Correa, André Luiz [UNESP]Santos, Lucas Mikael da Silva dos2022-01-042022-01-042021-10-29http://hdl.handle.net/11449/215686Esta dissertação teve como pretensão investigar de forma crítica todos os pressupostos da econometria referente ao valor de probabilidade (p-valor) e questionar se a sua utilização deve ser vista como uma metodologia absoluta dentro dos modelos estatísticos em Economia. Para tanto, analisou-se de maneira detalhada toda a literatura acerca do debate sobre a inferência estatística: em especial, sobre as estimações a partir da estatística frequentista e bayesiana, ao mesmo tempo, que se desenvolveu um paralelo com a teoria econométrica. A pesquisa inicialmente detalha a discussão Fisher x Neyman-Pearson, fato que será importante para entender os principais problemas do p-valor e como todos esses obstáculos metodológicos foram transportados para a econometria. Após isso, é apresentado de maneira resumida todo o desenvolvimento da teoria econometria e como o p-valor se tornou uma pedra angular para a metodologia tradicional da mesma. Por fim, da mesma forma, também é destacado de maneira sucinta os principais fundamentos e dilemas da teoria bayesiana. Conclui-se que apesar da estatística bayesiana também ter diversos pontos a serem questionados referente a sua utilização, não existe uma justificava plausível para o domínio metodológico da inferência frequentista - fato que é ilustrado pelo uso do p-valor - dentro do mainstream econométrico.This dissertation aimed at critically investigating all econometric assumptions regarding the probability value (p-value) and questioning whether its use should be seen as an absolute methodology within statistical models in economics. To this end, a detailed analysis was made of all the literature on the debate about statistical inference: in particular, about estimations from frequentist and Bayesian statistics, and, at the same time, a parallel with econometric theory was developed. The research initially details the Fisher x Neyman-Pearson discussion, a fact that will be important to understand the main problems of the p-value and how all these methodological obstacles were transported to econometrics. After that, the whole development of econometric theory is briefly presented and how the p-value became a cornerstone for the traditional econometric methodology. Finally, the main foundations and dilemmas of Bayesian theory are also briefly highlighted. The conclusion is that although Bayesian statistics also has several points to be questioned regarding its use, there is no plausible justification for the methodological dominance of frequentist inference - a fact that is illustrated by the use of the p-value - within the econometric mainstream.porTeoria econométricaInferência bayesianaInferência frequentistaEconometria bayesianaP-valorBayesian inferenceFrequentist inferenceBayesian econometricsP-valueEconometric theoryUma análise crítica sobre os fundamentos da econometria clássica: valor de probabilidadeA critical analysis on the fundamentals of econometrics classic: probability valueDissertação de mestradoAcesso aberto33004030080P0