samlet årgang - Økonomisk Institut - Københavns Universitet

WHAT AFFECTS STUDENTS’ PERFORMANCE? 365
for our data. The first year GPA is introduced to test whether the data exhibit a »progression
in learning«, i.e. whether doing well in the first year implies good performance in
the second year.
Before moving on to the estimations, some data averages are presented. As men
tioned, 20% of the students are admitted as quota-1 students. Of the remaining 80%,
roughly 50% come from a Scandinavian country. In total, approximately 50% of the
students are Danes. 16 The average age of the quota-1 student is 20.38 years, while the
average age of the quota-2 student is 23.78 years. Finally, we note that the average
admission GPA of the quota-1 students is 10.1, while the average admission GPA for
the quota-2 student is 8.87 (Scandinavian) and 8.46 (non-Scandinavian).
3. The estimated equation
We shall now describe the equation that we estimate. Our dependent variable is the
grades in each 1 st and 2 nd year course that has a written examination. The independent
variables are the following: a dummy variable for whether the students have gained
access by quota-1 or quota-2 and a variable for the age of the students when admitted
to the program. For quota-2 students in particular we include the GPA from highschool,
the motivational average and an indicator (dummy) for whether or not the student
comes from a Scandinavian country.
For each course, we run regressions of the following type:
GRADE i = 0 + 1 DQ i + 2 GPA i *DQ i + 3 MOTAVR i *DQ i
+ 4 DSCAND i *DQ i + 5 AGEADMIT i + error i
where GRADE is the grade for a student i who has participated in the exam under study;
DQ is the dummy for admission quota, taking the value 1 if the student is admitted
via quota-2, and 0 if the student is admitted via quota-1; GPA is the high-school gradepoint
average for a quota-2 applicant (this is why it is multiplied by DQ in the model
equation); MOTAVR is the figure for the average given to the motivational essay of the
student (multiplied by DQ for same reasons as above); DSCAND is a dummy for home
country of the student (DSCAND =1 if the student comes from a Scandinavian country
and 0 otherwise). Again, this background variable is of interest for quota-2 applicants
only, and this is why it has been multiplied by DQ in the model equation; AGEADMIT
16. Some discretion may also enter in the admission process as, other things equal, country diversity matters
to the program: the IB office does attempt to put together a diversified, international class.
(1)