SciELO - Scientific Electronic Library Online

vol.23 issue1Combining neural networks andgeostatistics for landslide hazardassessment of San Salvador metropolitan area, El SalvadorAn alternative to classical latent class models selection methods for sparse binary data: an illustration with simulated data author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links

  • Have no similar articlesSimilars in SciELO


Revista de Matemática Teoría y Aplicaciones

Print version ISSN 1409-2433


LLINAS SOLANO, Humberto; ARTETA CHARRIS, Martha  and  TILANO HERNANDEZ, Jorge. The regression logistics model in case the response variable assumes one of three levels: estimations, proof of hypothesis and model selection. Rev. Mat [online]. 2016, vol.23, n.1, pp.173-197. ISSN 1409-2433.

This approach follows the following scheme: first, the vector score and the information matrix from the logistics models and saturated multinomials with three possible response levels starting from the first and second derivative of the function of likelihood with respect to the parameters of the models; the relationship between the vector score and the information matrix; the multivariant standardization of the entry variables of each model; the respective asymptotic distributions; proof of comparisons and model selections that include the polytomic variable with three levels, logistic logistical and saturated models, logistical and submodel, logistical with null model, and logistical with the submodel of a less explanatory variable.

Keywords : logistic model; logit multinomial; vector score; Fisher´s information matrix; asymptotic distributions; hypothesis testing.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )