The main arenas where poverty and schooling interact to the detriment of girls (and children in general) are the household and the State. Shortages of school places and low school quality remain major problems in Ethiopia and Guinea, even though some one-fifth of recurrent government expenditure is spent on education in both countries.

At household level, an important reason why children are not sent to school is that the direct costs parents have to meet are often too great. Such costs exist even where school fees are not charged. In both focus countries, parents incur some expenditures for school uniforms, text and exercise books, and may also have to provide other inputs in cash or kind. The second reason is that poorer households may depend upon the labour of their children in order to supplement household income - either directly, on the farm or in the market place, or indirectly, by children undertaking household tasks which liberate adult labour for other remunerated work.

The factors noted above are the main levers which convert poverty at the level of the State and the household, into circumstances where not all children in the eligible age group attend primary school. There is however, an array of non-economic factors that cause rationed enrolment opportunities to be unequally allocated between boys and girls. They operate in the domains of the household, the school, the labour market and the society as a whole.

Gendered roles in society change the balance of incentives for girls and boys to attend school. In both countries, conventional opinion encourages women to see their future as centred on the home and the family. It implies that incentives for girls to attend and to perform well in school are less important than those for boys. In Ethiopia girls are married as early as the age of eight years in some areas. Being kidnapped for marriage from - or on the way to - school is not uncommon.

In Guinea, initiation ceremonies are sometimes performed during term-time. Absenteeism often follows, particularly for girls who, unlike boys, are often ashamed to return to school. The labour market in both countries has a critical influence upon the perceived benefits of schooling. Many jobs are reserved - formally or informally - for males. The gender balance of formal employment is strongly male, and gender discrimination in the labour market is apparent. Thus, the perceived benefits of schooling are greater for boys than for girls.

The environment of the school is often more hospitable to boys than to girls. Toilet and other facilities are often inadequate for girls. Harassment from boys or male teachers sometimes occurs, and the journey to school has greater attendant risks for the safety of girls than of boys. In both countries, girls spent more time performing non-school activities during school hours. In addition, teachers indicated their belief that boys are more intelligent than girls and it is not surprising, therefore, that female pupils conform to the expectations that they will be less successful than their male peers.

Finally, much household behaviour is responsive to the broader contextual conditions of the society and the labour market. But it is clear that the gendered division of labour within the household sharply affects the relative chances of girls and boys attending school. The studies revealed that girls are expected to perform household chores, and to look after younger siblings more often than boys. Moreover, where a girl's allegiance after marriage is mainly to her future husband's family, the balance of perceived benefits to parents favours the education of sons over daughters.

 For policymakers the evidence from Ethiopia and Guinea offers significant lessons, not least that:

• economic development is neither a necessary nor a sufficient condition for securing moves towards gender equity in schooling
• adverse cultural practice runs deep in the domains of society, households, labour markets and schools and its relationship with income growth is uncertain
• adverse cultural practice has to be removed if gender equity in schooling is to be achieved
• removing adverse cultural practice will require specific policy interventions. It will not suffice simply to wait for history's long march to do the trick.