1

SEM V- (6+3 U)
STATISTICAL TREATMENT TO THE DATA

CONCEPT OF ERRORS
The reliability, reproducibility and the accuracy are the basis of analytical chemistry.
The function of the analytical chemist is to obtain a result as near to the true value as
possible by the correct application of the analytical procedure employed. But every
measurement made, no matter how symmetrically and carefully subject to some degree
of uncertainty or error. Determination of this degree of uncertainty is not easy. All that
one can aspire is to decrease the margin of uncertainty so that the result becomes
sufficiently accurate to an acceptable level.
To get consistent results the experiment can be repeated several times. This is known
as replicate analysis. Mathematical and statistical methods are employed to determine
the most correct value.

Indeterminate or Random errors –

These are the errors whose origin or causes cannot be easily located and they do not
have definite values as these errors are of a fluctuating or random type. This can never
be eliminated completely.

Determinate of systematic errors –

They can be measured and accounted for the source for the determinate errors is
always known. They can have fixed values.
By an appropriate choice of equipment, calibration of apparatus these can be minimized
to an acceptable level.

They are classified as follows :-

1) Instrumental errors
2) Methodical errors
3) Operational errors
4) Personal errors

Instrumental Error :-
These arise due to some faults or shortcomings of the instruments which are used in
the analysis.

1. The faulty construction of balances like unequal arm lengths.

2. Every instrument possesses a definite least count i.e. the minimum magnitude of
the property that can be measured. Reading of the scale introduces an
uncertainty in the measurement in the last digit.

1. The use of uncalibrated or improperly calibrated weights graduated glassware

like burettes, pipettes etc.

2. The attack of reagents upon glassware, porcelain etc. resulting in the

introduction of foreign materials.
1
 2

3. The use of reagents containing impurities.

4. Faulty or less sensitive component in electronic circuitry in case of instruments

like potentiometer, conductometers etc.

5. Every instrument is designed to measure a given property in a certain range. If

the magnitude of the measured property is beyond this range, the response may
not be correct. E.g. a normal glass electrode measures pH of the solutions
satisfactorily upto pH 10. If such a system is used to measure exact pH of a
solution with pH 12 the response of the system will not be proper which will
introduce an error.

Minimistation of errors :-

1. Control of instrumental errors is possible with proper calibration of the instrument

at regular intervals. Glassware should be calibrated for the temperature at which
the analysis is to be carried out. Analytical weights also should be calibrated
from time to time.

2. Sensitivity of the instrument decreases with time; therefore maintenance and

periodic calibration of instruments is necessary.

Methodical errors :-
They arise due to limitations of analytical methods chosen for analysis and from
incompleteness of a reaction e.g.

1. In gravimetric estimations where precipitation of a particular component is

carried out, errors may arise due to –
a) Appreciable solubility of a precipitate.
b) Incomplete precipitation reaction.
c) Incomplete separation of precipitate by filtration.
d) Incomplete washing of the precipitate which leaves the impurities.
e) Excess washing of the precipitate which dissolves the precipitate to
certain extent.
f) Co-precipitation, post-precipitation process.
g) Decomposition of the precipitate during its ignition or drying.

2. In volumetric analysis errors may arise due to excess addition of titrant .

In volumetric analysis, a change in the colour of an indicator signifies the
completion of a reaction; when the volume of titrant is added slightly in excess of
the theoretical value. Such an end point can actually never be taken as the
correct technical end point or equivalence point.

3. Incomplete reaction :- Methodical errors arise due to a reaction under study not
going to completion or some side reactions taking place. E.g. In Kjeldahl’s
method estimation of nitrogen in an organic compound is carried out. The
method is based on the oxidation of the organic compound; by digestion with hot
2
 3

conc H2SO4.If the oxidation is incomplete an error will be introduced in the

percentage estimation of nitrogen.

Minimisation of Methodical errors :-

They can be minimized of following methods –

1. Running a blank determination :-

This involves all steps performed in the actual analysis of the sample under
exactly same experimental conditions in the absence of the sample. The object
is to find out the effect of the impurities introduced through the reagents.
2. Running control determination :-
It involves the determination under identical experimental conditions upon a
quantity of a standard substance which contains the same weight of the
constituent as it contained in the unknown sample. The value obtained from the
analysis is compared with the true value.

3. Use of independent method of analysis :-

The same sample can be analysed by different methods. Thus two techniques
which are totally different can be applied to the same sample and the results can
be compared. E.g. In the determination of the strength of a hydrochloric acid
solution both by titration with a standard solution of a strong base and by
precipitation and weighing as AgCl. If the results obtained by the two radically
differ,methods are concordant.

4. Running parallel determinations :-

Replicate analysis is carried out for the sample to reduce the uncertainty. The
sample can be analysed by the same analyst simultaneously in duplicate or
triplicate. The results obtained will reflect on the precision of the method. The
sample can also be analysed by two different analyst by the same method.

Operational Errors :-
They arise due to insufficient knowledge or a total ignorance of handling the equipment
and not taking the necessary precautions in measurements. The causes may be,

1. Non-representative sampling i.e. improper sampling.

2. Mechanical loss of material by various means during different stages of analysis.

3. Lack of experience of the analyst resulting in errors in weighing, volume reading

etc.

4. Introduction of foreign material in the experimental sample due to carelessness

of the analyst.

5. Underwashing or overwashing of precipitates.

6. Loss of precipitate during filtration.

3
 4

7. Weighing a crucible when it is hot and cooling it in a desiccators with a poor

desiccant.

8. Improper selection of ignition temperature.

9. Insufficient drying or ignition.

Minimisation of operational errors :-

They can be minimized by proper knowledge of handling the equipment and by taking
various necessary precautions in measurement.

Personal errors or Human errors :-

They arise due to faulty ideas, improper technique, carelessness, ignorance and
physical limitations of an analyst.

1. Blowing the pipette or warming the bulb of the pipette by the palm of the hand to
remove the so-called ‘last drop’ of the liquid being delivered by a pipette. This
leads to a larger volume of a solution being transferred than is indicated on the
pipette. The pipette is always calibrated to deliver a particular volume indicated
on the bulb when the liquid drains freely from the pipette.

2. While noting down the weight of a sample or volume from a burette many a times
readings are not noted down correctly. Also while transferring a sample from
one container to another causes contamination or loss of sample.

3. Failure to add the required amount of a reagent at a proper time in an analysis

may be due to ignorance. E.g. in iodometric titration, addition of the starch
indicator in the beginning of the experiment introduces an error.

4. Personal errors may also arise due to physical limitations such as colour-
blindness, which may lead to incorrect measurement of end-point of a titration.

5. A very important personal error is due to prejudice or bias. In other words an

analyst has a fixed idea which is unreasoned and unfair in favour of a particular
result. E.g. in volumetric analysis if the first two titre values in a titration are both
11.8 ml. The analyst thinks he is getting correct readings. Thus the third time he
tries to obtain the end point at 11.8 ml. Being intellectually honest i.e. by having
an objective approach while recording measurements can eliminate such an
error.

Minimisation :- Taking experience and making measurements with objectivity and

precaution minimize them. Back-checking instrumental calibration and calculations will
also minimize personal errors.

==================================================================

4
 5

PRECISION & ACCURACY :-

Precision is used to describe reproducibility of results. Whereas the accuracy denotes
the closeness of the observed value in a measurement to its acceptable value.

PRECISION :- It is defined as the agreement among the various numerical values in a

set of observations which have been made under exactly identical experimental
conditions.
Precision for a set of observations in a measurement can be expressed using –
1. Range = W max – W min
2. Absolute deviation
3. Average deviation
4. Relative average deviation

ACCURACY :- It is a measure of the degree of exactness of a measurement to an

accepted standard. It denotes the closeness of a measurement of a property i.e. the
observed value to the accepted or standard value. If the two values are very near then
accuracy would be very high.

Accuracy is measured in terms of –

i) Absolute error & (ii) Relative error

Absolute error :- It is the difference between a measured value (Xi) and the true value
(Xt) with the appropriate sign. It may be positive or negative.

Absolute error = Xi - Xt

Relative error :- It is defined as a ratio of the absolute error to the true value.

Relative error = Xi – Xt
---------
Xt

This is generally expressed as percentage or ppt.

5
 6

Comparison of accuracy and precision –

PRECISION ACCURACY

1) It is the concordance observed for 1) Accuracy is the nearness of a

observations of the same property measured value to its true value.
under identical conditions.

2) It signifies the reproducibility 2) It signifies the reliability of a given

of the measurement. Measurement.

3) Precision is calculated for a 3) Accuracy for a single observation

definite number of observations can be calculated.
in a particular set.

4) It is expressed in terms of average 4) It is expressed as absolute error or

deviation. relative error.

5) Precision can be evaluated even if 5) Accuracy can be calculated only if

the true value is not known. the true value is known.

Measures of the Central Tendency and Dispersion :-

The information about the data can be communicated in the form of

Central Tendency and

Dispersion

Central Tendency indicates the central value around which the observations are found
to cluster.

Dispersion of data indicates a spread or range.

There are three main measures of the central tendency :-

1. The arithmetic mean (average)

Mean of all the observations in a set of observations.

2. The median :-
It is the value of the middle number when the observations are arranged in an
ascending or descending order (when the number of observations is odd).

Eg. For the following data the median = 5.7

4.8 4.9 5.7 6.3 6.4

6
 7

When the number of observations is even, the average of two middle values is
taken as the median.

Eg. 2.1 2.2 2.6 2.8 3.5

The median = 2.7

3. The mode :-
The observation that occurs maximum number of times in a given set is termed
as the ‘mode’.

The measures of dispersion are grouped into two categories –

Group I :- These measures provide dispersion and hence precision of a single
measurement from the arithmetic mean of the data :-

1. Absolute deviation :- (d) : It is the numerical difference without regard to sign

between the observed value and the mean or median for the same set of values.
Greater value of ‘d’ represents less precision.

d = | xi – |

2. Relative deviation :- It is the ratio of the deviation to the arithmetic mean.

RD= d /

3. Average deviation :- It is arithmetic mean of all absolute deviation irrespective

of the sign of deviations.

4. Relative average deviation :- It is the ratio of average deviation to the

arithmetic mean for the same set of values.

RAD = /

It is generally expressed as parts per hundred (pph) or parts per thousand (ppt).

Group II :-
These parameters provide the dispersion for the overall set of results.

1. Range – (W) :- It is the difference between the highest and the lowest values
present in a set of observations. Highest the range, lower would be the
precision.
W = W max – W min

7
 8

2. Standard deviation – (s) :- It gives the spread of the numerical data about its
mean.
It is defined as square root of the mean of the squares of the individual deviations
from the mean of the values.

The quantity (N – 1) is called a degree of freedom of the set of observations.

When the number of observations is very large, standard deviation is given as,

3. Variance :-
Variance = s2 , where s = standard deviation.

4. Coefficient of variation :-

Coeff. of variation = s / * 100

8
 9

Constant and proportionate errors :-

Determinate errors may be constant errors or proportionate errors depending on their
effect on the result of the final analysis.
Constant errors :-
In this type, the absolute error introduced in a measurement has a constant value for
different amounts of the substance taken for analysis i.e. the absolute error is
independent of the sample size. But the relative error changes and is found to be
inversely related to the sample size. Therefore taking as large amount of the substance
as possible can reduce this error.
e.g. consider a 10 mg weight which actually weighs 10.5 mg. A constant error of 0.5 mg
will be introduced in each and every measurement. If a precipitate is shown to have a
weight of 500 mg when a measurement is done with a faulty weight the actual weight of
the sample would have been

500.0 + 0.5 = 500.5 mg

Absolute error = Xi – Xt
= 500 – 500.5
= -.05 mg

Relative error = X1 – Xt -.05

---------- = ------- = 0.09%
Xt 500.5

Similarly for a substance which weights 50.0 mg the actual weight would have been
50.0 + 0.5 = 50.5 mg.
Absolute error = 50 – 50.5 = 0.5 mg.
& Relative error = -0.5 / 50.5 = 0.9%

Proportionate errors :-
In this case the magnitude of the absolute error of the measurement is found to be
proportional to the sample size while the relative error remains constant irrespective of
value measured.
e.g. consider determination of iron-contaminated aluminium as its oxide, Al2O3 .As
Al2O3 is precipitated the contaminated Fe also gets precipitated as Fe2O3. Thus the
Al2O3 will always contain some Fe2O3. If more precipitate of Al2O3 is formed more Fe2O3
will get precipitated. Thus absolute error will be the amount of Fe2O3 which will be
proportional to the sample weight.

Amt of Mg of Al2O3 Mg of Al2O3 Mg of Absolute Relative

sample expected obtained Fe2O3. error error
0.50 gm 100 104 4 4 4%
0.25 gm 50 52 2 2 4%
0.10 gm 20 20.8 0.8 0.8 4%
Thus the absolute error is directly proportional to the sample size while the relative error
is constant.
=================================================================