Marriage patterns in South Africa: Methodological and substantive issues Debbie Budlender, Ntebaleng Chobokoane and Sandile Simelane 1 Abstract The relative dearth of analytical work on marital status to date arises, in part, from the difficulties associated with collecting and interpreting data. The problems are particularly acute in South Africa as a result of the wide diversity in marriage forms, cultures, religions and languages. Inadequacies in coverage of large segments of the population during the apartheid years add to the difficulties. The paper describes some of the historical, legal and social reasons for the data difficulties. It then describes and examines data from various household surveys, the 1996 population census and administrative sources on type, prevalence and timing of marriage to assess the extent of the problems. Because census and survey data reflect perceptions of marriage, while administrative data generally record the legal system, the paper can consider differences between the two views of marriage, as well as differences between sources which purport to measure the same view. The paper concludes that despite the many problems in the data, careful analysis can reveal important trends in marriage patterns in the country. Keywords Marriage, South Africa, Household Survey data, legal reform Introduction Marital status is of interest to a wide range of disciplines. Economists, sociologists, anthropologists and demographers, among others, will often include marital status in their investigations. SAJDem 9(1):1–26 1 1 Debbie Budlender, Centre for Actuarial Research, University of Cape Town, Private Bag, 7700 Rondebosch. Email: [email protected]Ntabaleng Chobokoane and Sandile Simelane are from Statistics South Africa.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Marriage patterns in South Africa:Methodological and substantive issues
Debbie Budlender, Ntebaleng Chobokoane and Sandile Simelane1
Abstract
The relative dearth of analytical work on marital status to date arises, in part, from the
difficulties associated with collecting and interpreting data. The problems are
particularly acute in South Africa as a result of the wide diversity in marriage forms,
cultures, religions and languages. Inadequacies in coverage of large segments of the
population during the apartheid years add to the difficulties. The paper describes some
of the historical, legal and social reasons for the data difficulties. It then describes and
examines data from various household surveys, the 1996 population census and
administrative sources on type, prevalence and timing of marriage to assess the extent of
the problems. Because census and survey data reflect perceptions of marriage, while
administrative data generally record the legal system, the paper can consider differences
between the two views of marriage, as well as differences between sources which
purport to measure the same view. The paper concludes that despite the many problems
in the data, careful analysis can reveal important trends in marriage patterns in the
country.
Keywords
Marriage, South Africa, Household Survey data, legal reform
Introduction
Marital status is of interest to a wide range of disciplines. Economists,
sociologists, anthropologists and demographers, among others, will often
include marital status in their investigations.
SAJDem 9(1):1–26
1
1 Debbie Budlender, Centre for Actuarial Research, University of Cape Town, Private Bag, 7700 Rondebosch.
Figure 1 Per cent distribution of the married population by type of marriage,
RSA, 1995–1999
in the census is outside the range covered by the set of OHSs. Yet even in this
group, the percentage point difference between the census values and the
nearest OHS values (for 1997) are smaller than the difference between the
nearest OHS value and the next nearest value (1996 and 1998).
The patterns for African and Indian population groups are the least stable.
For the latter, the instability can be explained by the relatively small sample
sizes, especially when using OHS data. For the former, the instability could
reflect variability in how the different terms are interpreted. However, there is
no clear pattern in terms of instrument effects.
Looking at differentials in marriage form between urban and rural areas, it
is apparent that traditional marriages are a rare phenomenon in urban areas,
accounting for less than 20 per cent of all marriages. Both the census and the
OHS data show that traditional marriages are generally more common among
non-urban dwellers than civil marriages. This finding is not surprising given
that African people predominate in non-urban areas to an ever greater extent
than in urban and, as shown above, traditional marriages are more common
among Africans. The African predominance in non-urban areas does not,
however, explain all of the difference between the urban and non-urban
marriage patterns. If we confine our attention to Africans, the 1996 census
reveals that 30.7 per cent of married people in urban areas were married by
traditional rites, compared to 58.4 per cent in non-urban areas.
Types of union
Polyandry is so rare in South Africa that polygamy is treated as synonymous to
polygyny. As a result, the focus of this sub-section is on estimates of the extent
of polygyny. However, it is generally accepted that polygyny is no longer
widespread in the country (South African Law Commission 1999:3). The only
source of data on polygyny in South Africa is the 1998 Demographic and
Health Survey. These data show 4 per cent of married women reporting that
their husbands have other wives beside themselves. When disaggregated by
population group, the percentage is 7 per cent among married African women,
3 per cent among coloured, 2 per cent Indian and 1 per cent white. These
percentages may be overestimates as some women may have interpreted
“other wife” broadly to include women with whom the husband cohabits but
to whom he is not necessarily formally married. The possibility of this
happening is increased by the fact that the word for “wife” is the same as the
word for “woman” in some South African languages.
Marriage patterns in South Africa 17
The timing of marriage
Our direct estimates of mean age at first marriage are based on responses to a
question in the OHSs of 1997 and 1998 which asked how old the person was
when they first married or lived with a partner. (This question was not asked in
all the OHSs.) The responses could be biased downwards to the extent that
people report first age at which they started living together with someone
rather than first age at marriage.
This bias is likely, on the one hand, because living together is more
commonly reported for younger people than older people. (Of course, living
together might also be more common today than it was in the past, so that we
cannot assume that those who are now in their forties behaved like those
currently in their twenties when they were that age.) For example, there are
only 1.7 married people aged 16–19 years for every person of this age who is
reported as living together with a partner. Among 20–29 year olds, there are
3.2 married people for every person reported as living together. The ratio
increases for each age group, to 27.7:1 for 69–69 year olds and 38.3:1 for
those aged 70 and above. The age group 20–29 accounts for 35 per cent of all
people living together, with another 35 per cent in the age group 30–39 years.
On the other hand, the bias might be limited as the OHS 1998 data show
virtually no differences in the mean and median age of first marriage or
co-habitation between those who are currently married and those currently
living together. Further, living together is recorded as the current situation for
a relatively small percentage of people (see Table 1). We suspect that living
together is under-reported, and that similar under-reporting of living together
will occur in respect of the age at first marriage or cohabitation question. If this
is true, the bias in the calculation of first age at marriage will not be large.
The 1997 and 1998 OHSs suggest that the mean age at marriage for South
Africans is around 25 years. The breakdown by sex provides a mean of
28 years for males and 24 years for females, yielding an average difference of
four years between married couples (Table 7). These figures are higher than
those reported by the United Nations for most other African countries (United
Nations 1988:66–71). Table 7 also suggests that Africans (males in particular)
tend to marry much later than the other groups. Indians, on the other hand,
have the lowest mean age at marriage (24 years). Calculation of variances for
male and female of the different population groups reveals that these are
significantly larger for African and coloured men and women than for Indians
and whites. Further, while Africans marry later and Indians earlier, both
18 Southern African Journal of Demography 9(1)
groups have an average difference in the ages of husbands and wives of four
years, a difference which is larger than those for coloured and white couples.
Table 7 Mean age at marriage (direct method) by population group, sex and source
Population groupOHS 1997 OHS 1998
Male Female Total Male Female Total
African 28.4 23.7 25.7 28.1 23.5 25.5
Coloured 26.3 24.0 25.0 26.1 23.8 24.9
Indian 25.6 21.7 21.8 25.6 21.8 23.5
White 25.8 23.0 24.3 26.4 23.5 24.9
RSA 27.5 23.5 25.3 27.4 23.5 25.2
Source: OHS 1997 & 1998
The data do not show marked differences in the timing of marriages
between urban and non-urban areas. However, the average difference in the
ages of the spouses differs markedly between the two areas. Whereas urban
husbands have an average of three years of seniority over their wives, the
difference in non-urban areas is five years. This pattern is again at least partly
explained by the predominance of African people in non-urban areas.
The administrative data provide information about median age at marriage
rather than mean age. The medians from the administrative data and the OHSs
are not strictly comparable, in that the former refers to the age at which the
most recent marriage was entered into, while the OHS refers to the age of first
marriage. As South Africa has relatively high rates of divorce and remarriage,
this distinction could be important.
According to the OHS of 1998, the median age at first marriage is 22 years
for females and 26 years for males (Table 8). These values are significantly lower
than the 30 years for females and 34 years for males recorded in the
administrative data for 1998 (Statistics South Africa 2000). The difference in
respect of current and first marriage would explain at least some of the disparity.
The average age at marriage can be estimated indirectly by calculating a
measure called the singulate mean at marriage (SMAM). There are four points
that must be kept in mind about SMAM. First, the measure relies heavily on
the proportions of people who are reported as single in surveys or censuses. In
the case of South Africa this creates problems because of ambiguity as to how
to treat those who report themselves as “living together”. To circumvent this
Marriage patterns in South Africa 19
problem, the estimates below were calculated treating those living together as
“single”.
Table 8 Median age at marriage by population group and sex, 1998
Population group Male Female Total
African 27 22 24
Coloured 25 23 24
Indian 24 20 22
White 25 22 23
RSA 26 22 24
Source: OHS 1998
Second, while the SMAM is sensitive to age reporting errors, this is not a
matter of particular concern in South Africa, as the data collected in censuses
and surveys typically shows only a small extent of exaggeration in the age
range used in the calculation of a SMAM. Third, the computation of SMAM
from current status data in a single census or survey assumes that age at first
marriage has been constant over time and that differences in mortality or
migration rates by marital status are negligible. Fourth, SMAM results are
biased towards the more recent past. If age at marriage is increasing, then the
proportion single at younger ages will be higher now than a few years before,
and the SMAM should lie between the two.
Table 9 presents SMAM estimates from the 1996 census and the 1995 and
1999 OHSs, as the endpoints of the period covered by the more reliable OHSs.
As expected, the SMAM figures are higher than the estimates derived from the
direct method of estimating the average ages at which people get married.
Table 9 nevertheless confirms the late marriage for Africans shown by the
direct method. Indirect estimates for urban and non-urban are not shown as
they show no differences in the timing of marriage by place of residence.
The SMAM figures suggest a slight increase in the mean age at marriage for
all population groups and both sexes over the period 1995 to 1999. The
period is a short one and the increase is small. The pattern is nevertheless
worth monitoring in the coming years as it occurs alongside the slight increase
in the proportion of living together over the same period reported above.
20 Southern African Journal of Demography 9(1)
Table 9 Singulate mean age at marriage (SMAM) by population group and source
Source OHS 1995 Census ’96 OHS 1999
RSA
Male 31.9 31.4
Female 29.6 29.3
Total 30.8 30.3
African
Male 33.1 32.4 32.4
Female 30.4 30.2 30.6
Total 31.8 31.2 32.7
Coloured
Male 30.3 29.2 31.0
Female 30.5 28.6 28.7
Total 30.1 28.8 30.2
Indian
Male 28.2 26.9 28.9
Female 24.4 23 26.6
Total 26.6 25.3 27.7
White
Male 26.2 27.4 28.4
Female 24.1 25.5 26.6
Total 25 26.3 27.2
Source: OHS 1995 & 1999; Census ’96
The prevalence of marriage
The ambiguity regarding the classification of the “living together” category
also affects estimations of the prevalence of marriage. This is so because this
index is dependent on the percentages of people single between the ages 45
and 54. As before, we treat those living together as single.
Table 10 shows the average percentages of people ever married by age 50 by
population group and sex over the period 1995 to 1999. All the data sets
provide comparable estimates. The prevalence of marriage among South
Africans during the period 1995–1999 averaged 83.4 per cent.
Marriage patterns in South Africa 21
Two distinct patterns emerge regarding the prevalence of marriage by
population group. During the period 1995–1999 marriage was almost
universal among whites (94.8 per cent) and Indians (92.2 per cent), treating
those “living together” as single. By contrast, the estimates for Africans and
coloureds suggest that marriage was far from universal among the two groups.
When those “living together” are considered as single, the proportions ever
married were 80.1 per cent and 83.1 per cent for Africans and coloureds,
respectively. Cohabitation is more prevalent – or is reported more often –
among Africans and coloureds than among Indians and whites.
Table 10 Average per cent aged 50 and above ever married by population group
and sex, 1995–1999
Population group Male Female Total
African 79.8 80.4 80.1
Coloured 83.3 82.8 83.1
Indian 92.0 91.8 92.2
White 94.3 95.3 94.8
RSA 83.2 83.5 83.4
Source: OHS 1995–1999
The analysis of the prevalence of marriage by place of residence does not
indicate a large difference between urban and non-urban locations. The
proportion of the population ever married by age 50 averages 84 per cent in
urban areas, and 82 per cent in non-urban areas.
Conclusions
The main aim of this paper is to describe and assess the available data on
marriage in South Africa. Despite the wide range of data sources available,
several weaknesses in these data have been identified which present problems
for the analyst trying to describe and understand marital patterns in the country.
One of the primary problems is the wide range of marriage practices in the
country, and differing cultural understandings as to what constitutes marriage.
This problem is aggravated by linguistic issues, in that the terms used in the
data collection instruments will be understood differently by different
informants and fieldworkers. Further, the instruments are not consistent in
the way they use terms and divide marriages into groups.
22 Southern African Journal of Demography 9(1)
A second important source of data disparities is the difference between the
legal definition of marriage – which is what is recorded in the administrative
data – and the sociological perception of the people involved and other
members of their household as to what constitutes marriage. Examination of
the administrative data reveals missing information in respect of type of
marriage, as well as fluctuations in provincial statistics from one year to the
next. These deficiencies suggest that registration of marriage, like that of births
and deaths, might not be recording even the legal situation as well as it should.
This clearly complicates comparison of the legal and sociological ‘pictures’ of
marriage in the country.
The recent change in the law, in the form of the Recognition of Customary
Marriages Act, will in the future further complicate the statistics as the
coverage of administrative data will increase, but by an unknown amount. One
of the aims of this paper was to get some idea of how great this increase might
be. The analysis above, however, shows smaller differences than expected
between data that exclude customary marriages and those that include them.
Further, the fact that registration is not compulsory, and the past and ongoing
possibility of marriage by both customary and civil law simultaneously will
obscure the meaning of observed changes in the statistics.
When looking at type of marriage, survey and census data provide very
similar pictures despite different grouping of categories in the two
instruments. Calculations of mean age at marriage produce higher values than
in most other African countries. These cannot necessarily be explained by a
higher proportion of people living together as the question about age at first
marriage includes age of first co-habitation. Further, the figures for median
age of marriage in the administrative data are even higher than those in survey
and census data. This latter difference is at least partly explained by the fact
that the administrative data refer to the current marriage, while the census and
survey data refer to first marriage. However, it is not clear that this would
account for most of the observed difference.
Data on the prevalence of marriage seem generally consistent. Data on both
marriage and living together reveal significant differences between population
groups, as well as some differences between urban and non-urban areas. In
terms of polygyny, only one source of data is available. Unfortunately, the way
this question was phrased in the DHS might well have been misinterpreted and
the derived estimate is thus open to question.
Marriage, entry into marriage, and their measurement are clearly not
Marriage patterns in South Africa 23
simple concepts, particularly in multicultural, multilingual societies such as
South Africa. Nevertheless, despite the inevitable inconsistencies that are
inherent to data on marriage from different sources, careful analysis of these
data allows researchers to investigate nuptiality and to identify trends in
marriage patterns in the country.
ReferencesBah, S. 1999. “The Improvement of Marriages and Divorce Statistics in South Africa:
Relevance, registration issues and challenges”, in Bah, S. and S. Rama (eds).Towards Improving the Registration of Marriages and Divorces in South Africa:Proceedings from a national workshop. Pretoria: Statistics South Africa, pp.5–10.
Budlender, D. 1999. “How Can Statistics on Marriages and Divorces Help Us toUnderstand the State of the South African Family and Household” in Bah, S. andS. Rama (eds). Towards Improving the Registration of Marriages and Divorces inSouth Africa: Proceedings from a national workshop. Pretoria: Statistics SouthAfrica, pp.11–17.
Central Statistical Service. 1991. Marriages and divorces, 1991. Report No. 03-07-01(1991). Pretoria: Central Statistical Service.
Central Statistical Service. 1994. Marriages and divorces, 1994. Report No. 03-07-01(1994). Pretoria: Central Statistical Service.
Goldblatt, B. 1999. “Living together without legal protection”. Gender ResearchProject Bulletin 4:3–7.
Hajnal, J. 1953. “Age at Marriage and Proportion Marrying”. Population Studies 7(2):111–135.
Mbatha, L. 1997. “Marriage proposals – can they work?” Gender Research ProjectBulletin 2:6.
Mbatha, L. 1998a. “Recognising customary marriages”. Gender Research ProjectBulletin 3:4–5.
Mbatha, L. 1998b. “Full legal recognition for customary marriages?” Gender ResearchProject Bulletin 3:6–7.
Nhlapo, T. 1999. “The Customary Marriages Act: Background and rationale,” inBah, S. and S. Rama (eds). Towards Improving the Registration of Marriages andDivorces in South Africa: Proceedings from a national workshop. Pretoria: StatisticsSouth Africa, pp.3–4.
South African government. Recognition of Customary Marriages Act, Act 12 of 1998.
South African government. Recognition of Customary Marriages Act, 1998 (Act No 12of 1998) Regulations, R1101, 1 November 2000.
South African Law Commission. 1999. The harmonisation of the common law and theindigenous law: Report on customary marriages. Project 90. Pretoria: South AfricanLaw Commission.
24 Southern African Journal of Demography 9(1)
Statistics South Africa. 1999. Marriages and divorces, 1996. Report No. 03-07-01(1996). Pretoria: Statistics South Africa.
Statistics South Africa. 2000. Marriages and divorces, 1997. Release P0307. Pretoria:Statistics South Africa.
Statistics South Africa. 2002. Marriages and divorces 1999. Release P0307. Pretoria:Statistics South Africa.
United Nations. 1988. First Marriage: Patterns and Determinants. New York.Department of International Economic and Social Affairs.