Indonesian Higher Education 117
be more likely to attend public secondary schools. Furthermore, according
to Vella (1998), when the relationship between school choice and higher
education participation is purely through the observable variables,
researchers can control by including the appropriate conditioning variables
in the equation. However, if the unobservable characteristics affecting the
school choice are correlated with unobservable characteristics affecting
higher education participation, the failure to include an estimate of the
unobservable characteristics will lead to sample selection bias.
The next step is the estimation of the logit model to explain an
individual’s choice with regard to higher education attendance. This logit
model essentially estimates the probability that individual i will attend
higher education or Hi = 1 and it may be written as
(4) Pr(Hi = 1|Zi) = h((Eij − Cij),Zij)
= h(X^b)
where Zij are independent variables that include personal characteristics
and family background, academic abilities, socio-economic status (SES),
school quality, labour market incentives and regional variables. Personal
characteristics and family background indicators include sex, religion,
whether living in urban or rural areas when 12 years old, marital status
and number of siblings under 16 years old. Students with more siblings
aged less than 16 years are also expected to have a lower chance of
attending higher education. Sex and religion are included in the model
to control for sex and religious discrimination in higher education
attendance. Moreover, it is hypothesized that residents of urban areas
have a higher probability of access to higher education. The academic
abilities variables include whether the individual has failed or passed
a class in primary education and their standardized final exam score
in lower and upper secondary education. SES indicators are parents’
educational background and whether the student is working when
studying in lower and upper secondary school. People who work while
attending upper secondary are presumed also to come from a low socio-
economic background as they have to contribute to family expenses.
Hence, they have a lower chance of higher education attendance. The
proxies for school quality consist of the number of classmates and the
number of hours of activities in schools.