The crisscross method to evaluate data quality in fertility surveys

In this paper, I compare direct and indirect estimates of recent fertility to evaluate the quality of birth history data. The direct estimates are based on birth histories and computed in the same way as the published estimates in DHS. The indirect estimates are computed using the crisscross method (Schmertmann, 2002), based on cohort parity increments between successive surveys. Comparisons between direct and indirect estimates are used (1) to show that the data are affected data quality problems, and (2) to help identify the type of data quality problems (omissions of recent births, displacements, omissions of early births, differences in sample implementation). It could potentially be used to (3) help correct fertility estimates.