Errors in Chemical Analysis

Steps in a Typical Quantitative

Analysis
• Data of unknown quality are useless!
• All laboratory measurements contain
experimental error.

• It is necessary to determine the

magnitude of the accuracy and
reliability in your measurements.
• Then you can make a judgment about
their usefulness.
Experimental Uncertainties

• No measurement is perfectly
accurate or exact. Many
instrumental, physical and human
limitations cause measurements to
deviate from the "true" values of the
quantities being measured. These
deviations are called "experimental
uncertainties," but more commonly
the shorter word "error" is used.
 Terms and Definitions

• Replicates - two or more

determinations on the same sample
Example: One student measures Fe (III)
concentrations six times.
• The results are listed below:
– 19.4, 19.5, 19. 6, 19.8, 20.1, 20.3 ppm
(parts per million)
• 6 replicates = 6 measurements
 Terms and Definitions
• Mean: average or arithmetic mean

• Median: the middle value of replicate

data
• If an odd number of replicates, the
middle value of replicate data
• If an even number of replicates, the
middle two values are averaged to
obtain the median
Calculation: Mean and Median
• Example I: measurements of Fe (III) concentrations:
• 19.4, 19.5, 19. 6, 19.8, 20.1, 20.3 ppm (parts per
million)
• What are the mean and median of these measurements
• Mean = 19.4 + 19.5 + 19.6 + 19.8 + 20.1 + 20.3
6
= 19.78 ppm = 19.8 ppm
• 6 replicates
• An even number of replicates !!!
• Median = 19.6 + 19.8
2
= 19.7 ppm
Calculation: Mean and Median
• Example II: measurements of Fe (III)
concentrations: 19.4, 19.5, 19. 6, 19.8,
20.1 ppm (parts per million)
• What are the mean and median of these
measurements
• Mean = 19.4 + 19.5 + 19.6 + 19.8 + 20.1
5
= 19.68 ppm = 19.7 ppm
• An odd number of replicates !!!
• Median = 19.6 ppm
Calculation: Mean and Median
• Example III: measurements of Fe (III)
concentrations:19. 6, 19.4, 20.1, 19.5, 19.8
ppm (parts per million)
• What are the mean and median of these
measurements
• Mean = 19.4 + 19.5 + 19.6 + 19.8 + 20.1
5
= 19.68 ppm = 19.7 ppm
Calculation: Mean and Median

• 5 replicates: An odd number of

replicates !!!
• Rearrange the number from the low to
the high:
• 19.4, 19.5, 19. 6, 19.8, 20.1
• Median = 19.6 ppm
 Terms and Definitions
• Precision - describes the reproducibility of
measurements.
• How close are results which have been
obtained in exactly the same way?
• The reproducibility is derived from the
deviation from the Mean:
• Deviation from the mean
– Standard deviation
– Variance
– Coefficient of variation
 Terms and Definitions

• Absolute Deviation (D) of an

element of a data set is the absolute
difference between that element and a
given point. Typically the point from
which the deviation is measured is the
value of either the median or the mean
of the data set.
 Terms and Definitions

• Average absolute deviation or

Average deviation or mean absolute
deviation is the average of the
absolute deviations and is a summary
statistic of statistical dispersion or
variability.
 Terms and Definitions

• Standard Deviation is a measure of

the dispersion of a collection of
numbers
 Standard Deviation
• measures the spread of the data about the
mean value. It is useful in comparing sets
of data which may have the same mean
but a different range.
• Example, the mean of the following two
is the same: 15, 15, 15, 14, 16 and 2, 7,
14, 22, 30
• However, the second is clearly more
spread out. If a set has a low standard
deviation, the values are not spread out
too much.
Example:
Find the standard deviation of 4, 9, 11, 12, 17, 5, 8, 12,
14
First work out the mean: 10.222
Now, subtract the mean individually from each of the
numbers in the question and square the result. This is
equivalent to the (x - xbar)² step. x refers to the
values in the question.
x 4 9 11 12 17 5 8 12 14
(x – x )² 38.7 1.49 0.60 3.16 45.9 27.3 4.94 3.16 14.3
Now add up these results (this is the 'sigma' in the
formula): 139.55
Divide by n-1. n is the number of values, so in this case
is 8: 17.44
And finally, square root this: 4.18
 Terms and Definitions

• Variance is a measure of statistical

dispersion
• Variance is the average squared
deviations about the mean. Thus,
variance is the square of the standard
deviation.
 Terms and Definitions

• Coefficient of Variation (CV) is a

normalized measure of dispersion of
a probability distribution. It is
defined as the ratio of the standard
deviation to the mean
 Problem

The following results were obtained for the

analysis of aspirin in aspirin tablets.
Determination % Aspirin w/w
• 1 91.30
• 2 91.25
• 3 91.20
• 4 91.31
• 5 91.26
 Problem
Calculate
• a) The mean of the data set
• b) The median of the data set
• c) The absolute deviation from the mean of
the 3rd determination
• d) The absolute average deviation from the
mean of the data set
• e) standard deviation
• f) variance
• g) coefficient of variation
 Terms and Definitions

• Accuracy - the closeness of the

measurement to the true or accepted
value.
• This "closeness" called as the error:
absolute or relative error of a result
to its true value.
 Accuracy & Precision

XXX X X
X

X X

Precise – Not Accurate Accurate – Not Precise

XXX
X
X
X
X
X

Not Precise – Not Accurate Accurate and Precise

 Terms and Definitions
• Outlier - Occasional error that
obviously differs significantly from
the rest of the results.
Calculation: Mean and Median
• Example I: measurements of Fe (III) concentrations:
• 19.4, 19.5, 19. 6, 19.8, 20.1, 20.3 ppm (parts per
million)
• What are the mean and median of these measurements
• Mean = 19.4 + 19.5 + 19.6 + 19.8 + 20.1 + 20.3
6
= 19.78 ppm = 19.8 ppm
• 6 replicates
• An even number of replicates !!!
• Median = 19.6 + 19.8
2
= 19.7 ppm
 Mean and True Value

__
Mean : X Xt = true value
Accuracy and Precision
 Terms and Definitions

• Precision: Described by the standard

deviation, the variance and the
coefficient of variation (all are function of
the deviation from the mean).
• Accuracy: Described by the absolute and
relative error.
 Terms and Definitions
• Absolute Error (E) - the difference between
the experimental value and the true value.
Has a sign and experimental units:
E = xi − x t
= Experimental value – true (acceptable)
value
Absolute Average Error (AAE)
= ∑ (xi − xt)
 Terms and Definitions
• Relative Error (Er) - the absolute error
corrected for the size of the measurement
or expressed as the fraction, %, or
parts-per-thousand (ppt) of the true value.
Er has a sign, but no units.
Er = xi − x t
xt
• parts per hundred (pph) = Er x100
• parts per thousand (ppt) = Er x1000
Relative Average Error (RAE)
 Calculations: Absolute and
Relative Error
• Example: measurements of Fe concentrations:
• 19.4, 19.5, 19. 6, 19.8, 20.1, 20.3 ppm
• Assumed we already knew the true value of Fe (III)
concentration at 20.0 ppm. What are absolute and
relative errors of each measurement?
• E = 19.4 - 20.0 = -0.6 ppm Er =(-0.6/20)x100% = - 3%
• E = 19.5 - 20.0 = -0.5 ppm Er =(-0.5/20) x100%= -2.5% ~ -3%
• E = 19.6 - 20.0 = -0.4 ppm Er =(-0.4/20) x100% = -2%
• E = 19.8 - 20.0 = -0.2 ppm Er =(-0.2/20) x100% = -1%
• E = 20.1 - 20.0 = 0.1 ppm Er=(0.1/20)x100% = 0.5%
• E = 20.3 - 20.0 = 0.3 ppm Er = (0.3/20)x100% =1.5% ~ 2%
 Problem
A standard, National Bureau of Standards
(NBS), sample known to contain 21.37%
arsenic was analyzed with the following
results.
Determination % Arsenic
• 1 21.40
• 2 21.35
• 3 21.32
• 4 21.39
• 5 21.30
 Problem

Calculate
• a) The mean of the data set
• b) The median of the data set
• c) The absolute error of the 2nd
determination
• d) The absolute average error of the data set
• e) The relative error of the 4th determination
in ppt
• f) The relative average error of the data set
in ppt
 Types of Errors

• Systematic or determinate errors

affect accuracy!
• Random or indeterminate errors
affect precision!
• Gross errors or blunders lead to
outlier’s and require statistical
techniques to be rejected.
 Systematic or Determinate
Errors
• 1. Instrument errors - failure to calibrate,
degradation of parts in the instrument, power
fluctuations, etc.
• 2. Method errors - errors due to no ideal
physical or chemical behavior -
completeness and speed of reaction,
interfering side reactions, sampling
problems
• 3. Personal errors - occur where
measurements require judgment, result from
prejudice, color acuity problems
 Systematic or Determinate
Errors
Potential Instrument Errors
• Variation in temperature
• Contamination of the equipment
• Power fluctuations
• Component failure
All of these can be corrected by
calibration or proper instrumentation
maintenance.
 Systematic or Determinate
Errors
Method Errors
• Slow or incomplete reactions
• Unstable species
• Nonspecific reagents
• Side reactions
These can be corrected with proper
method development.
 Systematic or Determinate
Errors
Personal Errors
• Misreading of data
• Improper calibration
• Poor technique/sample preparation
• Personal bias
• Improper calculation of results
These are blunders that can be
minimized or eliminated with proper
training and experience
The Effect of Systematic Error - normally "biased” and
often very "reproducible".
1. Constant errors - Es is of the same magnitude,
regardless of the size of the measurement.
• This error can be minimized when larger samples
are used. In other words, the relative error decreases
with increasing amount of analyte.
• Er = (Es/Xt )x100%
• eg. Solubility loss in gravimetric analysis
• eg. Reading a buret
2. Proportional errors - Es increases or decreases with
increasing or decreasing sample size, respectively.
In other words, the relative error remains constant.
• Typically a contaminant or interference in the
sample
 Detection of Systematic
Method Errors
• 1. Analysis of standard samples
• 2. Independent Analysis: Analysis
using a "Reference Method" or
"Reference Lab"
• 3. Blank determinations
• 4. Variation in sample size: detects
constant error only
 Random Errors
• caused by uncontrollable variables which
normally cannot be defined
• The accumulated effect causes replicate
measurements to fluctuate randomly around
the mean.
• Random errors give rise to a normal or
Gaussian curve.
• Results can be evaluated using statistics
• Usually statistical analysis assumes a
normal distribution
Normal or Gaussian Curve
• named after Carl Friedrich Gauss
• characteristic symmetric "bell shape
curve" that quickly falls off towards
plus/minus infinity
Normal or Gaussian Curve
 Gross Errors
• gross errors cause an experimental value to
be discarded.
• Lead to outlier’s and require statistical
techniques to be rejected.
• Examples of gross error are an obviously
"overrun end point“ (titration), instrument
breakdown, loss of a crucial sample, and
discovery that a "pure" reagent was
actually contaminated.
• We do NOT use data obtained when gross
error has occurred during collection.
Study Hard!