Does matrilineal kinship weaken the marital bond Evidence from Ghana

1
Does matrilineal kinship weaken the marital bond?
Evidence from Ghana

• Stephen Obeng Gyimah, PhD
• Department of Sociology
• Queens University
• Kingston, Ontario
• gyimahs_at_post.queensu.ca
• Baffour K. Takyi, PhD
• Department of Sociology
• University of Akron
• Akron, Ohio
• btakyi_at_uakron.edu
• Prepared for presentation at the annual meetings
  of the Canadian Population Society, Winnipeg,
  June 2-5 2004.

2
Research context

• Theoretical formulations about family
  transitions, particularly marital outcomes, in
  sub-Saharan Africa assume that the changes
  underway in the region are the result of
  modernizing processes, particularly Africas
  encounter with the outside world due to its long
  history with European imperialism and
  colonialism.
• Yet still, others argue that the regions rapid
  rate of urbanization, estimated at about 4.4 per
  cent per annum and changes in womens economic
  position arising from increasing education may be
  the driving force behind the fragmentation and
  changes that we are seeing in African families.

3

• While a common theme behind recent scholarship on
  the African family focuses on the role of social
  change in explaining family transitions, few
  examine the potential influence of internal
  processes within African societies and their
  impact on family life.
• In this paper, we examine the role of family ties
  (which we define in terms of lineage or kinship
  allegiance) on the duration of first marriages in
  Ghana.
• Despite this growing research interest on the
  African family, surprisingly little of existing
  deals with marital stability and dissolution .

4

• The limited research on marital disruption is
  that, until quite recently, the focus of most
  African demographic researchers was in explaining
  the regions high fertility levels that were
  viewed as detrimental to socioeconomic
  developments.
• Because of the interest on fertility-related
  issues, a sizable amount of existing demographic
  studies on postcolonial Africa investigate the
  determinants of high birth rates, and the
  conditions under which a transition from high to
  low birth rate would occur in the region.
• Given the paucity of studies on marital
  instability, findings from our study could
  provide insights into family dynamics and
  processes in the sub-Saharan region.

5

• In exploring how family ties affect marital
  stability in the context of Ghana, we are
  particularly interested in assessing a key
  hypothesis that is derived from the institutional
  theoretical framework that suggests that family
  ties, particularly matrilineal ties, undermine
  the marital bond and thus, increase womens risk
  for marital disruption (see e.g., Oppong, 1974
  1977 Poewe, 1978 Hagan, 1983 Hutchison, 1990).
  

6
Why Ghana?

• Three main factors influenced our choice of Ghana
  as the setting for testing our hypothesis. First,
  roughly half of Ghanas population is comprised
  of the matrilineal Akan and the other half are
  non-matrilineal (Gaisie, 1981).
• These two family arrangements serve as the basis
  of Ghanaian social organizations and are
  reproduced in succeeding generations through
  carefully crafted socialization processes and
  systems of rewards (Dodoo and Tempenis, 2000).
• These differential rewards, expectations, and
  influence is likely to have a differential impact
  on family dynamics in the country. This is
  particularly the case in matrilineal societies
  where individuals tend to defer more to the
  preferences and needs of their lineage kin than
  in non-matrilineal settings.

7

• Second, evidence is emerging to indicate that a
  transformation is underway in many African
  families (see e.g., Bledsoe, 1990 Lloyd and
  Gage-Brandon, 1993).
• With regard to Ghana, Amoateng and Heaton (1989)
  found women who married in the 1970s to be twice
  as likely to report a divorce as those who
  married in earlier decades. Lloyd and
  Gage-Brandon (1993), and Gage and Njogu (1994)
  have also reported that during the 1970s, about
  40 of ever-married women between the ages of
  40-49 in Ghana reported a marital dissolution,
  with this figure increasing to a high of 60.8 by
  the 1980s (Gage and Njogu 1994).
• Despite these observations, researchers disagree
  on the forces that are influencing these recent
  transitions in family life in Ghana.

8

• Another reason for using Ghana is that data from
  Demographic and Health Surveys conducted in that
  country during the 1980s and 1990s have measures
  that allow us to distinguish between the two
  predominant family traditions and ties
  matrilineal and non-matrilineal ties for the
  study.
• Using this categorization, we estimate hazard
  models to examine the duration of first marriage
  to address the question of whether women from
  matrilineal family backgrounds experience more
  disruption than their non-matrilineal
  counterparts.
• Because the disruptive-effect thesis is
  predicated on the assumption that in matrilineal
  societies women tend to have more support and
  also autonomy in their social relations than
  their non-matrilineal counterparts (Clignet,
  1970 Poewe, 1978 Takyi, 2001), we hypothesize
  that the risk of marital disruption will be
  greater among matrilineal than non-matrilineal
  women.

9
Background

• Studies on marital stability have in the recent
  past explained the reported fragility of
  African marriages in terms of structural forces.
• This thesis argues that social changes arising
  from what is considered as modernizing influences
  and the introduction of Western ideas and norms
  into the region have something to do with recent
  marital transitions (Kaufmann and Meekers, 1998
  Takyi, 2001 McDonald, 1985 Caldwell, 1982).
• While acknowledging the possible influence of
  micro-level factors such as age at marriage, and
  increasing levels of womens education on marital
  stability, the undue emphasis on structural
  forces ignore the potential role of internal and
  institutional forces such as family ties that
  could potentially affect family and marital
  outcomes in sub-Saharan Africa.

10

• The issue of whether kinship ties influence
  marital outcomes is important given the
  overwhelming influence of lineage or kinship ties
  in Africa.
• Indeed, research from the region have
  increasingly pointed to the saliency of lineal
  ties to our understanding of childbearing and
  rearing patterns (Page, 1989), and overall
  reproductive behavior (Caldwell, 1982
  Lesthaeghe, 1984 Caldwell and Caldwell, 1987),
  and also divorce (Takyi, 2001).
• In the context of West Africa, Caldwell (1995)
  also notes that family members rarely leave the
  extended family, with women traditionally
  worshipping in their households of origin even
  after marriage.

11

• More importantly, women and their children tend
  to be accepted readily back into their families
  of origin if their marriages break up, a practice
  that is common among matrilineal kin members.
  These dynamics, coupled with the ambiguous
  position of the conjugal family vis-à-vis lineal
  could affect intra-family relations which could
  contribute to the destabilization of the marital
  union.

12
Theoretical framework

• Three main mechanisms through which matrilineal
  family ties can affect the stability of the
  conjugal unit.
• 1) under the matrilineal family system, married
  couples rarely pool their resources together for
  the benefit of the conjugal family unit (Oppong,
  1983a, 1983b).
• While feminists scholars would argue that such a
  practice provide women with some level of
  autonomy from their spouses, the lack of a joint
  account is consistent with the traditional belief
  under the matrilineal system (especially as
  practice in Ghana) that men and women should give
  priority to their matrilineal kin over their own
  spouses (Clark (1999).
• Takyi (2001) and Dodoo (1998) have argued that
  the practice of maintaining separate marital
  accounts, plus the spousal allegiance of the wife
  to her own maternal family of origin, could
  undermine the authority of the husband, weaken
  and compromise the marital unit, and provide
  women and men with different cost and benefits
  calculations, which could ultimately influence
  the wifes marital decisions.

13
Mechanisms linking lineage ties to marital
outcomes

• 2. the second borrows its ideas from rationale
  choice theory (see e.g., Klein, 2002). Consistent
  with this view, it has been suggested that the
  transactions that lead to the exchange of
  resources between family members in the form of
  bridewealth payments prior to the consummation of
  the union may influence the wifes actions within
  the marriage.
• 3. The third is the status of women of
  matrilineal descent . Research on the status of
  women in the developing world have pointed to the
  dependency-relationship that arises between
  couples when the wife rely on the husband for
  financial support.
• Under the matrilineal system of descent, family
  members are guaranteed significant social
  support, benefits and freedom not found among
  non-matrilineal societies. These benefits may
  include for example valued resources such as
  access to land (see e.g., O'Rourke, 1995).
  Besides, non-biological members tend to get
  significant support and benefits outside their
  own conjugal family unit.

14
Summary of key hypotheses

• Matrilineal family ties could weaken family
  cohesion thereby increasing the rate of marital
  disruption. Hence, we expect the risk of marital
  dissolution to be higher among matrilineal than
  non-matrilineal women.
• The relative degree of autonomy that matrilineal
  women derive from their non-conjugal family
  members could in turn increase womens relative
  risk for divorce. Thus, we expect rates
  matrilineal rather than non-matrilineal women to
  have lower duration of marriages.

15
Methodology

• The present analysis is based on pooled data from
  ever-married women interviewed in the 1988, 1993,
  and 1998 Ghana Demographic and Health Surveys
  GDHS.
• The GDHS is a nationally representative,
  stratified, self-weighting probability sample
  survey of women aged15 to 49 years.
• The merged file yielded a sample size of 10, 843
  ever-married women.

16
Measures

• Duration of first marriage was used as the
  dependent variable.
• The main independent variable, kinship structure,
  was derived from the question on ethnicity.
  Coded as matrilineal or non matrilineal.
• For the multivariate models, we controlled for
  socio-demographic factors that have been linked
  to marital stability, including religion,
  education, place of residence, presence of
  children in the household, type of marital union,
  age at first marriage, history of premarital
  births, and marriage cohort.
• These are factors that have been shown to
  influence marital outcomes.

17
Analytical model

• Because the duration of first marriage is
  right-censored at the time of the survey,
  standard regression procedures are deemed
  inadequate.
• For our study, Cox proportional hazards models
  are used to assess the links between matrilineal
  family ties and the duration of first marriage.
• We examined the proportionality assumption by
  using the the stphtest in STATA

18
Findings
19

• Figure 1 describes the relationship between the
  duration of first marriage among matrilineal and
  non-matrilineal women.
• The survival plot suggests a significant
  difference between the matrilineal and
  non-matrilineal women when it comes to their
  first marriages.
• At all durations, we find evidence that the
  proportion of matrilineal Akan women who
  experienced a marital disruption was higher than
  non-matrilineal women.
• For example, at 15 years of marriage, about 25
  of Akan marriages had been dissolved compared
  with only 10 of non-Akan marriages

20
Multivariate Findings
21

• Overall, the results indicate that matriliny has
  a strong and robust effect on the duration of
  marriage.
• In Model 1, there is evidence that the
  matrilineal Akans have a signicantly higher risk
  of marital disruption -- the risk being 57
  percent higher--than their non-Akan counterparts.
  
• In Model 2, the effect of kinship ties remains
  strong and robust, suggesting that religion does
  not explain kin differences in marital duration.
  In Model 3, the effects of kin are shown to be
  independent of socio-economic factors of
  education and place of residence. Across models,
  there is evidence of a robust effect of kin ties
  on marital duration affirming its independence
• These results suggest that the ethnic differences
  in marital duration cannot be attributed to
  socio-economic and demographic factors, and thus
  provide support for our hypotheses.

22
Summary

• In all our models, we found empirical support for
  the reported anthropological findings with regard
  to the potential linkages between matrilineal
  social organizations and marital instability in
  Africa.
• The fact that family ties is an important
  predictor of marital duration should in no way be
  construed to mean that other structural processes
  are not relevant in the context of Africa.

23
Summary

• Indeed, as the case is around the world, social
  changes and globalization are having immense
  influence on social relations and family
  processes in Africa as well.
• Thus, in arguing that cultural processes are
  relevant for our understanding of marital
  outcomes in Africa, what we argue here is that,
  in addition to the widely reported variables
  associated with structural changes, researchers
  need to consider the cultural influences which
  could confound marital decisions and outcomes in
  a region where family ties are salient in
  everyday discourse.