Pesquisa eleitoral: uma avaliação do desempenho dos institutos de sondagem.

Abstract

Electoral Research is a method used by research institutes to consult, by sampling, the predisposition of vote of the voters. She has been occupying a special highlight, especially in an election year, making it difficult to conceive a campaign without using at least a handful of queries so propagated "public opinion". In this context, there are errors and contradictions between the research institutes that go unnoticed by the general public. Given the above, the present study was intended to assess the performance of research institutes (IBOPE, CONSULT and 6SIGMA) in the 2008 elections for Mayor in the city of Campina Grande, PB. The criteria used to evaluate the performance of research institutes based on published results were: the winner's name error, Error of order placement of candidates and Error estimates outside margins of error considered, second degree of confidence. The realization that the institutes have met with mandatory requirements contained in article 33, the electoral law 9.504/1997, was carried out by analysing the technical specifications for each search. To verify that the institutes have calculated and disclosed correctly the sample size in your searches, formula (n) usually used for calculating the size of a sample. The research analysis as a tool of managerial support was carried out by the observation of variables sex, age, educational level and income, present in search disclosed by IBOPE and registered with the TRE under number 029/2008, where he evaluated the profile of the electorate in Campina Grande. All successful research institutes in three performance evaluation criteria and have fulfilled all the requirements laid down in article 33 of the electoral law 9.504/1997. As regards the sampling plan adopted by research institutes, it was noted that all three samples have availed themselves of size and margin of different bugs (IBOPE: n = 1.204 voters e = 3.0%; CONSULT: n = 850 voters e =3.1% and 6SIGMA: n = 4.096 voters e e = 2.0%). IBOPE and CONSULT worked with 95% confidence level, already the 6SIGMA worked with a confidence level of 99%. Replacing these data in the formula (n), only the Group 6SIGMA calculated and disclosed correctly their values. Errors in calculations submitted by IBOPE institutes and CONSULT based formula (n) did not prevent those institutes of guessing the name of the winner for the post of Mayor elections of 2008 in the city of Campina Grande/PB. However errors that cause underestimation in the sample size, as released by CONSULT is subject to variability, conveys little confidence and therefore are less valuable to strategic decisions. In addition, could have led the Institute to commit mistakes in their results. With respect to the profile of the electorate of Campina Grande which gave victory to the candidate Veneziano Vital do Rêgo, there hasn't been a segmentation very evident in the variables analyzed, but the observation of the characteristics of the society and the political guidelines thereof are of fundamental importance for the planning of campaigns.