By Alia Aghajanian

Four out of ten Nigerian school-age children are not in school (see UNESCO report). Even more unsettling is how this statistic is drastically different for girl children, the poor, and the North East and North West regions of Nigeria. A study of the trends in inequality in access to schooling between 2003 and 2013 was carried out by Cora Mezger, which analyses successive rounds of the Nigerian Demographic and Health Survey datasets, for 2003, 2008 and 2013. The study finds the following:

Unequal access to education by income group

• A child from the poorest quantile[1] of the Nigerian population has a predicted probability of being out of school that is almost double that of a child from the richest quantile of the population.
• The gap in predicted probabilities of being out of school between the poorest and richest quantiles was 21 percentage points in 2003, but this gap increased to 32 percentage points in 2013.
• The average predicted probability of being out of school has increased from 2003 to 2013, rather than decreased for the poorest quantile of the population.

Unequal access to education by gender

• Female children are more likely to be out of school compared to male children.
• While a gender gap still persists, it has been closing over the past decade.

Unequal access to education by geographical region

• There is a strong North-South divide in access to education. The North East and North West regions have the highest predicted probability of being out of school. The North Central region comes second. On the other hand the Southern regions have the lowest predicted probability of being out of school.

How does one go about measuring divergence in access to education across certain groups?

The above conclusions result from a robust analysis that follows a similar approach to a study by Keith Lewin and Ricardo Sabates in 2011, which uncovered trends in access to education in six Anglophone and seven Francophone Sub-Saharan African countries.

Access to education is measured based on responses to the following survey questions “Has [NAME] ever attended school?”; “Did [NAME] attend school (at any time) during the 20xx-20xx school year?” and “During this school year, what level and grade is [NAME] attending?” Then four mutually exclusive categories for children aged between 6 and 15 are identified:

1. Those who are currently attending school at the age appropriate level.
2. Those who are currently attending school but are 2 years older than the appropriate age for their level.
3. Those who are currently attending school but are 3 or more years older than the appropriate age for their level.
4. Those who are not currently attending school.

The econometric model used (in case you are wondering, a multinomial probit!) can estimate the marginal effect of certain factors on the predicted probability of belonging to each of the groups above. More specifically the probability of belonging to the last three groups compared to belonging to the first group. The factors that are considered in the model are:

1. Wealth – a variable indicating the quintile that the household belongs to, based on an index of household assets constructed using principal component analysis
2. The sex of the child
3. The sex of the head of household
4. The household composition (number of children below the age of 5, number of adults)
5. Mother’s and father’s level of education.
6. The area of residence (urban/rural)
7. Regional variable based on Nigeria’s geopolitical zones.

By controlling for these variables in the econometric analysis, the marginal effect of wealth, gender and region can be teased out when holding all of these variables constant.

Read the paper for a detailed description of this methodology and more results!

[1] Each quintile contains a fifth of households from the sample.