Estadística adimensional: del dato al gráfico estimado por inducción. al rescate de un método en crisis de fundamento teórico desde hace dos siglos

Abstract

The essay tells the author’s past experience in his statistics university courses; it treats later some elements of the deep crisis of a sector of traditional statistics theory: induction (or inference). It recovers the descending data ordering (Pareto) which is applied to build a structural model of nonnegative univariable continuous distribution functions. Using regional rural data (Nariño, Colombia, 2008), it shows basic components of Adimentional Statistics plus a non-parametric technique proposal (2009) that separates the media from the structural distribution to build the Cummulative Distribution Function (CDF) using only data plus two additional premises on extreme values. It does not need Probability Density Functions (PDF) and suggests their change –through consense- to adimensional normalized histograms in order to compare regressions built with different methods from common data. The idea might be applied to other science fields and specific statistical themes and even more, it might be used in statistics teaching and practice in order to make them simpler and more effective in the education of young periphery researchers and other nations youth -once it fulfills its due evaluation process-. The math part is summarized in graphs and tables. It concludes asserting the importance of ethics, responsible subjectivity and self-criticism in statistical handling by all participants in a humanist perpective.