Contraception-fertility link in Sub-Saharan Africa and in other developing countries

The present study evaluates various hypotheses advanced to explain this anomaly. To this end, the study disaggregates the data from 59 Demographic and Health Surveys (DHS) in 451 regions of these countries -- 200 from sub-Saharan Africa and 251 from other countries. Both regional- and national-level analyses clearly show a much lower association between the level of contraceptive practice and the total fertility rate in sub-Saharan Africa. A variety of explanations were statistically evaluated. Confining the comparisons of the correlations in the two parts of the world to the use of modern methods reduces the difference but by no means erases it. None of the other explanations seem to shed any light -- neither substituting more current measures of fertility, nor confining the analysis to marital fertility, nor taking into account the fact that postpartum insusceptibility is so much longer in Africa. Comparing the associations after separating contraceptive use into spacing and limiting proved only modestly significant. Substituting the total demand for family planning (adding unmet need and intention to use) for the standard contraceptive prevalence measure did nothing to reduce the difference. The key consideration seems to be that the populations in sub-Saharan Africa were clearly at the beginning of their fertility transition and were being compared with Asian and Latin American populations at much more advanced stages of the transition. By comparing the correlation for the same regions that were surveyed more than once, researchers were able to discern a clear trend in sub-Saharan Africa toward a higher correlation over time. The implication is that the association will eventually be the same in sub-Saharan Africa as elsewhere. At the beginning of the transition, there is little variation in fertility or contraceptive prevalence and thus only limited association between the two variables. As the transition develops, the variation of both measures and their covariance increases, leading to higher correlations. Support for this interpretation comes from the low correlation between fertility and contraceptive prevalence in the subset of regions of countries in Asia and Latin America still in the early stage of their transition. (Author abstract, modified)