This study in Bangladesh found that inter-cluster variation in the use of modern reversible methods of contraception was significantly attributable to the educational levels of the female family planning workers working in the clusters. Women belonging to clusters served by educated workers had a higher probability of being contraceptive users than those whose workers had only completed primary education. At the household level, important determinants of use were socioeconomic status and religion. At the individual level, the woman being the wife (...) of the household head and having some education were positively related to her being a user. The model also found that inter-household variation was significantly greater than inter-cluster variation. Finally, the study concludes that after controlling for various covariates at all three levels, the clusters do not have significantly different levels of use of modern reversible methods of contraception. There are, however, some special areas where contraceptive use is dramatically low, and these contribute significantly to the observed inter-cluster variation. (shrink)