132 Mohamad Fahmi
Parents education background (FATHSHHE and MOTHSHHE) have
similar effect in all groups as the coefficients are positive and significant.
Among the school group, students from private secular school have the
most advantage when their mothers attend upper secondary education
or higher education. On the other hand, father’s education background
(FATHSHHE) affects the probability of higher education access differently.
Except in private secular group, father’s education in upper secondary
and higher education increases the probability of student to attend post-
secondary education. The influence of father’s education is strongest to
students from private Christian school.
Male students in all type of school have higher probability to continue
to higher education than female students. Male students in private Christian
and public schools have more chance to attend post-secondary education
than students from private Islam and private secular ones. Moreover,
number of siblings (SIBLING) also significantly affects the higher education
participation. The coefficients on SIBLING are all negative and significant
in all type of school indicating that the more siblings under 16 year old
living in a household the less probability of a student in that household
continuing to higher education.
There are mix signals on coefficients teacher-student ratio (TSRATIO),
score of final exam (ZNEMSMA), and failed in upper secondary school
(SMAFAIL). TSRATIO is a variable that provides information about
school quality. ZNEMSMA and SMAFAIL are variables that inform about
student abilities. In public schools, coefficient of ZNEMSMA is positive,
whereas coefficients of ZNEMSMA and TSRATIO are negative. Intuitively,
these results inform that the factors of ability are more important than
school quality in higher education participation. On the contrary, positive
coefficients on ZNEMSMA, TSRATIO and negative coefficient on SMAFAIL
in private secular and private Islam schools inform that high ability and
good school quality are important factor for student in those private schools
to access higher education.
5.3 The Non-Linear Decomposition of the Logit of
Higher Education Participation
In this subsection, I provide the result of non-linear decomposition
estimation to identify the causes of the choice of school type differences on
higher education participation. The summary statistics in Table 5.5 indicates