The Reliability of Calendar Data for the Reporting of Contraceptive Use: Evidence from Rural Bangladesh

Rebecca Callahan, Johns Hopkins Bloomberg School of Public Health

The use of calendar methods for collecting data on contraceptive use experience has become standard practice in large-scale population surveys such as the DHS and the NSFG. The reliability of these methods for capturing accurate contraceptive histories remains questionable, however, given that only one previous study has assessed calendar reliability (Strickler, 1997). This study uses contraceptive use reports obtained with a 43-month retrospective calendar (40 months at baseline) from women participating in both rounds of a large, household survey in rural Bangladesh. Reporting concordance of contraceptive use during the 3-5 months of survey overlap is measured with a Kappa statistic. Logistic regression is used to explore predictors of overall reliability in reporting as well as method-specific reliability. Young age, school attendance, and use of long-term methods will be positively associated with reporting reliability. The complexity of contraceptive history will modify the relationships between predictors and reporting reliability.