What are the usual sources of data on vital events ? What types of error usually occur in census data on age ? How are such errors adjusted ?
10 marks · directive: Discuss
The core of a good answer
Vital events data come from civil registration, sample registration systems, population censuses, sample surveys and hospital or other administrative records; census age data suffer from age heaping (digit preference for ages ending in 0 and 5), avoidance of certain ages, systematic over or under statement, and omission of infants, which are measured by Whipple's and Myers' indices and corrected by grouping into 5 or 10-year groups, graduation and smoothing formulae, and reconciliation with other sources.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
Three parts: sources (with their strengths), types of age error, and methods of measuring and adjusting them.
Answer outline
Introduction
Vital events are births, deaths, foetal deaths, marriages, divorces and migration. Their rates need both event counts (numerators) and population at risk by age and sex (denominators, mostly from the census).
Sources of vital statistics
- Civil registration: the continuous, permanent, compulsory and universal recording of vital events (live births, deaths, foetal deaths, marriages, divorces); its primary purpose is legal documents, and it is also a source of vital statistics.
- Sample registration: continuous recording of events in a sample of areas, used where civil registration is incomplete.
- Population census: age-sex structure, fertility questions (children ever born, births in last year), indirect mortality estimates.
- Sample surveys: demographic and health surveys and other household surveys.
- Administrative records: hospital records and medical certification of cause of death.
Types of error in census age data
- Age heaping or digit preference: excess at ages ending in 0 and 5 (and even numbers), common where people do not know exact age.
- Digit avoidance: deficits at ages ending in 1, 9 or culturally unlucky numbers.
- Systematic misstatement: exaggeration of age by the very old; understatement by young women; shifting of girls' ages near marriage age.
- Under-enumeration: infants and young children (0 to 4) often missed; differential coverage by sex.
- Coverage and coding errors: double counting, mistakes in processing.
Measuring the errors
- Whipple's index: 5 × (population aged 25, 30, ..., 60) ÷ (population aged 23 to 62) × 100; 100 means no heaping on 0 and 5, 500 means all ages reported end in 0 or 5.
- Myers' blended index: measures preference for each terminal digit 0 to 9.
Adjustment methods
- Grouping single years into 5 or 10-year groups, chosen so that heaped ages fall inside groups (for example 23 to 27, 28 to 32).
- Smoothing and graduation: moving averages or graduation formulae applied to the single-year or grouped counts.
- Comparison with previous census survivorship (cohort analysis) and with stable population models.
- Post-enumeration surveys to estimate and correct coverage error.
Conclusion
Vital rates combine registration numerators with census denominators, so errors in census age reporting, especially heaping on digits 0 and 5, must be measured by indices such as Whipple's and Myers' and corrected by grouping and graduation before computing age-specific rates.
Examples and data to cite
- The United Nations scale grades Whipple's index below 105 as highly accurate and above 175 as very rough age data.
Facts to remember
- Whipple's index ranges from 100 (no heaping) to 500.
- Age heaping is corrected by grouping and graduation.