Using data collected during school-based surveys in two regions in Ethiopia, bivariate and multivariate analysis has been carried out to identify some of the important constraints to primary schooling. Problems in obtaining a representative sample of children not in school for this type of analysis are identified, and allowances are made for these problems, where possible.
The results provide some evidence that well-nourished children from wealthier households whose mother and father are literate are more likely to attend and complete primary school. Furthermore, starting school at the official starting age of seven years and not repeating a grade increases the chances of completion. For children living within the vicinity of the school, opportunity costs do not appear to have a significant impact on the probability of attending or completing school, although girls in and out of school have a heavier work burden, relative to boys.
Whilst similar factors are found to influence the probability of both boys and girls attending and completing school in the sample, their impact on girls is greater – under-nourished girls from households lacking possessions, whose parents are illiterate, are 20 per cent less likely to attend school and 16 per cent less likely to complete school compared to a boy with these characteristics.
Starting school above the official age and repeating a grade are also shown to reduce the chances of girls completing school by a slightly greater amount than boys. Statistical analyses of the kind used in this paper are not able to control adequately for cultural constraints to schooling, although these are recognised as potentially providing an additional important explanation for the gender gap.