Calculation of average chronological values – an undeservedly neglected method of statistical data processing

In real analytical practice involving parallel determinations there are certain objective reasons (limited measurement time, available resources, etc.) that prevent performing sufficient number of measurements required for rigorous statistical data processing. Such parameters as standard deviations are particularly unreliable for small data sets (they are usually too high in comparison with the results obtained for more representative data sets). This problem can be minimized by changing the character of data processing, i.e. calculating so-called chronological averages instead of average arithmetical values. This, however, requires ranking the data not in the order as they were measured, but in the ascending order. The essence of calculating chronological averages is that the minimum and the maximum values of the initial data set are replaced with a single number, namely their arithmetical average, which is then averaged with other data. As a result, the total number of data values decreases by one but the changes of the averages would be negligible. At the same time, the standard deviations of the modified data sets become more characteristic and approaching the standard deviations of extended data sets. Replacing the arithmetical means by chronological average values leads to sufficiently smaller distortions of the initial data sets than, for example, in the case of excluding outliers by using the well-known “3s” criterion.