Extraido del articulo: A Decision Analysis Approach to Multiple-Choice Examinations que hay en http://inforfamn12.uv.es/moodle/login/index.php
In Spain, candidates to postgraduate medical schools are selected in the order
established by the total score they achieve in a multiple-choice questionnaire (the MIR examination)
which consists of n = 250 questions, each of which has k = 5 possible answers,
of which one, and only one, is correct. Candidates are requested to mark one answer from
each question or to leave it blank. A point is awarded for each correct answer, zero points are
awarded for blank answers and c = 1/3 points are subtracted for each incorrect answer. If more
than one possibility is marked, the answer is considered to be incorrect. Since, in this case,
c/(1+c) = 1/4, the argument above shows that the candidate’s optimal strategy is to mark all
questions such that, p∗ ≥ 1/4, that is to mark all questions for which the probability of the more
likely answer is, at least, 1/4. In particular, this is automatically achieved if the candidate may
rule out at least one of the answers. It also follows that random guessing with no knowledge
(so that p∗ = 1/5) diminishes the candidate’s expected score.
Viene a decir que si no sabes cual es entre las 5 no la marques, pero que si dudas entre 4 3 o 2 la marques porque por probabilidad sumaras puntos