Methods and tools to evaluate the quality of vital statistics
Other direct methods to measure completeness of birth registration
Two other methods are commonly used to estimate birth completeness.
Capture–recapture (direct) methods
When registered births are linked to births from another data source that cover the same population and period, methods such as the Chandrasekar–Deming technique can be applied to estimate completeness of registration.1 The data sources that can be linked to CRVS data include an existing routine reporting system or government records (for example, a health centre or hospital, baptism or burial records, school enrolment), a Health and Demographic Surveillance System (HDSS) site, a household survey or a population census. This method compares the overlap of events between two or more sources to estimate the ‘true’ number of events without assuming that either source is correct.
The following principles apply to capture–recapture methods:
- The data sources need to be independent of each other.
- The population of each source must have the same geographic boundaries and definition of residency.
- There should be minimal migration in and out of the population if the other data source is a retrospective household survey or a population census.
- The data sources must be able to be successfully linked.
- The process of linking datasets should not be too costly and time intensive.
Household reporting of birth registration
Many DHSs and MICSs collect data on birth registration, where registration is defined as ‘children whose birth certificate was seen by the interviewer or whose mother or caregiver says the birth is registered’. UNICEF collates this information for all countries with such data, and defines completeness as the ‘percentage of children under age 5 whose births are registered at the moment of the survey2. The definition used could potentially overestimate birth registration completeness, because it relies on the respondent (parent or caregiver) reporting whether the birth is registered or not and it ignores the children who died before being registered.
Occasionally, household censuses ask this question as well, but response data would need to be filtered out so that birth registration rates were calculated just for children and not for children and adults combined.
1 Chandrasekar et al (1949). On a method of estimating birth and death rates and the extent of registration. Journal of American Statistics Association.
2 UNICEF (2013). Every child’s birth right: inequities and trends in birth registration, UNICEF, New York.
Moultrie T (2013). The relational Gompertz model, in Tools for demographic estimation, IUSSP, Paris, pp. 54–68.
Moultrie T (2013). Evaluation of data on recent fertility from censuses, in Tools for demographic estimation, IUSSP, Paris, pp. 44–50.