Technical data

252 Understanding Your Agilent ChemStation
11 Evaluating System Suitability
Definitions for Reproducibility
Confidence Interval CI
The confidence interval is calculated to give information on how good the
estimation of a mean value is, when applying it to the whole population and
not only to a sample.
The 100 × (1 - α) % confidence interval for the overall mean is given by
where:
percentage point of the t distribution table at a risk probability of α)
For the extended statistics in the sequence summary report the 95%
confidence interval may be used (α = 0.05).
The t distribution (or ‘student distribution’) must be used for small sample
volumes. In case of large sample volumes the results for the t distribution and
the normal (gaussian) distribution do not differ any more. Therefore in case of
30 or more samples the normal distribution can be used instead (it would be
very difficult to calculate the t-distribution for large numbers, the normal
distribution is the best approximation of it).
95% Confidence Interval for 6 samples:
1 - α = 0.95
N = 6
The correct value for t has to be taken from the t distribution table for 5 (N-1)
degrees of freedom and for the value α/2, being 0.025. This would give the
following calculation formula for CI: