THE RELIABILITY OF REPORTING OF CONTRACEPTIVE BEHAVIOR IN DHS CALENDAR DATA - EVIDENCE FROM MOROCCO

This report addresses the consistency of reporting in the contraceptiv e calendar in the 1992 and 1995 Morocco Demographic and Health Surveys . Because a panel design was used in these surveys, the same women wer e interviewed in both years, providing a unique opportunity to examine the reliability of responses. Measures of reliability for various asp ects of contraceptive-use dynamics are computed, and the impact of rep orting errors on contraceptive failure, discontinuation, and switching rates is estimated. Results suggest that reporting of contraceptive b ehavior in Moroccan DHS calendar data appears to be relatively reliabl e at the aggregate level. Individual respondents, particularly those w hose contraceptive patterns have been complex, have a lower level of r eliability. The observed inconsistencies do not appear to affect aggre gate-level estimates of contraceptive prevalence; however, measures of contraceptive-use dynamics are less stable.