[Use of various accordance measures in the study of sick leave frequency distribution per man-year]

In the studies of a wide number of phenomena very often an attempt to describe their distributions with the help of certain theoretical distributions is undertaken. Pearson's chi2 and Kolmogorovs lambda traditional tests, when applied to the evaluation of accordance between the empirical distribution and the distribution calculated on the basis of theoretical probability function, depend directly on the dimensions of population studied, and they are not useful in the case of big quantities. In the paper the author presents a great number of measures for two (empirical and theoretical) distributions accordance and their application to the study of sick-leave frequency distribution per man per year. On the basis of received results a high level of distributions accordance between the sick leave cases distribution and the negative binomial distribution has been observed. The distribution of sick leave shows a high accordance with the logarithmically-normal distribution. In the light of the above mentioned distribution study, the imperfection of Pearson's chi2 and Kolmogorov's lambda tests can be ascertained. In the case of big quantities the evaluation of distributions accordance seems to require the application of the accordance measures based on relative quantities.