Errors in age reporting in Bengali populations
Edmonston, B.; Bairagi, R.
Rural Demography 8(1): 63-87
1981
ISSN/ISBN: 1010-3783 PMID: 12338519 Document Number: 391380
Reports on the major errors found in age reporting on population censuses and sample surveys among the Bengali speaking populations of what is now West Bengal, India and Bangladesh from 1881 to the present, and identifies cautions necessary in using Bengali age data. The problem of adjustment is not discussed. Age reporting is confused by the use of 2 calendars, the western (Julian) and a Bengali calendar in which the year begins in the middle of April. The Arabic calendar is also used among Moslems. Moreover, birthdays are not especially emphasized in Bengali culture. 4 types of age misstatement are examined: preference for particular ages, for age groups, for age digits (as years ending in 0 or 5), and the interaction of random error. The influence of the progress of literacy on age reporting is also examined. Special problems with reporting children's ages are noted. Underreporting of children 0-4 was found, both from underenumeration and from a tendency to inflate ages. A slight tendency for age preferences (especially ages 4 and 5) was found, and random error increased with age up to 7 years. Digital preference becomes more manifest with adolescence and early adulthood, 0 and 5 the most frequent digital endings. The effect becomes more pronounced among people over 50. Problems with data grouped by 5 year periods are thus intensified. Errors surrounding the age of majority seem especially critical for females. A tendency is to deflate the ages of girls until after menstruation and marriage, when their ages are adjusted to those more suitable for married women. Young unmarried females are heaped at 10-12; young marrieds at 15. Digital preference persists despite literacy improvement. Single year age data may be improved by analytical approaches such as standard smoothing techniques or developing alternative age groupings to minimize the distortion, or by more careful and sensitive data collection techniques. Data on 5 year age groups are seen to have become marginally acceptable in recent decades, showing some effect from improved literacy.