Recently, pre eleccciones surveys have shown, in many different cases, strong differences between their estimates and the results given in the polls. But this phenomenon of inaccuracy has harmed no single surveys as possible developed meters results, although this is not their naturaeza nor objective, but tmabién have detectadoserios problems and differences with the election results forecasting models specifically oriented. What happen? How can they be corrected, if possible, these issues accuracy of forecasts and surveys genersl taken as such in particular?