100 scale (average values of 74.7 per cent and 80.9 per cent, respectively).
Additional quantitative variables include travel time to reach university
(in minutes), age, and year of registration. A number of dummy variables
provide information on student characteristics, such as gender (1=female),
foreign language (1=non-native speaker), work (1=worked while taking the
course), web (1=internet available at home), and live away from home. Fur-
ther information on the background of students is provided by cathegorical
variables referring to high school type, education and occupation for both
father and mother, and province of residence.13
4 Methodology
We are interested in estimating the parameters characterizing the relationship
between teaching and learning. We assume that learning is the output of an
educational production function that reflects the match between two types of
factors: academic input and student input.14 Academic input broadly refers
to teaching (lectures, classes, seminars, tutorials, office hours, etc.). Student
input is assumed to reflect a number of individual factors, among which the
three main ones are ability, effort, and motivation. Assuming linearity, the
relationship can be described as
yi = β1x1i + β2x2i + εi (1)
where yi is learning for individual i, x1 is academic input, x2 is student input,
and εi is an error term reflecting all other factors that affect learning, with
i = 1,...,N.
We measure learning by academic performance (test score) and teaching
by lecture and class attendance. It is more difficult to find an appropriate
measure for student input, given that factors such as ability, effort and mo-
tivation are not directly observable. This would not be a problem for the
estimation of β1 if student input and attendance were uncorrelated. How-
ever, ability, effort and motivation are all likely to be positively correlated
13Tables 6-10 in the data appendix provide descriptive statistics for quantitative indi-
cators of performance, attendance, ability, effort and motivation, by sub-groups defined
according to students’ characteristics.
14See e.g. Pritchett and Filmer (1999), Lazear (2001), Todd and Wolpin (2003) and
Coates (2003).