STATISTICS FOR ANALYTICAL CHEMISTRY

Vvian Yomaira Achig Merino,1 Jordy Mateo Cusangua Vargas,1 and Carla Lizbeth Parra Altamirano1
1
Instrumental Chemical Analysis II , School of Chemical Sciences and Engineering, Yachay Tech University
(Dated: 07/06/2022)

A. standard sample of pooled human blood serum contains 42.0 g of albumin per litre. Five laboratories
(A-E) each do six determinations (on the same day) of the albumin concentration, with the following results
(g 1−1 throughout): Comment on the bias, precision and accuracy of each of these sets of results.

Laboratories Albumin concentration Comments

A 42.5 41.6 42.1 41.9 41.1 42.2 precise, little bias, accurate
B 39.8 43.6 42.1 40.1 43.9 41.9 poor precision, little bias, little accurate
C 43.5 42.8 43.8 43.1 42.7 43.3 precise but biased to high values, poor accuracy
D 35.0 43.0 37.1 40.5 36.8 42.2 poor precision, biased to low values, poor accuracy
E 42.2 41.6 42.0 41.8 42.6 39.0 precise, little bias, accurate

Laboratory A’s results present three characteristics. First, all the results lie between 41.6-42.5, which means that
they are accurate with respect to the real value of 42.0 g. Their second feature is that they are relatively close to each
other and they are not too high for the correct result. It is observed that two types of errors have occurred. First,
there are random errors these causes replicate results to differ from one another so that the individual results fall on
both sides of the average value (42.0 g in this case). Random errors affect the precision of an experiment. In the case
of laboratory A, the random errors are small, so we say that the results are precise. In addition, there are systematic
errors, these cause all the results to be in error in the same direction. In A some results are high to the real value but
others are lower for this reason we say that the results present little bias. The data obtained from laboratory B show
poor precision due to the dispersion of the data is too big, The average of Bâs six results (41.9) is very close to the
true value, so there is evidence of little bias, the values obtained also are in the range of 39.8-43.9 what means that
they are little accurate to the real value. Related to the values obtained in C we could say that they are too high for
the real value which means they present bias, also the values are in the range between 42.7-43.8 which means that
they are precise but not accurate due to they are far from 42.0g. Laboratory D it is observed the same characteristics
as in D because they present bias due to the low values and poor accuracy related to the real value, but contrary
to C they are not precise because they are in a range of 35.0-43.0. Finally, for laboratory E we observe the same
characteristics of case A, they are precise due to their range being 39.0-42.6, also they are accurate to the real value
(42.0 g), and they present little bias.

B. Using the same sample and method as in question A, laboratory A makes six further determinations of
the albumin concentration, this time on six successive days. The values obtained are 41.5, 40.8, 43.3, 41.9,
42.2 and 41.7 g 1−1 . Comment on these results.

According to what is observed in this data we could say that they are poor precise because they are in the range
between 40.8-43.3, furthermore, they present little bias because some values are high to the real value (42.0 g), and
they are mean accurate. Comparing these values with the values of laboratory A from question 1 we could say that
it reflects reproducibility ( between-day precision) rather than repeatability (within-day precision).

C. Ten replicate analyses of the concentration of mercury in a sample of commercial gas condensate gave
the following results: 23.3 22.5 21.9 21.5 19.9 21.3 21.7 23.8 22.6 24.7 ng ml−1 .

a) Calculate the mean, standard deviation, relative standard deviation and 99% confidence limits of the mean.

• Mean
23.3 + 22.5 + 21.9 + 21.5 + 19.9 + 21.3 + 21.7 + 23.8 + 22.6 + 24.7
M ean = = 22.3ngml−1 (1)
10
Statistics for Analytical Chemistry .

• Standard deviation
rP
i = (xi − x̄)2
s= (2)
n−1

q
(1)2 +(0.2)2 +(−0.4)2 +(−0.8)2 +(−2.4)2 +(−1)2 +(−0.6)2 +(1.5)2 +(0.3)2 +(2.4)2
s= 10−1 (3)

s = 1.4ngml−1 (4)

• Relative standard deviation

s
RSD = 100 (5)
x̄

1.4
RSD = 100 = 6.2% (6)
22.3
• 99% confidence limits of the mean
s
C.I = x̄ ± tn−1 √ (7)
n

1.4
22.3 ± 3.25 √ = 22.3 ± 1.4ngml−1 (8)
10

b) Six replicate analyses on another sample gave the following values:13.8 14.0 13.2 11.9 12.0 12.1 ng ml−1 . Repeat
the calculations for these values.

• Mean
13.8 + 14.0 + 13.2 + 11.9 + 12.0 + 12.1
M ean = = 12.83ngml−1 (9)
6
• Standard deviation
rP
i = (xi − x̄)2
s= (10)
n−1

q
(13.8−12.83)2 +(14.0−12.83)2 +(13.2−12.83)2 +(11.9−12.83)2 +(12.0−12.83)2 +(12.1−12.83)2
s= 6−1 (11)

s = 0.95ngml−1 (12)

• Relative standard deviation

s
RSD = 100 (13)
x̄

0.95
RSD = 100 = 7.4% (14)
12.83
• 99% confidence limits of the mean
s
C.I = x̄ ± tn−1 √ (15)
n

0.95
12.83 ± 4.03 √ = 12.83 ± 1.6ngml−1 (16)
6

2
Statistics for Analytical Chemistry .

D. In an evaluation of a method for the determination of fluorene in seawater, a synthetic sample of

seawater was spiked with 50 ng ml−1 of fluorene. Ten replicate determinations of the fluorene concentration
in the sample had a mean of 49.5 ng ml−1 with a standard deviation of 1.5 ng ml−1

a) Calculate the 95% confidence limits of the mean.

s
C.I = x̄ ± tn−1 √ (17)
n

1.5
C.I = 49.5 ± 2.26 √ (18)
10

49.5 ± 1.1ngml−1 (19)

b) Is the spiked value of 50 ng ml−1 within the 95% confidence limits?

yes it is, because the 95% confidence interval of the mean is: 49.5 -0.93ng ml−1 < u < 49.5 +0.93ng ml−1

E. In order to evaluate a spectrophotometric method for the determination of titanium, the method was
applied to alloy samples containing different certified amounts of titanium. The results (% Ti) are shown
below. For each alloy eight replicate determinations were made.

Sample Certified value Mean Standard deviation

1 0.496 0.482 0.0257
2 0.995 1.009 0.0248
3 1.493 1.505 0.0287
4 1.990 2.002 0.0212

a) For each alloy, test whether the mean value differs significantly from the certified value
The statistic t is calculated from:

√
n
t = (x̄ − u) (20)
s

where:
x̄= sample mean
s = sample standard deviation
n = sample size.

(a) Sample 1
√
8
t = (0.482 − 0.496) = 1.54 (21)
0.0257

(b) Sample 2
√
8
t = (1.009 − 0.995) = 1.59 (22)
0.0248

(c) Sample 3

√
8
t = (1.505 − 1.493) = 1.18 (23)
0.0287

3
Statistics for Analytical Chemistry .

(d) Sample 4

√
8
t = (2.002 − 1.990) = 1.60 (24)
0.0212
The null hypothesis was rejected because the difference between u and x̄ is said to be significant at the P = 0.05
(or 5%) level. Since the observed value of |t| is less than the critical value t7 = 2.36 (P=0.05) the null hypothesis is
retained: there is no evidence of the systematic error. None of the means differs significantly from the certified value.

F. The following data give the recovery of bromide from spiked samples of vegetable matter, measured by
using a gas-liquid chromatographic method. The same amount of bromide was added to each specimen

Tomato: 777 790 759 790 770 758 764 mg\g

Cucumber: 782 773 778 765 789 797 782 mg\g

a) Test whether the recoveries from the two vegetables have variances which differ significantly.
In this case the null hypothesis is that the two methods have the same X̄1 and X̄2 result. Hence, it have proved
that Ho : µ1 = µ2 , or µ1 - µ2 = 0, so we need to test whether (X̄1 − X̄2 ) differs significantly from zero.

FIG. 1: Variance Test in R

F = 1.6958 not significant at P = 0.05. This result show insufficient evidence to reject the null hypothesis. It
conclude that the variance are different hence the have equal variance.
b) Test whether the mean recovery rates differ significantly

FIG. 2: Mean Test in R

These show that P(|t| >1.2819) = 0.2241. Since this probability is much greater than 0.05, the result is not
significant at the 5% level.

4
Statistics for Analytical Chemistry .

G. The following results show the percentage of the total available interstitial water recovered by
centrifuging samples taken at different depths in sandstone.

Depth of sample (m) Water recovered, %

7 33.3 33.3 35.7 38.1 31.0 33.3
8 43.6 45.2 47.7 45.4 43.8 46.5
16 73.2 68.7 73.6 70.9 72.5 74.5
23 72.5 70.4 65.2 66.7 66.7 69.8

• Show that the percentage of water recovered differs significantly at different depths.

Comparison of several means method was used to Show that the percentage of water recovered differs significantly
at different depths. Using R to calculate means and variances.

FIG. 3: Mean and Variance of data in R

Thus, this method call three two steps.

1. Within-sample mean square

V ariances
σ2 = n
σ 2 = 8.1
2. Sample mean square
(X̄i −x¯g )2
P
Between-sample mean square = n−1 . degrees of freedom
Between-sample mean square = 2121.9
3. F-Stadistic F = 262.
So, it could conclude that exist highly significant difference between depths. All pairs, except
deepest pair, differ significantly from each other

5
Statistics for Analytical Chemistry .

H. The following table gives the results of the determination of chromium in organic materials for two
different methods. In each case the mean is the average of five values. For each material test whether the
mean results obtained by the two methods differ significantly

2
S1
Mean 1 Mean 2 s.d 1 s.d 2 t= 2
S2
·2 Probability
Pine needles 2.15 2.45 0.26 0.26 3.45(2) = 6.9 0.01
Beech leaves: 5.12 7.27 0.80 0.44 3.31(2) = 6.6 0.01
Aquatic plant: 23.08 32.01 2.63 4.66 3.41(2) = 6.28 0.05

Thus, for pine needles and beech leaves, the method 1 and 2 means has significant difference. However to Aquatic
plant it necessary to calculate the pooled Standard Deviation.

s
(n1 − 1)S12 + (n2 − 2)S22
sp =
n1 + n2 − 2

r
4(2.63)2 + 4(4.66)2
sp = ⇒ sp = 3.784
8
So, t value was calculated.

X̄1 − X̄2 32.01 − 23.08

t= q = √ = 3.731
sp n11 + n12 3.784( 0.20 + 0.20)

At five percent level, for (n1 + n2 -2) eight degrees of freedom liberated + value is 2.306. So:
tcalculated 3,731 ≥ t is calculated 2.306
There is a Significant difference between the mean source of two.

I. Six analysts each made six determinations of the paracetamol content of the same batch of tablets. The
results are shown below:

Analist Paracetamol content

A 84.32 84.51 84.63 84.61 84.64 84.51
B 84.24 84.25 84.41 84.13 84.00 84.30
C 84.29 84.40 84.68 84.28 84.40 84.36
D 84.14 84.22 84.02 84.48 84.27 84.33
E 84.50 83.88 84.49 83.91 84.11 84.06
F 84.17 84.70 84.11 84.36 84.61 83.81

• Test whether there is any significant difference between the means obtained by the six analysts

Using Excel to calculate the One-Factor ANOVA

6
Statistics for Analytical Chemistry .

FIG. 4: One-Factor ANOVA in Excel

The hypothesis being tested is:

Null Hypothesis, Ho : µA = µB = µC = µD = µE = µF
Alternative Hypothesis, Ha : At least one means is not equal
The test statistic, F = 2.568749
The p-value is 0.0476177.
Since the p-value (0.0476) is less than the significance level (0.05), it can reject the null hypothesis.
It could conclude that least significant difference (0.25) indicates that A differs from B, D and E

J. A new flame atomic-absorption spectroscopic method of determining antimony in the atmosphere was
compared with the recommended calorimetric method [6]. For samples from an urban atmosphere the
following results were obtained:

Sample N Antimony found (mg m3 )

New Method Standard Method
1 22.2 25.0
2 19.2 19.5
3 15.7 16.6
4 20.4 21.3
5 19.6 20.7
6 15.7 16.8

• Do the results obtained by the two methods differ significantly?

7
Statistics for Analytical Chemistry .

FIG. 5: Paired t-test

The result P( t ≥ ± 3.43) = 0.02 is lest greater than 0.05. the two methods differ significantly at P = 0.05.

K. The following results were obtained when each of a series of standard silver solutions was analysed by
flame atomic-absorption spectrometry

Concentration ng ml−1 0 5 10 15 20 25 30
Absorbance 0.003 0.127 0.251 0.390 0.498 0.625 0.763

a.Determine the slope and intercept of the calibration plot, and their confidence limits

Determine the slope and intercept of the calibration plot were determined in Excel program with R program.

FIG. 6: Concentration vs Absorbance

8
Statistics for Analytical Chemistry .

FIG. 7: Values of slope and intercept of the calibration plot in R program

According the figure 6,7. We determine that the values slope and intercept are:
Intercept: 0.0021
Slope:0.0252
The equation for the regression line is thus:

y = 0.0252X + 0.0021 (25)

Also, we determine the Pearson correlation to know if the relation between Concentration and Absorbance according
to data is very close indeed to 1; and in this case we have 0.9997. On another hand, we must first calculate the
statistic Sy/x call residual standard error which estimates the random errors in the y-direction.

FIG. 8: Values of residual standard error which estimates the random errors in the y-direction in R program

According the residual standard error we have:

Sy/x = 0.007026, where is a measure used to assess how well a linear regression model fits the data
Standard error for the slope: Sb =0.00479
Standard error of intercept: Sa =0.000266

9
Statistics for Analytical Chemistry .

FIG. 9: confidence limits for a and b

FIG. 10: confidence limits for a and b

So 95% confidence limits for a are thus:

For intercept:
t=2.57

10
Statistics for Analytical Chemistry .

a=0.021 ± 0.0007
and 95% confidence limits for b are thus:
For Slope:
t=2.57
b=0.021±0.0123
a.Estimate the confidence limits for the silver concentrations in:
i.a sample giving an absorbance of 0.456 in a single determination, and

FIG. 11: Values of confidence limits for the silver concentrations in a sample giving an absorbance of 0.456 in a single
determination

According to figure 11; Y-value of 0.456 corresponds to a concentration of 18.04 ng ml−1 . Also, the Sy/x = 0.300.
Finally, the confidence limits are:
t=2.57= (0.300 x 2.57)= 0.771. Then, 18.04±0.77ngml−1 .
ii. sample giving absorbance values of 0.308, 0.314, 0.347 and 0.312 in four separate analyses.

We determine the Q-test, then we eliminate the value more disperse in relation to other values. In this case the
value absorbance of 0.347 . Since, the Q-test shows that the absorbance reading of 0.347 can be rejected as an
outlier, the mean of the remaining three readings being 0.311, Then, we determine the mean:

FIG. 12: Mean of absorbance values of 0.308, 0.314, 0.347 and 0.312 in four separate analyses

We have that the a concentration is 12.28 ng ml−1

c.Estimate the limit of detection of the silver analysis.

11
Statistics for Analytical Chemistry .

The absorbance at the limit of detection is given by:

a + 3Sy/x = 0.0021 + (3 ·0.00703) = 0.0232
Also, we have the concentration of
x = 3.3σ/m=0.84 ng/mL

L. The gold content of a concentrated seawater sample was determined by using atomic-absorption
spectrometry with the method of standard additions. The results obtained were as follows.

Gold added, ng per ml of concentrated sample 0 10 20 30 40 50 60 70

Absorbance 0.257 0.314 0.364 0.413 0.468 0.528 0.574 0.635

a. Estimate the concentration of the gold in the concentrated seawater, and determine confidence limits
for this concentration

FIG. 13: Concentration of the gold and determine confidence limits

FIG. 14: Concentration of the gold and determine confidence limits

Therefore, according Figure 13 and 14, we can determine the next values
a = 0.2569

12
Statistics for Analytical Chemistry .

b = 0.005349

Au concentration is 0.2569/0.005349 = 48.0 ng ml−1

The value of Sy/x is 0.003693, so is 0.9179.
In this case t = 2.45. Then, the 95% confidence limits for the concentration are :
48.0 ±(2.45 0.9179) = 48.0±2.2ngml−1

M. In a study of the complex formed between europium (III) ions and pyridine-2, 6-dicarboxylic acid
(DPA), the absorbance values of solutions containing different DPA:Eu concentrations were
determined, with the following results:

Absorbance 0.008 0.014 0.024 0.034 0.042 0.050 0.055 0.065

DPA:Eu 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Absorbance 0.068 0.076 0.077 0.073 0.066 0.063 0.058
DPA:Eu 1.8 2.0 2.4 2.8 3.2 3.6 4.0

c.Use these two rows of data to determine the slopes and intercepts of two separate straight lines. Estimate
their intersection point and its standard deviation, thus determining the composition of the DPA-europium
complex formed

We use Excel program, where we draw the points that give us the problem and we use a perform lineal fit to
calculate the intercept point between two lines. Therefore, we obtained the next values in the figure 15.

FIG. 15: intercept point between two lines data

Slope of the blue line is: 0.00411 and their intercept is: -0.0005 Slope of the orange line is: -0.0067 and their
intercept is: 0.0876 Then, we have that the intercept of the two lines is: (1.843, 0.081).

13