Applying statistics to social-type research is generally carried out by a staff who is not specialized on Mathematics, therefore mistakes are found in the methodology used due to the lack of command of the subject. Such an experience has allowed to recognize which are the most frequent mistakes in this knowledge area, specially the pedagogic deficiencies. The present paper is intended to reveal these deficiencies, and to stress the contributions statistics makes to research as well as its usefulness for investigation. It also brings together a sample review work of masters and doctoral dissertations defended in CEPES (Center of Studies for the Improvement of Higher Education).