Sex Ratios in South African Census Data 1970-1996

The sex ratio of a population or in an age group is the number of males per 100 females in that group. Attention has been drawn to the sex ratios in the 1996 Census of South Africa. A focus has been on the low sex ratios for those age 20-34. Concern has been expressed that these low sex ratios are implausible and differ from what was found in earlier censuses. Some have suspected that these low sex ratios are an indication of poor quality of the 1996 Census overall.

• The non-African population exhibited low sex ratios in the younger working ages in every census 1970-1996. The dip in sex ratios at the younger working ages for the population of South Africa as a whole was masked by high sex ratios at those ages among Africans in censuses before 1991. The high sex ratios at the younger working ages among Africans were likely due to international inmigrants who were predominantly male and to changes in the geographic regions which were included in the enumeration in the 1980-1991 censuses. These changes in geographic coverage led internal migrants in South Africa, who were mainly male, to have a similar effect on sex ratios by age as international migrants.

• When sex ratios by age are calculated based on those born in South Africa and for which, as much as possible, similar geographic regions are included, a pattern similar to that seen in 1996 is found in all censuses since 1970.

• In virtually every country in the world, implausibly low sex ratios are found in the younger working ages. This is because geographically mobile young men in those ages are undercounted relative to women of the same age. The data from South Africa have come to show the same implausible pattern of low sex ratios at the younger working ages as seen in other countries.

• Examination of cohort survival proportions by sex from 1985 to 1991 for those born in South Africa indicates more complete enumeration of those in their thirties than of those in their twenties for both sexes. The greater intensity of this for males than females is consistent with the age patterns found in the sex ratios.

• Examination of cohort survival proportions by sex from 1985 to 1991 for Africans born in South Africa shows substantially too many people reporting themselves as age 61-65, likely due to the desire to qualify for pensions.

• The low sex ratios for those in their twenties and early thirties compared to those older or younger in the 1996 Census of South Africa indicate a problem common to almost all censuses; they do not reflect a special problem with data collection in South Africa or with the population of South Africa. Like all national census offices, South Africa needs to seek ways to reduce the undercount of mobile young adults, especially mobile young men.

Dataset(s): South Africa Censuses of 1970, 1981, 1985, 1991, and 1996