It is difficult for research in Mathematics Education to have any real impact on teaching practice in the classroom. Conscious of this, we have developed a research model for curricular innovation, involving the teachers responsible for its implementation, improving their training, and providing scientific information whilst maintaining the integrity of the phenomena analyzed "where they occur and exactly as they occur". The model is based on an analysis of the principal existing trends, and is tested in the development of Mathematical subjects at distinct educational levels, chief among them university courses in Engineering, where mixed research methods are used to analyse the effect that carrying out commands in a Computer Algebra System (CAS) has on the performance and attitude of students taking various Mathematics courses. Each teacher leads the study in his/her group and collaborates with the rest of the groups as an observer and assistant. Teaching protocols are employed in experimental and control groups; observation protocols, testing objectives, and individual questionnaires and interviews are also employed. The complete analysis provides data that confirms the fitness of the specific innovations applied (with a notable improvement of more than one point over 10 with [alpha] = 0, 05), which enables a direct and solidly-based change to the programmes and the development of their corresponding curriculums.