Confidence intervals, for their inferential nature, are useful when you want to estimate parameters of a population, based on the information provided by a random sample. One of the characteristics possessed by these elements is that they define the estimation error on a confidence interval - the sample size, population standard deviation (assuming that it is known), and the level of confidence - is its complete independence, in the sense that neither is conditioned by the value of the other two. However, in a study involving trainee teachers, who had studied the subject two years prior to the completion of this investigation, it was found that, when asked about the concept of accuracy in a confidence interval, some established wrong links between sample size and the level of confidence relationships, an example of this is to assume that increasing the sample size results is an increase in the level of confidence.