The learning at school level is evaluated periodically and if required, the student must repeat it. It is assumed that initially no one meets the evaluation, but as you acquire knowledge you must pass the evaluation, otherwise the exam should be repeated. Thus, the subsequent results are not independent. For this purpose, the learning model presented by Kalbfleisch in his book of probability and statistical inference was applied. This enabled that two groups were compared and exposed under different conditions in which the learning (A) and need for improvement (B) parameters were analyzed, by maximum likelihood estimation. For the data processing, we used the statistical software R. The obtained results determinate that there are effects of the learning processes: A < B and A < 1; In particular group B shows better effect due to the fact that A is smaller and the likelihood contours, which was 15% for A, show a significant difference between groups.

Keywords: Learning evaluation, maximum likelihood estimation, statistical inference.