Evaluation of

Analytical Data
  Systems of Measurements

Metric System – is a system of units based on

the decimal number system

SI Units- the standard units of measurements

(meters, kilograms, seconds, Kelvin,
mole)

English System – pounds, inches, quarts,

foot, yard
 Writing Measurements in Scientific Notation

Scientific Notation – is the representation of a

number in the form A X 10n, where A is a
number with a single nonzero digit to the left of
the decimal point, and n is a whole number.
Example: Convert the following to its
proper scientific notation
a.) 210.3 liters
b.) 9,900,000,000,000,000 atoms
c.) 0.000,000,000,13 meters
 When the exponent is positive, you shift the
decimal place to the right. When the exponent
is negative, you shift the decimal place to the
left..
 Significant Figures
Variation in measurement reading is normal, and
all measurements exhibit such variations, they are
the results of limitations inherent in making any
measurements. Measuring device has its own
limitation that affects its accuracy. Yet the
uncertain digits are still significant.

 Significant figures – are those digits in a

measured numbers (or in the result of
calculations with measured numbers) that include
all certain digits plus a final one that is somewhat
uncertain. This last digit with some uncertainty in
it is called the least significant digit.
 Counting Significant Figures
1. All digits are significant except zeros at the beginning
of the number
Example : Count the number of significant figures in
each of the given numbers
 1.2345
 3.06
 0.3009
2. End zeros are significant if the number contains a
decimal point.
Example : Count the number of significant figures in
each of the given numbers
 3600.
 3.600
3. End zeros may or may not be significant if
the number has no decimal point. Either
rewrite the number in scientific notation to
clarify the number of significant figures
intended, or, if appropriate, add a decimal
point.

Example : Count the number of significant

figures in each of the given numbers
 600
 250
4. Certain values, such as those that arise
from the definition of terms are exact and
there is no uncertainty in an exact numbers.
Exact numbers have infinite number of
significant figures.

Example : 1L = 1000 mL
1 kg = 1000g
 Rounding of Numbers
 Rounding – is the procedure of dropping non-
significant digits in a calculation result and
perhaps adjusting the last remaining digit
upward.

Rules in Rounding of Numbers

 If the figure following the last number to be
retained is less than 5, all the unwanted figures
are discarded and the last number is left
unchanged.
Example: 3.6247 round of to 3 significant figure
 If the figure following the last number to be
retained is greater than 5, or is 5 with other
digits following it, the last figure is
increased by 1 and the unwanted figures
are discarded.

Example:
7.5647 round of to 4 significant figures
6.2501 round of to 2 significant figures
 If the figure following the last figure to be retained
is 5 and there are no figures or only zeros
following the 5, the 5 is discarded and the last
figure is increased by 1 if it is an odd number or
left unchanged if it is an even number. (Zero is
considered an even number)

Example : 3.250 round of to 2 significant figures

7.635 round of to 3 significant figures
8.105 round of to 3 significant figures
 Calculations with Proper Significant
Figures
 The results of an addition or subtraction
should be reported to the same number of
decimal places as that of the term with the
least number of decimal places.

Example: 191.092
+ 5.6
32.4524
 The answer to a multiplication or division is
rounded of to the same number of
significant figures as is possessed by the
least precise term used in the calculation.

Example: 183.05 X 0.25 = 46

 Fundamental Terms
Replicate determinations are often carried to
obtained best results, however they are
seldom exactly the same and they tend to
cluster about a central value. The benefit of
carrying a replicate determinations or
measurements is first, the central value is
likely to be closer to the true value than any
single value and second, the variation in
replicate values may tell us something about
the reliability of the measurements.
 Central Value
 Arithmetic Mean

Arithmetic Mean is a central value used

commonly and it is also called average value , X

X = X1 + X2 + X3 + X4 + ….+ Xn
n
X = ΣXi
n
 Median
Median is a middle value in a set of values,
this is less commonly used by an analyst.
For odd number of determination, the
median is the middle value; and for even
number of determination, median is the
average value of 2 middle value.
Example: Find the median for the sets
of values:
a.)20.4, 20.6, 20.1, 20.7, 20.0
b.)20.4, 20.6, 20.1, 20.7
Solution:
Arrange the values in increasing or decreasing
order
Median = 20.4
Median = 20.4 + 20.6 = 20.5
2

 “The mean and the median will approach

the same value when the size of the data
set increases“
 The mean is a better central value than the
median for small data sets. The main
advantage of the median is its relative
insensitivity to a single divergent value in
the data set. The median in certain instances
may be useful, particularly in dealing with
small samples.

Example: Calculate the mean and the median for

each of the following sets of data.
 Set A: 6.37, 6.33, 6.41, 6.80
 Set B: 6.37, 6.33, 6.41, 6.93
Solution:
 For Set A:
Mean = X =6.37 + 6.33 + 6.41 + 6.80 = 25.91 = 6.48
4 4

Median = 6.37 + 6.41 = 6.39

2
 For Set B:
Mean = X = 6.37+6.33+6.41+6.93 = 26.04 = 6.51
4 4
Median = 6.37 + 6.41 = 6.39
2
True value μ is the accepted value.
 % Error = Xi – μ x 100% or X – μ x 100%
μ μ
 ppt Error = Xi – μ x 1000 or X – μ x 1000
μ μ
Example:
Calculate the absolute error, percent error, and
parts per thousand error of the mean of the
following data set.
 Xi (mg) : 8.33, 8.29, 8.28, 8.34, 8.36
 Given: μ = 8.27 mg
 Solution:
 X = 8.33 + 8.29 + 8.28 + 8.34 + 8.36 = 8.32
5
 absolute error = 8.32 – 8.27 = 0.05 mg
 % error = 8.32 – 8.27 X 100% = 0.6 %
8.27
 ppt error = 8 .32 – 8.27 X 1000 = 6
8.27
 Precision
The term precision refers to the agreement among a
group of experimental results, it implies nothing about their
relation to the true value. Precision is commonly stated in
terms of the standard deviation, average deviation or
range.

A. Average Deviation
Average Deviation, d = sum of individual deviations
divided by the number of measurements.
d = Σ/ Xi – X/
n
 Average deviation is not preferable used because it is not
statistically interpretable and it gives equal weight to large
and small deviations, which are not equally probable.
B. Standard deviation
Standard deviation, s ( root mean square
deviation) is calculated by equation below:

s = Σ ( Xi – X)2
√ n-1
N-1 = for sample
N= population
Example :Quantitative analysis of student A
obtained the following results for the
determination of isooctane in gasoline.
Determination # Percent isooctane, / Xi – X / / Xi – X /2
Xi

1 3.83 0.08 0.0064

2 3.97 0.06 0.0036

3 3.94 0.03 0.0009

4 3.88 0.03 0.0009

5 3.94 0.03 0.0009

6 3.09 0.01 0.0001

X = 3.91 Σ/ Xi – X /2 = 0.0128
s = Σ ( Xi – X)2 = 0.0128 =
√ n-1 √ 6-1
 relative average deviation = d / X
 relative standard deviation = s / X
C. Range
The range R, is the absolute difference between the
largest and the smallest values in the data set. It is not a
good estimate of precision in large data sets, becoming
increasingly better in smaller sets until for two
observations it, together with the mean, describe the data
completely.

Example: Find the range for the following data set

Set A: 6.37, 6.33, 6.41, 6.80
Set B: 6.37, 6.33, 6.41, 6.93
Solution:
 For Set A:
R = 6.33 – 6.80 = 0.47
 For Set B:
R = 6.33 – 6.93 = 0.60
 Types of Error

1. Determinate Errors ( systematic Errors)

Sources :
 a.) Method
Due to undesirable behavior of samples or reagent used in analysis.

Example :
Precipitation of analyte may be weigh less if the precipitate is not
sufficiently insoluble or may be weigh more because some
substances will precipitate together with the analyte.

 b.)Equipment and Materials

Example : volumetric flask whose marking were made at 20 oC is used at

25oC
 c.) Personal judgment
 d.) Mistakes
 Types of determinate errors:
 Proportional
-magnitude varies in proportion to the
amount of analyte and therefore, the
amount of the sample ( consider absolute
error)

 Constant
- does not depend on the magnitude of the
measured quantity or amount of analyte.
 Detection, Reduction, and compensation of Error:
Example: Method Error
 Selection of method ( some are dependent on the
particular sample being analyzed)
 Control the variables or conditions( eg: maintain the acidity
of the solution)
 Evaluate the magnitude of determinate error, then
compensate rather than eliminate it. Standard samples can
be used to compensate both proportional and constant
errors.
Amount of A in unknown = result for unknown
Amount of A in known result for known
 Doing blank determination or calibration method( measured
error of the device is then added to its normal experimental
value to give correct value) are ways to evaluate magnitude
of error.
2. Indeterminate Errors ( Random Errors)
 These errors cannot be determine or controlled
 Both magnitude and direction of indeterminate errors vary
nonreproducibly from one measurement to the next and
are never the same except by chance.
 These errors are the cumulative effect of many small,
noncontrollable variables and personal judgments that lead
to uncertainty in a measured value.
 Cannot be eliminated or reduced for a given procedure but
can be evaluated to supply information about the reliability
of the data.
Example: 1. Fluctuation of room temperature and humidity
can contribute errors in Weighing
2. Manner of handling material being weighed
3. Subjectivity of the operator who reads the
weight
Distribution of Measurements: The Error Curve (
Gaussian distribution)

Figure 1: Theoretical distribution of indeterminate error

for a large number of measurements
From the figure, it shows that;
 1. An equal number and type of positive and
negative error occurs
 2. Small errors occur much more frequently than
large errors.
 3. Very large errors occur very infrequently.
 4. The value that occurs most often is the one
with no error, namely, the true value.
 The symmetry of the error curve suggests that
positive and negative errors tends to cancel one
another in the averaging process and this
cancellation becomes more perfect as the number
of measurements increases.
 Reliability of the Mean, Re
 Re ( as an estimate of the true value)
increases as the square root of the number
of measurements or k (proportionality
constant) increases.
Re = k√n
 To increase the reliability of the mean, the
number of measurements should be
increase.
Re1 = k√n1
Re2 k√n2
 Statistics Applied to Small Data Sets
Reliability of s and X can be calculated by

1. Confidence Intervals ( Confidence Level)

 How reliable or good the mean is as an estimate of
the true value, μ.
 In the absence of determinate errors, the
confidence interval about the mean with in the true
value lies
 μ = X + ts/√n where t- statistical constant( depends on
the confidence level and number of measurements, see
table 1)
Table 2. Values of t for calculating Confidence Intervals
# Degrees of Confidence Confidence Confidence
determinations freedom Level , 90% Level , 95% Level , 99%
n ( n-1)

2 1 6.314 12.706 63.657

3 2 2.920 4.303 9.925
4 3 2.353 3.182 5.841
5 4 2.132 2.776 4.604
6 5 2.015 2.571 4.032
7 6 1.943 2.447 3.707
8 7 1.895 2.365 3.499
. . . . .
Example: Calculate the confidence intervals at 95% Confidence
level in the determination of % Calcium in limestone.

Determination, n % calcium ( Xi – X)2

1 14.35 0.0004

2 14.41 0.0016

3 14.40 0.0009

4 14.32 0.0005

5 14.37 0

n=5 X = 14.37 Σ(Xi – X)2= 5.4X10-3

Solution:
Standard deviation, s:
 s = √Σ ( Xi – X)2 = √ 5.4 X10 -3 = 3.7 X 10 -2
√ n-1 √ 5-1
 μ = X + ts/√n = 14.37 + 2.776 (3.7 X 10 -2)/√5

 μ = 14.37 ± 0.045 or ( 14.325 – 14.415)

2. Comparing Values: Test of significance
( hypothesis testing)
In testing significance, a null hypothesis should be
formulated and need to be prove or disprove

A. Comparing a Mean (X) with a true value ( μ)

Null hypothesis: There is no significant
difference between the mean and the true value.

 ( X – μ) > ts/√n then significant difference EXIST!

 Example: The value of zinc(ppm) in vegetables for 4
determinations is given. Is there a significant difference
between the mean of the results and the certified true
value? Given, μ = 0.082 ppm Zn
# of Zn in vegetables ( Xi –X) (Xi-X)2
determination (ppm)
n
1 0.083 3.0 X 10 -3 9.0 X 10 -6

2 0.088 2.0 X 10 -3 4.0 X 10 -6

3 0.087 1 X 10 -3 1.0 X 10 -6

4 0.086 0 0

n= 4 X = 0.086 Σ( Xi –X)= 6.0 X 10 -3 Σ(Xi-X)2

ppm =1.0X10 -5
 Solution:
Based on the equation:
 ( X – μ) > ts/ √n
s = √ Σ ( Xi – X)2 = √1.0 X 10 -5 = 2.2 X 10 -3
√ n-1 √ 4-1
 0.086 – 0.082 > 3.182( 2.2 X 10 -3)/√4
 4.0 X 10 -3 > 3.5002 X 10 -3

therefore, hypothesis is rejected.

 SIGNIFICANT DIFFERENCE does EXIST

 B. Comparing two Means
Null hypothesis: There is no significant difference between the
two mean of 2 different data set.
 μ = X + ts /√n
1 1 1 1
 μ = X + ts /√n
2 2 2 2
 If we assume μ = μ and s = s , then
1 2 1 2
 X – X = + tsp √n + n
1 2 1 2

√ n1 n2
 where Sp is called pooled standard deviation

 Sp = √( n1 -1) S12 + ( n2 – 1) S22

 √n1 + n2 -2
 If ( X – X ) > + tsp √n + n
1 2 1 2

√n1 n2
then SIGNIFICANT DIFFERENCE does EXIST! and null
hypothesis is rejected.
 Example : A ship of copper ore from Chile was purchased by a
local metal refiner. The analysis certificate made out while the
ship was being loaded, showed that % Cu = 14.66 with the
standard deviation of 0.07% for 5 measurements. When the ore
arrived at the refinery, it was analyzed with the following results.

# of determination % Cu, Xi ( Xi-X)2

1 14.58 2.5 X 10 -3
2 14.61 4.0 X 10 -3
3 14.69 3.6 X 10 -3
4 14.64 1.0 X 10 -4
n=4 X = 14.63 Σ( Xi-X)2 = 6.6 X 10 -3
 Solution:
s = Σ ( Xi – X)2 = 6.6 X 10 -3 = 0.047% = 0.05%
√ n-1 √ 4-1
From the supplier: X1 = 14.66% s1 = 0.07% n=5
From your analysis: X2 = 14.63% s2 = 0.05% n=4

Sp = √( n1 -1) S12 + ( n2 – 1) S22

√n1 + n2 -2

=√( 5 -1) (0.07)2 + ( 4 – 1) (0.05)2

√5 + 4 -2
Sp = 0.062

 At 95% confidence limit, t = 2.365, degrees of freedom is 7

 From the formula:
If ( X1 – X2) > +tsp √n1 + n2
√ n1 n2
then hypothesis is rejected
But,
( 14.63 – 14.66) < 2.365 ( 0.062) √9/20

0.03 < 0.098

Therefore hypothesis is accepted that
the
refiner should accept the shipment of ore.
 C. Comparing Two Precisions: F - test
To check the significant difference between
2 standard deviation of 2 data set. The test
is made by comparing the 2 variance. The
square of standard deviation is called
variance. The large variance is always the
numerator.
Fc = S12 = V1
S 22 V2
If the calculated value, Fc exceeds
tabulated statistical value, Ft, the
hypothesis of no difference is rejected.
Table 3. Values of Ft for Comparing Variances at the 95% Confidence
Level

Number of observations in numerator

Number of 3 4 5 6 7 10 ∞
determinatio
n in
denominator
3 19.00 19.16 19.25 19.30 19.33 19.38 19.50
4 9.55 9.28 9.12 9.01 8.94 8.81 8.53
5 6.94 6.59 6.39 6.26 6.16 6.00 5.63
6 5.79 5.41 5.19 5.05 4.95 4.78 4.36
7 5.14 4.76 4.53 4.39 4.28 4.10 3.67
10 4.26 3.86 3.63 3.48 3.37 3.18 2.71
∞ 2.99 2.60 2.37 2.21 2.09 1.88 1.00
Example: The director of a hospital clinical laboratory was trying to
decide whether or not to keep a young, recently hired technician. The
director wants to see if the new technician’s work was of the same
quality as that of the other staff. So, she asked both a senior
technician and the new technician to analyzed the same sample using
the same procedure. Below are the results.

Senior Technician New Technician

1.38% 1.28%
1.33% 1.36%
1.34% 1.35%
1.35% 1.40%
1.30% 1.31%
 F- test is used to determine if there is a
significant difference in the precision of the
data.
Ssenior = √Σ( Xi-X)2 = 0.029%
√ n-1
Snew = 0.046 %
Fc = ( 0.046)2 = 2.5
( 0.029)2
From the table Ft = 6.39. Since Fc< Ft,
therefore there’s no significant difference.
 Rejecting Data: Q-test
Occasionally in a set of results we find that
one value is much larger or smaller than the
others, and we decide whether to retain or
reject the value. Q-test is used to reject or
retain the data.

 Qc =/ questionable value – nearest value/

Range
If Qc < Qt(from the table) then questionable
data will be retain.
Table 4. Values of Qt for Rejecting Data
Number of Degrees of Confidence Level
observation, n freedom
90% 96% 99%

3 2 0.94 0.98 0.99

4 3 0.76 0.85 0.93
5 4 0.64 0.73 0.82
6 5 0.56 0.64 0.74
7 6 0.51 0.59 0.68
8 7 0.47 0.54 0.63
9 8 0.44 0.51 0.60
10 9 0.41 0.48 0.57
Example: In the determination of Chloride in water sample,
the following data were obtained. Based on the data
taken, determine if the questionable value needs to be
rejected or retained.
Trial # % Cl-
1 12.69
2 12.58
3 13.02
4 12.63
Solution:
Qc = / 13.02 – 12.69/ = 0.75
/13.02- 12.58/

from the table Qt = 0.76, and Qc< Qt therefore data will

be retain
 Method of Least Squares
Experimental points seldom fall exactly in a
straight line because of indeterminate errors
in the instrument readings. In many cases a
straight line relationship between the x and y
axis is needed to get the desired data.
Method of least squares will be use to make
the line straight.
 Assuming the relationship is linear, then the
equation of the line is,
Y = mX+ b
The value of m and b will be substituted in
the equation of the line, and new values of Y
will be calculated using the values of X. The
new values of Y will be graph versus the
value of X. A straight line will be obtained,
thus desired data can be determined.
4.The concentration of glucose in aqueous solution can be determined by
oxidation with ferricyanide ion, Fe(CN)6-3. The yellow ferricyanide ion is
reduced by glucose to colorless ferrocyanide ion, Fe(CN)6-4. The
absorbance of the ferricyanide ion, measured at 420 nm, decreases as the
concentration of glucose increases. The decrease in absorbance is
proportional to the concentration of glucose.

Using the least squares calculate

a. the equation of the best straight line through these points
b. the 90% confidence interval of the intercept
c. the 90% confidence interval of the slope
d.The solution of unknown concentration gave an absorbance of 0.422,
calculate the concentration of the unknown
 Xi Xi- (Xi-) 2 Yi Yi- (Yi-)2 (Xi-)(Yi-)
0.005 0.016 0.00026 0.75 0.2876 0.08271 4.60E-03
0.010 0.011 0.00012 0.642 0.1796 0.03226 1.98E-03
0.020 0.001 1E-06 0.498 0.0356 0.00127 3.56E-05
0.030 0.009 8.1E-05 0.29 0.1724 0.02972 1.55E-03
0.040 0.019 0.00036 0.132 0.3304 0.10916 6.28E-03

2 2
S xi = 0.105 S( Xi-) = 0.056 S (Xi-) = 8.20E-04 S Yi= 2.312 S Yi-= 1.0056 S (Yi-) = 0.25512 S (Xi-)(Yi-)= 1.44E-02
 0.021 = 0.4624