Widiastuti Setyaningsih

European Master in Quality Analytical in Laboratory

Gdansk University of Technology
Email widiastuti_setyaningsih@yahoo.com
Mobile +48889664465 or +6287878121767

QM0502–METHOD VALIDATION

Deadline : February 04, 2011

Supervision : Professor Piotr Konieczka

ASSIGNMENT NUMBER 09
Based on the series of results for the four samples (8 repetitions) calculate the method repeatability. Based on the results
determined for certified reference material sample determine the trueness value (as recovery value)

Data: results ng/g:

sample 1 sample 2 sample 3 sample 4

1 11.1 13.8 22.7 33.2
2 11.5 13.7 22.8 33.9
3 12.0 14.2 22.4 34.1
4 12.3 13.2 22.0 33.0
5 11.8 13.6 21.9 32.4
6 11.2 14.9 23.2 32.3
7 11.4 14.4 23.1 31.8
8 11.9 13.8 23.0 31.0

CRM data: ng/g:

1 12.5
2 12.8
3 12.9
4 12.3

CRM U k
12.32 0.58 2

WORKSHEET DESCRIPTION

This MS Word document as a support for descriptions of the attached separate EXCEL book for the whole work out

On the same EXCEL book, separate sheets are assigned for workouts, plots and tables of Repeatability and Trueness
solution.
THE METHOD REPEATABILITY

The repeatability of a method based on a series of replicates can be determined using the following basic steps.
 Test for a result with gross error-Q-Dixon Test
Q-Dixon test is applied with a significance level,  = 0.05, to check any outlier in the given data.

Table 1. Q-Dixon Test

Statistic Sample 1 Sample 2 Sample 3 Sample 4

Number of results - n 8 8 8 8

Range - R 1.2 1.7 1.3 3.1

Q1 0.083 0.235 0.077 0.258

Qn 0.250 0.294 0.077 0.065

Qcrit (f=7, α=0.05) 0.507

As the calculated values of Q 1 and Qn < Q crit, It indicates that there are no outliers in series of measurements, so
the repeatability can be calculated taking into account all the results.

 Test for homogeneity of variances: Fmax Hartley Test

To check the homogeneity of the variation of the given data, The F max-Hartley test is applied because of Equi-
numerous series (a significance level of  =0.05 was chosen) with hypothesis :
Ho: The samples are taken from the same population (Homogeneous)
H1: The samples are heterogeneous since any of sample are not taken from the same population

Table 2. Fmax Hartley Test

Statistic Sample 1 Sample 2 Sample 3 Sample 4

Sum 93.2 111.6 181.1 261.7
Mean 11.7 14.0 22.6 32.7
Standard Deviation 0.417 0.529 0.493 1.048
RSD 0.036 0.038 0.022 0.032
CV (%) 3.583 3.793 2.176 3.204
CV (%) Repeatability 3.189
F max 3.038
Fmaxo (k=4, f=7, α=0.05) 8.44

As Fmax < Fmaxo so the Ho is accepted at 95% confidence level, there is no evidence the variances are
heterogeneous and we assure that there is no statistically significant difference in variation values

The calculated repeatability value can be, however, calculated as a mean value from the coefficient of variation
(CV) for four series.
  Determination of coefficient of variation for each series.
Repeatability should be calculated as a mean value of coefficient of variation (CV) values for the four series.

Table 3. Coefficient of Variation (CV)

Statistic Sample 1 Sample 2 Sample 3 Sample 4

Sum 93.2 111.6 181.1 261.7
Mean 11.7 14.0 22.6 32.7
Standard Deviation 0.417 0.529 0.493 1.048
RSD 0.036 0.038 0.022 0.032
CV (%) 3.583 3.793 2.176 3.204
CV (%) Repeatability 3.189

 Determination of the mean coefficient of variations of all series.

The determined repeatability value is equal to:
CV repeatability = (3.583+3.793+2.176+3.204):4 = 3.189 %

THE TRUENESS VALUE

We can determine the trueness of a given analytical method based on recovery values R (%) obtained for the analysis of a
given standard (or certified reference material CRM) with concentration 12.32 ng/g. A series of four independent
determinations are conducted.

Table 4. Recovery Value, R (%)

Statistic Value
Number of results - n 4
CRM 12.32
Mean (X det) 12.625
SD 0.275379
Udet 0.275379
udet 0.137689
ucrm 0.29
Ucrm 0.58
U (k=2) in % 5.15
R (%) 102.48

From the data given from the certified reference material, the detail calculation to obtain recovery value R (%) as explained
below
n

Mean X 
X
i 1
i
= 12.625 ng/g
Det
n

(X  X m )2
Standard deviation i 1
i
= 0.275379 ng/g
SD 
n 1

The expanded uncertainity for the determined values Udet will be calculated as;
Udet  2 x SD = 0.275379 ng/g
n
The recovery value % R for the analysis of the CRM
 X 
% R =  Det 100% = 102.48 %
 X CRM 
where X CRM = concentration in the CRM.
2 2
The expanded uncertainty of this recovery value Urec can be calculated using the equation; Urec  kx
uCRM  uDet
 X Det  X CRM 
 2 
 
where uCRM = standard uncertainity of the calculated mean value, calculated as
uCRM = UCRM : 2 = 0.29 ng/g
uDet = standard uncertainity of the calculated mean value, calculated as
U
uDet = Det = 0.137689 ng/g
k
Substituting all the calculated values in to equation, Urec in percentage % U rec = 5.15 %

The trueness as recovery value can be expressed as; % R = (102.5 ± 5.2) %

If the mean of the determinations X Det and the value of the certified reference materials X CRM don’t differ significantly,
the ratio will fit to 1. Hence the trueness indicated as recovery interval contains the ratio 1 (100%) as can be seen from
Figure below

Since the value of 100% is included in the range of calculated trueness value, i.e. determined recovery value ± the
uncertainty of its determination (R ± U), it can be concluded that the obtained results are correct in a statistically
significant manner.
REFERENCES

EMQAL. 2010. QM0502 - Method Validation (PDF file for Module Presentation) by Professor Piotr Konieczka

EURACHEM/CITAC. 2000. Quantifying Uncertainty in Analytical Measurement. Second edition