19
studied using OLS regression model while the determinants of zero scores in subtasks are
studied using Probit model. The analysis is primarily based on student performance in Hausa
since English assessment was only carried out on government school children.
In sum, while the analysis in this report defines quality primarily in terms of learning outcomes,
it is not possible to comparable all countries in all domains of learning. Some measures of
student learning focus on grade-specific sample (TIMSS and PIRLS) while others sample
students based on their age (PISA). These international assessments are sample-based and only
reflect quality based on children who participate in the assessment exercise. Moreover, not all
countries participate in these surveys, creating a missing data problem. While most non-
participating countries have national assessment system, the data is neither released in public
domain nor comparable to student performance from other countries. Analysis of learning
outcomes in this study therefore primarily relies on TIMSS and PISA. This provides measures of
education quality in terms of student performance in math, science and language. In spite of the
sample-based nature of the assessments, they offer important insights into the relative capacity
of participating countries to transmit basic cognitive skills to students. Lastly, only a handful of
OIC member states has conducted early (primary) grade evaluation of student learning. The
number of OIC countries which participated in internationally coordinated assessment of
primary school children is very small (2 in SACMEQ and 8 in EGRA). Therefore the analysis of
trends in learning outcomes is primarily based on children enrolled in upper-primary and/or
secondary grades.