Comparaciones de perfiles de salud de individuos sanos en diferentes sectores económicos

Abstract

The present study shows that tests based on the rank of observations are a more flexible alternative to parametric models when the normality assumption is not fullfilled. The objective of these tests is to detect differences between two or more populations, either in location or in scale, with the characteristic that those affected are almost as efficient as their parametric counterparts in terms of Asymptotic Relative Efficiency when it is applicable. After an appropriate data cleaning process, the Principal Component Analysis (PCA) and Benefit of the Doubt Approach (BOD) were considered as tools for building individuals health indicators for the individuals involved in the economic activities. The projections in the new axes for the most numerous groups are the used variables for the inference process divided into the interest groups.