The purpose of the present study is to raise questions regarding the divergences among evaluators in respect to the grades attributed in mathematics exams. These interrogations are relevant since, in the university entrance exams, many important Brazilian institutions adopt essay questions. When the exam is based on multiple choice, the correction is automatized and the error, if it occurs, can be easily spotted. Also, there is an exact comprehension of the manner in which it affects all those who are taking the exam. This is not the case in essay question exams, in which the correction is not automatized and suffers significant influence of those who correct them. This study will show that certain elements present in the students' answers in the exam can bring about strong divergence among the evaluators. These divergences are an alert to the entrance exam committees, who must pursue some form of grade standardization for these exams.
Docimology; multi-correction of exams; grading scale; grade standardization
