[Bias in the enumeration of single marital status for females in population censuses: a proposal for correction.]

This paper investigates the effect of consensual union dissolution on the observed levels of women never in union as estimated from the marital status variable in Latin American population censuses. We compare census data with data from Demographic and Health Surveys (DHS) for those countries and years where both sources are available (Bolivia, Brazil, Colombia, and Peru). Results show that the proportions of women never in union based on the census 'marital status' variable are systematically higher than those estimated from the DHS. This overestimation is due to individuals formerly in consensual unions declaring themselves as single at the time of the census. The large proportion of single women that have children is strong evidence of this and at the same time appears to be a valid and practical option to correct the bias.