ABL90 FLEX reference manual 7. Performance characteristics
7-5
The coefficient of variation is reported as a percentage and calculated from the mean
(or measuring level) and standard deviation as follows:
CV%=
Standarddeviation
Measuringlevel
100
Confidence interval provides a range of values estimated from a study group that is
highly likely to include the true, but unknown, value ("confidence interval" applies to
the results of a statistical analysis). A 95% confidence interval means that there is only
a 5% chance that the true value is not included in the interval.
Repeated measurements using one analyzer on samples assumed to be identical will
not necessarily yield identical results. The degree of variation in the results is a
measure of the precision of the analyzer.
The table on the next page describes the parameters used to characterize precision
obtained via the performance tests on the ABL90 FLEX analyzer. [Ref. 9]
Parameter Description
S
0
Repeatability
This is a standard deviation obtained from repeated
measurements within a short interval of time using:
The same instrument and location
The same measurement procedure
Identical portions of the same sample
One operator per instrument
S
0
for each level is pooled for all test instruments and test
days.
S
x
Reproducibility is obtained from repeated measurements over
several days using:
Random instrument
Random sample
Random operators
Reproducibility for each level is pooled for all test instruments
and test days.
TE
A
, total analytical error is a quality specification that sets a limit for both the random
error (imprecision) and systematic error (inaccuracy) in a single measurement or single
test result. In Radiometer reference manual the following expression for total analytical
error is either expressed in an absolute number
TE
A
= (|Bias|+1.96×S
x
)
or in %
TE
A
= (|Bias%|+1.96×CV
x
) %
The formula we are using for total error allowable works at 95% probability to allow for
5% error.
Coefficient of
variation
(CV%)
Confidence
intervals
Repeatability/
Reproducibility
Total analytical
error