The ecological design did not permit us to obtain direct estimates of the effect of child labor in exposed versus unexposed populations. Therefore, we could not be certain, for instance, that the children experiencing greater morbidity and mortality in a given population were actually child laborers.