UMA PROPOSTA DE AMOSTRAGEM ESTRATIFICADA PARA PESQUISA DE ORIGEM E DESTINO (O/D)

The statistical sampling to affirm a null hypothesis in the degree of confidence should follow precise and correct methodological criteria and processes, because in the occurrence of estimation errors, besides being controlled, it may increase errors in the verification of results. The research aims to evaluate the sampling collected at km 193 of BR393 in the municipality of Paraíba do Sul, about 154 km from the capital of Rio de Janeiro. The purpose was to verify whether the sampling of the 4 different vehicle groups among 29 different vehicle types had sufficient sample to affirm, whether research of origin and destination O/D had statistical significance in the degrees of confidence of the research. The result presents three scenarios to simulate the best method. The stratified sampling was the most indicated, as it allows the grouping of homogeneous data in heterogeneous population data. The research data presented problems of underestimation in the number of people interviewed and in the execution time, so much in the applied method and in the simulation of the Arena software.