or deliver a volume to the uncertainty oI the
glass pipette. For example, most Iour-place analytical balances are accurate to; 0.0001 grams.
ThereIore, with care, an analyst can measure a 1.0000 gram weight (true value) to an accuracy oI
; 0.0001 grams where a value oI 1.0001 to 0.999 grams would be within the random error oI
measurement. II the analyst touches the weight with their Iinger and obtains a weight oI 1.0005
grams, the total error 1.0005 -1.0000 0.0005 grams and the random and systematic errors
could be estimated to be 0.0001 and 0.0004 grams respectively. Note that the systematic error
could be as great as 0.0006 grams, taking into account the uncertainty oI the measurement.
A truly random error is just as likely to be positive as negative, making the average oI several
measurements more reliable than any single measurement. Hence, taking several measurements
oI the 1.0000 gram weight with the added weight oI the Iingerprint, the analyst would eventually
report the weight oI the Iinger print as 0.0005 grams where the random error is still 0.0001 grams
and the systematic error is 0.0005 grams. However, random errors set a limit upon accuracy no
matter how many replicates are made.
Precision
The term precision is used in describing the agreement oI a set oI results among themselves.
Precision is usually expressed in terms oI the deviation oI a set oI results Irom the arithmetic
mean oI the set. In analytical chemistry good precision does not mean good accuracy. However,
It sounds reasonable to assume otherwise.

Why doesn't good precision mean we have good accuracy? In errors we know that there are
systematic and random errors. We also know that the total error is the sum oI the systematic error
and random error. Since truly random error is just as likely to be negative as positive, we can
reason that a measurement that has only random error is accurate to within the precision oI
measurement and the more precise the measurement, the better idea we have oI the true value,
i.e., there is no bias in the data. In the case oI random error only, good precision indicates good
accuracy.
Let`s add the possibility oI systematic error. We know that systematic error will produce a bias
in the data Irom the true value. This bias will be negative or positive depending upon the type
and there may be several systematic errors at work. Many systematic errors can be repeated to a
high degree oI precision. ThereIore, it Iollows that systematic errors prevent us Irom making the
conclusion that good precision means good accuracy. When we go about the task oI determining
the accuracy oI a method, we are Iocusing upon the identiIication and elimination oI systematic
errors. The statement that 'good precision is an indication oI good accuracy' should not mislead
anyone when determining the accuracy oI a method. Too many systematic errors can be repeated
to a high degree oI precision Ior this statement to be true.
Validation oI an Analytical Method
Once an analytical method has been developed it needs to be validated. The validation process
ensures that the results are reproducible and accurate. Some oI the important parameters that
need to be considered during the validation process are reproducibility, repeatability, bias, limit
oI detection and limit oI quantiIication.
The terms 'repeatability' and 'reproducibility' can be used instead oI the more general term
'precision.' The Iollowing is the deIinition:
O #epeatability (oI results oI measurements) - the closeness oI the agreement between the
results oI successive measurements oI the same measurand carried out under the same
conditions oI measurement.
- These conditions are called repeatability conditions.
- Repeatability conditions include the same measurement procedure, the same observer,
the same measuring instrument, used under the same conditions, the same location, and
repetition over a short period oI time.
O #eproducibility (oI results oI measurement) - the closeness oI the agreement between the
results oI measurements oI the same measurand carried out under changed conditions oI
measurement.
- A valid statement oI reproducibility requires speciIication oI the conditions changed.
- The changed conditions may include principle oI measurement, method oI
measurement, observer, measuring instrument, reIerence standard, location, conditions
oI use, and time.

When discussing the precision oI measurement data, it is helpIul Ior the analyst to deIine how
the data are collected and to use the term 'repeatability' when applicable. It is equally important
to specify the conditions used Ior the collection oI 'reproducibility' data.
O etection Limits
The detection limit is the smallest concentration oI an analyte that can be detected with a given
degree oI conIidence. The detection limit is commonly based on a signal to noise ratio oI 3
times the standard deviation oI the background signal. Where noise is deIined as the width oI the
baseline as shown in Figure 1 (below). Another concept similar to the limit oI detection is the
limit of quantitation. The limit oI quantitation is deIined as the smallest signal that can be
converted to an accurate measurement oI an analyte. The limit oI quantitation can be deIined in
various ways, but it is commonly set at a signal to noise ratio oI 10. At this set limit, the
analytical precision should be better than plus/minus 3.

Figure 1: Graph Showing the Measurement oI Signal to Noise Ratio, Ior a Detector Signal.
Signal A is Below the Detection Limit and Signal B is Above the Limit.
Mean
The deIinition oI mean is, "an average oI n numbers computed by adding some Iunction oI the
numbers and dividing by some Iunction oI n." The central tendency oI a set oI measurement
results is typically Iound by calculating the arithmetic mean ( ) and less commonly the median
or geometric mean. The mean is an estimate oI the true value as long as there is no systematic
error. In the absence oI systematic error, the mean approaches the true value () as the number oI
measurements (n) increases. The Irequency distribution oI the measurements approximates a
bell-shaped curve that is symmetrical around the mean. The arithmetic mean is calculated using
the Iollowing equation:

(X
1
X
2
X
n
) / n
Typically, insuIIicient data are collected to determine iI the data are evenly distributed. Most
analysts rely upon quality control (QC) data obtained along with the sample data to indicate the
accuracy oI the procedural execution, i.e., the absence oI systematic error(s). The analysis oI at
least one QC sample with the unknown sample(s) is strongly recommended.
Even when the QC sample is in control it is still important to inspect the data Ior outliers. There
is a third type oI error typically reIerred to as a 'blunder'. This is an error that is made
unintentionally. A blunder does not Iall in the systematic or random error categories. It is a
mistake that went unnoticed, such as a transcription error or a spilled solution. For limited data
sets (n 3 to 10), the range (X
n
-X
1
), where X
n
is the largest value and X
1
is the smallest value, is
a good estimate oI the precision and a useIul value in data inspection. In the situation where a
limited data set has a suspicious outlier and the QC sample is in control, the analyst should
calculate the range oI the data and determine iI it is signiIicantly larger than would be expected
based upon the QC data. II an explanation cannot be Iound Ior an outlier (other than it appears
too high or low), there is a convenient test that can be used Ior the rejection oI possible outliers
Irom limited data sets. This is the " test.
Standard Deviation
A useIul and commonly used measure oI precision is the experimental standard deviation
deIined as... "For a series oI n measurements oI the same measurand, the quantity s
characterizing the dispersion oI the results and given by the Iormula:
s | _ (xi- )
2
/ (n-1) |
1/2

x
I
being the result oI the i-th measurement and being the arithmetic mean oI the n results
considered."
The above deIinition is Ior estimating the standard deviation Ior n values oI a sample oI a
population and is always calculated using n-1. The standard deviation oI a population is
symbolized as s and is calculated using n. unless the entire population is examined; s cannot be
known and is estimated Irom samples randomly selected Irom it. For example, an analyst may
make Iour measurements upon a given production lot oI material (population). The standard
deviation oI the set (n4) oI measurements would be estimated using (n-1). II this analysis was
repeated several times to produce several sample sets (Iour each) oI data, it would be expected
that each set oI measurements would have a diIIerent mean and a diIIerent estimate oI the
standard deviation.
The experimental standard deviation of the mean for each set is calculated using the Iollowing
expression:

s / (n)
1/2

Apart Irom accuracy, precision, mean and deviation as related to chemical measurements in the
general Iield oI analytical chemistry. There are certain basic concepts in analytical chemistry that
are helpIul to the analyst when treating analytical data. They include; signiIicant Iigures and
uncertainity.
O SigniIicant Figures
When working with analytical data it is important to be certain that you are using and reporting
the correct number oI signiIicant Iigures. The number oI signiIicant Iigures is dependent upon
the uncertainty oI the measurement or process oI establishing a given reported value. In a given
number, the Iigures reported, i.e. signiIicant Iigures, are those digits that are certain and the Iirst
uncertain digit. It is conIusing to the reader to see data or values reported without the uncertainty
reported with that value.
The number oI digits needed to express results consistent with the precision oI the measurement
is called signiIicant Iigures. The digit '0 can be signiIicant or it can be used simply to indicate
the magnitude oI a number. For example: 0.01 has one signiIicant Iigure. The zeros simply
indicate the magnitude oI the number as 'one hundredth. While 0.01 has only one signiIicant
Iigure, 0.10 has two signiIicant Iigures, there is no diIIerence between 0.10, 0.12, 0.13 all have
2 signiIicant Iigures. In this case, the last zero is signiIicant.
Examples oI 3 signiIicant Iigures: 0.00312, 312, 12.0, 1.10, 1.01, 12,400, 93,200,000
It is not really possible to determine the number oI signiIicant Iigures in the number 93,200,000
because there is no indication iI the zeros are actually measured as :ero or are placeholders to
indicate the magnitude oI the number. A better way to represent the number is with scientiIic
notation.
9.32 x 10
-7
has 3 signiIicant Iigures
9.3200 x 10
-7
has 5 signiIicant Iigures
This way oI representing the data clearly indicates the precision oI the measurement.

The number 93,200,000.0 has 9 signiIicant Iigures. The placement oI the zero aIter the decimal
point indicates that the value was measured to the nearest 'tenth thereIore, all the numbers are
signiIicant.
Many experiments require some processing oI the data using addition, subtraction,
multiplication, division, logarithms, etc.
Calculations and reporting such as this are oIten seen:

35.6 g x 1.7 6.05201 grams

As a general rule, the reported value should have the precision oI the least precise value in the
calculation. In this case, 1.7 which has 2 signiIicant Iigures should be used as the basis Ior
reporting the precision oI the Iinal value.
35.6 g x 1.7 6.1 grams (2 signiIicant Iigures)
Note: iI the Iinal value Iollowing the last signiIicant Iigure is 5 or greater, the number is
rounded up, other wise it is rounded down.
Also: II you wish to take the average oI 3 numbers: (12.12 12.54 12.19)/3 12.28
4 signiIicant Iigures is reported. The divisor '3 is an absolute number and does
not play a role in deIining the number oI signiIicant Iigures.
O Uncertainty
Uncertainty is deIined as: "A parameter associated with the result oI a measurement, that
characterizes the dispersion oI the values that could reasonably be attributed to the measurand."
The parameter may be, Ior example, a standard deviation (or a given multiple oI it), or the width
oI a conIidence interval.
Uncertainty oI measurement comprises, in general, many components. Some oI these
components may be evaluated Irom the statistical distribution oI the results or series oI
measurements and can be characterized by standard deviations. The other components, which
also can be characterized by standard deviations, are evaluated Irom assumed probability
distributions based on experience or other inIormation.

REFERENCES
FiIield, F. W.; Kealey, D. !rinciples and !ractices of Analytical Chemistry, 5th ed.; Blackwell
Science, Ltd.: OxIord, 2000; 562 pages.

Skoog, D. A.; Holler, F. J.; Nieman, T. A. !rinciples of Instrumental Analysis. 1998, 849 pages.
LC/GC Magazine Online: http://www.chromatographyonline.com/lcgc/