Quantitative Analysis

For each selected country, two sets of quantitative data analysis were undertaken. The first one

intends to shed light on the relationships between poverty and learning outcomes in a bi-variate

setting. Attention is also paid to the performance of children from specific groups (e.g. rural vs

urban; girls vs boys). The second one relates to multivariate models of the determinants of

learning outcomes. To highlight the role of income poverty, we additionally present estimates

of multivariate models separately for children from economically rich and poor families. For

Nigeria and Pakistan, however, trends in access related indicators (e.g. enrolment rates, years

of schooling completed) and physical indicators of quality (e.g. class size) are also reviewed.

For multivariate models, the variables of interest were selected based on the conceptual and

methodological frameworks described in Section 1. The review of the national policy documents

and the international academic literature on school effectiveness suggests that student learning

in OIC countries is not only low, there is also significant inequality in access to quality education.

The former is owing to system-wide factors while the latter arises because of advantages

enjoyed by children from high SES families. Therefore the level of student achievement is

modelled following the framework of educational production function where student

achievement is examined in relation to individual, family, school and institutional factors.

In case of Jordan and Malaysia, the selection of explanatory variables is very comparable as data

comes from PISA 2012 round. However, as explained in section 1, it is not possible to maintain

a fixed set of explanatory variables in all four country case studies for two reasons. First, the

data source varies across the four countries. In case of Jordan and Malaysia, detailed analysis is

based on publicly available international data sets such as PISA and TIMSS. This corresponds to

In case of Nigeria, analysis focuses on children who participated in Early Grade Reading

Assessment (EGRA) In case of Pakistan, analysis is based on ASER, the largest household based

survey that provides reliable estimates on the schooling status of children aged 3-16 years

residing in all rural and few urban districts of

Pakistan

. Second, the sample in case of Jordan and

Malaysia is school based and contains rich set of information on teachers and school facilities.

Such information is limited in case of Nigeria and Pakistan.

In case of Pakistan, the study makes use of the rich and extensive ASER data available from

Pakistan to not only provide some key descriptive statistics on both the access to education by

different groups and parts of the country but also the quality of education available to them. The

latter are based not just onmeasures of physical quality indicators (such as availability of toilets

or boundary walls) but more nuanced measures such as extent of multi-grade teaching (i.e. more

than one class sitting together typically because of a shortage of teachers, rooms etc.) within

classrooms and even more importantly on what children actually know as measured by learning

outcomes in literacy and numeracy. In addition to descriptive statistics, regression analysis is

undertaken to identify key drivers of educational quality as measured by individual learning

outcomes in literacy and numeracy. This is achieved using probit models that specify learning

as a binary variable (1 if a child is able to achieve a specified learning level as measured using

ASER data, 0 otherwise) whilst controlling for a rich set of independent variables. As explained

later, ASER data measures children’s’ literacy (Urdu, Pashto, Sindhi depending on region) and

numeracy capabilities. Students are coded as being at ‘beginner’ level in numeracy if they cannot

identify any three digits from 0 to 9;level ‘0–9’ if they can identify single-digit numbers;‘11–19’

if they can identify double-digit numbers; ‘subtraction’ if they can conduct Grade 2–level

subtraction and ‘division’ if they can conduct Grade 3–level division successfully. Similarly, in

the literacy test students are coded as being at ‘beginner’ level if they cannot identify any three