test score ranges potentially from -36 to + 36, as it results from summing
up the outcome of 24 independent true/false questions, with 1.5 marks for
correct answers, -1.5 marks for wrong answers, and 0 for no answer, so that
random guessing implies a zero expected score. Test scores were then rescaled
to the range (-100, +100), to make them more easily interpretable and com-
parable with the results reported in the literature. In our sample, the average
rescaled test score is 58.7, with a range between -41.7 and 100.8 The figures
on lecture (LEC) and class (CLA) attendance are estimated percentages, as
reported by the students, out of a variable number of two-hour lectures and
classes for each of the three-week course units.9 On average, the students in
the sample attended about two-thirds of classes (67.4) and a slightly higher
percentage of lectures (70.8).10
Ability is proxied by four main indicators, based on students’ past per-
formance in both high school and university: high school grade (HSG), rang-
ing between 60 and 100, is the leaving certificate score; grade point average
(GPA) is the average mark on exams passed before taking Introductory Mi-
croeconomics;11 exams per annum (EPA), is a measure of speed in completing
course work, defined as the number of exams passed divided by the years of
registration; calculus is a dummy variable equal to 1 if the student has passed
the first-year calculus exam.12 The means of both GPA and EPA appear to
be quite low (76.9 and 2.1, respectively). Effort is measured by the average
number of hours of study per week (in addition to attendance) for Introduc-
tory Microeconomics (SSH), ranging between 1 and 35 around a mean value
of 10.9. Motivation is measured by two indicators: subject evaluation (SEV)
and teacher evaluation (TEV), self-reported assessments defined on a 0 to
8 The pass mark for the rescaled score is 50, given that in the Italian university system
the exam pass score is 18.
9 There were generally 8 lectures and 3 classes for each course unit.
10The figures for lectures are very similar to those reported by Romer (1993) and Rodgers
(2001), while substantially higher than those reported by Kirby and McElroy (2003). The
higher attendance rate for classes relative to lectures is the opposite of what was found by
both Rodgers (2001) and Kirby and McElroy (2003).
11The actual GPA, defined on a 18-30 scale, was rescaled to a 60-100 scale to make it
more easily comparable.
12 The interpretation of the last two variables might be considered difficult for first-year
students, who represent the majority of the sample. However, they can be considered
informative as they provide information on performance in the first semester for first-year
students, and overall performance for the remaing group of students in higher year of
registration.