At the beginning of the twentieth century the sex ratio for South Africans differed markedly according to racial group. Those for white South Africans remained almost invariable, with more boys than girls, while black South Africans had a clear majority of girls, a situation that the literature has almost completely overlooked. This high proportion of black girls was also present in most sub-Saharan countries. The reasons are still not completely clear. Sex ratios at birth show more births of boys than girls. Boys’ mortality was higher than girls’ mortality.
This study exploits a natural experiment to evaluate the gender bias effect associated with negative marking due to gender-differentiated risk aversion. This approach avoids framing effects that characterize experimental evaluation of negative marking assessments. Evidence of a gender bias against female students is found. Quantile regressions indicate that female students in higher quantiles are substantially more adversely affected by negative marking.