Decrease in prevalence of alcoholism in Siberia: tendency or predictive artifact?

The time series of patients with alcoholism registered by narcological institutions (morbidity) and their correlation with patients newly diagnosed as having alcohol dependence were considered. The development included only those administrative-territorial formations of Siberia, where this correlation is weak. The technique of identification of the main tendency of development of the time series (trend) was carried out by alignment of the time series. Alignment allows characterizing the feature of change of this time series over time in the most general view as function of time, assuming that further it will be possible to express influence of all major factors. For analytical alignment of time series of alcohol dependence across each territory various forms of trends expressing those or other qualitative properties of development of tendencies were used. Identification of the main pattern of change of levels of a series assumes its quantitative expression, to a certain extent free from random influences in the form of coefficient of determination that allows to characterize feature of change of this time series over time in the most general view as function of time, assuming that further it will be possible to express influence of all major factors. The coefficient of determination characterizes the part of variability (dispersion) of the effective sign y explained by regression in the general variability (dispersion). The coefficient of determination is calculated for assessment of quality of selection of the equation of regression. For the acceptable models it is supposed that the coefficient of determination has to be at least not less than 50 %. The coefficient of determination characterizes the part of variability (dispersion) of effective sign y explained by regression in the general variability (dispersion) of y. The coefficient of determination is calculated for assessment of quality of selection of the equation of regression on which basis predictive estimates were provided. Linear, polynomial, exponential, logarithmic forms of trends and a trend in the form of the exponential curve were used for analytical alignment of time series across each territory.