[Reorder the disorder: disturbance and standardization variables].

"Spatial variations in crude death rates by cause-of-death may result among others from differences in the age structures of the populations of the regions and/or from the small numbers of deaths at certain ages or for certain cause-of-death. This paper addresses, on the one hand, the issue of controlling the impact of age structures (or of confounding factors, generally speaking) and, on the other hand, that of the chance variations in the results of the various standardization methods. Standardization is considered here from the viewpoint of causal modelling: controlling for age structures is a special case of controlling for confounding variables when using aggregate data. From this point of view, we have evaluated and compared the different standardization techniques and their ability to control for confounding factors. The construction of significance tests and confidence intervals has been pursued in order to evaluate the methods from the point of view of the statistical robustness of the standardized indices." (SUMMARY IN ENG AND ITA)