ArticlePDF Available

Social Organization and Reproductive Behavior in Southern Ghana


Abstract and Figures

Abstract The objective of this research is to examine,the association between,social organization and reproductive behavior in one,setting in sub-Saharan Africa. The particular focus is on the effects of social organization on the diffusion of innovative reproductive ideas and behaviors. Social diffusion is assumed,to be strongly affected by patterns of informal social interaction, and these in turn are assumed to be deter- mined in part by the social organization of local communities (gender relations, employment activity, voluntary organizations). The research draws on data collected in six communities,in southern Ghana. The analysis reveals a weaker than expected association between,the social organization of the communities,and key reproduc- tive indicators (fertility preferences, age at first marriage, postpartum practices, use
Content may be subject to copyright.
Social Organization and
Reproductive Behavior in
Southern Ghana
Dominic K. Agyeman
John B. Casterline
2002 No. 167
Dominic K. Agyeman is Professor of Sociology, University of Cape Coast, Cape Coast,
Ghana. John B. Casterline is Senior Associate, Policy Research Division, Population
Council, New York.
This research was supported by an award from the National Institute of Child Health
and Human Development to the Population Council (R01-HD34524) and by an award
from the Rockefeller Foundation to the University of Cape Coast.
Social Organization and Reproductive Behavior
in Southern Ghana
Dominic K. Agyeman
John B. Casterline
This material may not be reproduced without written permission from the authors. For a
list of Policy Research Division Working Papers, including those available for down-
loading in PDF format, see
The objective of this research is to examine the association between social
organization and reproductive behavior in one setting in sub-Saharan Africa. The
particular focus is on the effects of social organization on the diffusion of innovative
reproductive ideas and behaviors. Social diffusion is assumed to be strongly affected
by patterns of informal social interaction, and these in turn are assumed to be deter-
mined in part by the social organization of local communities (gender relations,
employment activity, voluntary organizations). The research draws on data collected
in six communities in southern Ghana. The analysis reveals a weaker than expected
association between the social organization of the communities and key reproduc-
tive indicators (fertility preferences, age at first marriage, postpartum practices, use
of modern contraception). Closer examination of the six communities suggests that
the weak association is explained by the idiosyncratic histories of several of the
communities, in particular their histories of health and family planning provision.
Explanations for reproductive change that place social organization on center stage
must be enlarged to incorporate the potentially powerful influence of community
In this paper we join together two research traditions that have attempted to ex-
plain societal-level variation in fertility. Both traditions have a long heritage and a large
literature. The first posits that reproductive behavior is heavily influenced by the social
organization and cultural practices of a society. While this proposition can be applied to
any society, whatever its cultural heritage or level of income (or, for that matter, level of
fertility), perhaps in no region has this proposition held greater sway over research on
reproduction than in sub-Saharan Africa. The tradition goes back at least to the seminal
volume edited by Lorimer in the 1950s (Lorimer 1954), continuing with much original-
ity in Caldwell’s work in the late 1970s (Caldwell 1982), and revived once more in the
stimulating set of papers collected by Lesthaeghe (Lesthaeghe 1989) and the ambitious
multivolume National Academy of Sciences assessment of the demography of sub-Sa-
haran Africa (NRC 1993). The proposition can be broken into several distinct parts—
social organization can be distinguished from cultural practices, and both can be subdi-
vided further—but a common mode of analysis unites the various strands in this research
tradition: societal-level variation in reproduction is explained by the imperatives con-
tained in the cultural and social systems that have been passed down through the genera-
tions. The key imperatives might concern childbearing itself, or any of the chief proxi-
mate determinants of fertility, such as marriage and postpartum behaviors. This school
of thought struggles to explain reproductive change. This is not to say that the major
theoretical frameworks linking social organization and reproductive behavior cannot
accommodate reproductive change, but one must look outside this tradition for a full-
fledged theory of why and how reproductive change might occur.
Social change is the main preoccupation of the second research tradition that
informs the work in this paper. The innovation diffusion tradition posits that social change
can occur as the result of novel ideas, technologies, and behaviors spreading from one
social group to another (Rogers 1983). Obviously individuals must perceive that adop-
tion of the innovation will be to their advantage, but cost–benefit assessment that favors
the innovation is not by itself sufficient: individuals must be exposed to the innovative
ideas, technologies, and behaviors, and ordinarily this requires their introduction from
the outside, that is, innovation diffusion. There are several means by which innovations
can be introduced into a group. It might occur deliberately through the mass media or
formal schooling, as has been the case with modern contraception in sub-Saharan Africa
and other high-fertility societies in recent decades. Perhaps more interesting from a
social scientific standpoint, and more challenging as well, is innovation diffusion through
social networks. This channel is comprised of various mechanisms, including informal
day-to-day conversation with friends and neighbors, discussion with workmates, and
exchanges in formal group meetings. Social networks vary in size and structure from
one society to another, and this variation may well have implications for the ease and
rapidity of innovation diffusion (Valente 1995). A common argument, to which we re-
turn below, distinguishes between homogeneous and heterogeneous networks, with the
latter on balance facilitating innovation diffusion. Henceforth in this paper, we use the
phrase “social diffusion” to refer to innovation diffusion that occurs through social net-
works, broadly defined.
It is the link between social organization and social diffusion, in this instance
with respect to reproductive behavior, that is our primary interest. The aim is not to put
the validity of either research tradition to the test, but rather by joining them to gain
insights about how social organization can affect reproductive behavior. Our analytical
strategy is comparison of a discrete number of communities in southern Ghana, first
characterizing their social organization, then considering how the communities vary in
their receptivity to social diffusion, and finally examining covariation of both with indi-
cators of reproductive behavior (fertility and its proximate determinants). Because fer-
tility is declining in Ghana, with increased practice of modern contraception a major
contributing factor (Ghana Statistical Service and Macro International 1999; Blanc and
Grey 2000), there is good reason to posit that the receptivity of these communities to
innovative reproductive behavior will explain much of the observed cross-community
variation in reproductive behavior. In linking social organization to social diffusion, we
depart from the concerns of Caldwell (1982) and Lesthaeghe (1989), who consider so-
cial organizations in relation to modes of production and reproduction; such an approach
leads to a theoretical framework for reproductive behavior that relies heavily on eco-
nomic arguments. In contrast, we focus on the role of social organizations and other
social structures as channels of social diffusion.
It should be clear from the preceding paragraphs that we take the view that repro-
ductive behavior is socially determined and not simply an attribute of an individual.
Both research traditions—social organization and innovation diffusion—make this as-
sumption, whether or not explicitly stated. This means that group membership can have
considerable bearing on individuals’ fertility decisions. Our approach fits squarely in a
theoretical tradition in sociology that is most prominently represented by Durkheim. In
keeping with the aim of explaining group-level variation, the analysis that follows is
carried out at the community level.
We analyze data collected in six communities in southern Ghana: two in the Greater
Accra Region (Amanfro and Tubaman), two in the Central Region (Brenu-Akyinim and
Frami), and two in the Western region (Abuesi and Komfoeku) (see Figure 1). The six
communities were purposely selected for diversity in ecological setting (coastal versus
inland), primary economic activity (farming, fishing, trading), ethnic composition (Ga-
Adangbe and Akan) and corresponding kinship (patrilineal and matrilineal), and religion
(Christian, Muslim, indigenous). These differences are explored in greater detail below.
Data collection in these sites began in 1994 and has continued to the present. The
analysis here draws on three types of empirical materials gathered in the six communities:
Community portrait. In 1994/95 and again in 1998/99, project investigators devel-
oped in-depth portraits of the study communities, on the basis of extended stays in each
community during which formal and informal interviews were conducted and community
activities were observed. The community portraits focus on political organization, princi-
pal economic activities and sources of income, social and religious organizations, schools
and health services, and fertility-related customs and practices. These portraits achieve
our aim of describing key social organizational features of each community.
Organization inventory. In 2000, an inventory of organizations in each commu-
nity was constructed, through interviews with community members. The intention was
to obtain a comprehensive listing of organizations in each community: names, objec-
tives, number of members, and frequency of meeting.
Panel survey.Women of reproductive age (ages 18–50) and their male partners
in each community were interviewed at length in late 1998 and subsequently have been
reinterviewed at roughly six-month intervals in a “panel survey” design. (This paper
makes use of data from rounds 1–3.) The combined survey sample size for the six com-
Figure 1 Southern Ghana showing the study communities
munities is approximately 1,300 women and 700 male partners. Among the topics in-
quired about are reproductive behavior (including details of the respondent’s reproduc-
tive career and his/her contraceptive attitudes and practices); attitudes and preferences
concerning childbearing and childrearing; socioeconomic characteristics of the respon-
dent and his/her household; knowledge and attitudes about HIV/AIDS; mass media ex-
posure; geographic mobility; and social networks.
As noted above, our analytical approach consists of three stages: a characteriza-
tion of key social organizational features of each community; a consideration of how
these features in turn bear on the receptivity of the communities to social diffusion; and
an examination of differentials across communities in reproductive behavior, in light of
community differences in social organization and receptivity to social diffusion.
The basic features of the six communities are summarized in Table 1. In terms of
ecological setting, four of the communities are inland and two coastal. Fishing is a primary
economic activity in the two coastal communities, although farming figures importantly
in the local economy of Brenu-Akyinim. Judging by the mean number of household pos-
sessions (items such as a radio, television, refrigerator, bed with foam mattress, bicycle,
and so forth—ten items maximum) and the percentage of households that are electrified,
Amanfro and Tubaman are the most prosperous communities, probably because of their
proximity to Accra. The educational attainment of women is highest in Brenu-Akyinim,
whereas the educational attainment of men is higher in Amanfro and Frami.
There are clear differences among the six communities in their religious compo-
sition. Tubaman is almost entirely Muslim, and there is a substantial Muslim minority in
Abuesi (about one-quarter of respondents). The progression of Islam into West Africa
for the most part stopped short of the coast (Islam, of course, is the dominant religion in
many parts of the interior of West Africa), but there are nevertheless some significant
Muslim communities along the coast. Among the other four communities, affiliation
with orthodox Christian denominations predominates in Brenu-Akyinim, while Pente-
costal Christianity predominates in Amanfro. All communities except Tubaman permit
indigenous religious practices in addition to Christianity and Islam.
Finally, ethnolinguistically four of the communities are comprised primarily of
Akan-speaking people, one is predominantly Ga (Amanfro), while the sixth is Adangbe
(Tubaman). The Ga and Adangbe are usually classified together as Ga-Adangbe be-
cause of their very close ethnolinguistic affinity. Ethnolinguistic affiliation essentially
determines the dominant lineage rules in each community, which in turn have repercus-
sions for gender relations and other factors bearing on reproductive behavior (Lesthaeghe
1989). Hence the ethnic composition of the communities is a critical element in the
argument developed below.
Table 1 Basic features of the six communities
Feature Abuesi Amanfro Akyinim Frami Komfoeku Tubaman
Greater Greater
Region Western Accra Central Central Western Accra
Ecology Coastal Inland Coastal Inland Inland Inland
Primary economic Farming
activity Fishing Trading Fishing Farming Farming Farming
Mean number
of household
3.5 4.6 3.4 2.7 3.2 5.0
Percent of households
88 85 71 79 74 82
Percent JSS+
Women 27 45 58 49 30 32
Men 32 79 60 77 62 41
Religious OC = 29% OC = 17% OC = 71% OC = 50% OC = 38% OC = 4%
PC = 37% PC = 68% PC = 19% PC = 41% PC = 46% PC = 6%
M = 24% M = 7% M = 2% M = 0% M = 0% M = 90%
group Akan Ga-Adangbe Akan Akan Akan Ga-Adangbe
Calculated for reproductive-age women interviewed in round 3 of panel survey. Adjusted for community
differences in age and marital status (through poisson and logit regressions).
JSS+ = junior secondary schooling completed. Calculated for reproductive-age women and their male partners
interviewed in round 1 of panel survey. Adjusted for community differences in age and, for women, marital
status (through logit regression).
Percentage distribution of reproductive-age women interviewed in round 1 of panel survey.
OC = Orthodox Christian; PC = Pentecostal Christian; M = Muslim. The residual category consists of syncretic
and traditional groups.
Source: Community portraits; panel survey
Kinship and Gender Relations
In these six communities the family, clan, and lineage are central to the identity
and well-being of the individual. Generally, the four Akan communities follow the matri-
lineal inheritance system, while the two Ga-Adangbe communities follow the patrilin-
eal system. Yet among the Akan, certain positions such as the Asafohene
and the chief
priests of the gods are inherited patrilineally. Within the two Ga-Adangbe communities,
the residents of Tubaman stress the principles of patrikin and male dominance more
highly than the residents of Amanfro.
In the eyes of the people of Tubaman, a child is first of all a lineage member, and
one of the chief purposes of socialization is to produce a loyal and worthy addition to the
lineage. The residents of Tubaman, therefore, practice a system of patriarchy and patrilo-
cality (coresidence of husband and wife after marriage) that ensures male dominance
over females. Islam reinforces the patrilineal strains in the Ga-Adangbe culture (see
Lesthaeghe 1989: 34). In the community portrait of Tubaman, it is reported that deci-
sionmaking about most matters is strictly male-dominated, with the exception of those
areas of exclusive concern to women. (For instance, when the women meet alone, as in
the 31st December Women’s Movement, they are allowed to assume leadership roles.)
During worship, only males are allowed to play leadership roles; women are forbidden
to stand before men and preach. In every home the man is supposed to be the head, with
adult women supporting him as deputies, his sons following in authority, and his daugh-
ters having the least power.
This male dominance in the family and community in Tubaman contrasts sharply
with the position of men in the nearby market town of Amanfro and in the four farming
and fishing Akan-speaking communities. It is often the practice in Amanfro, as in other
Ga-Adangbe communities, that after marriage the husband continues to live in the men’s
compound of his father and the wife in the women’s compound of her mother. Men take
their meals together in the men’s compound, although the food may have come from
wives residing in several women’s compounds. Under this arrangement, neither partner
is under the strict control of the other for most of the day. Not only do these kinship
practices of the people of Amanfro have implications for power and authority in mar-
riages (for example, the degree to which spouses share interests and resources), they
also have implications for social interaction patterns. There is less opportunity for ex-
change of information between spouses and more opportunity for spouses to interact
with other persons. In general, Ghanaians interact most frequently with other residents
of the same compound (this is documented by the panel survey data on social network
partners); hence the traditional Ga-Adangbe residence pattern provides more opportu-
nity for nonspousal social interaction than would be the case were the spouses coresiding.
We return to this point below.
However, there is evidence of change in the customary coresidence patterns in
Amanfro, suggesting that the pattern of separate residence for spouses may have begun
to break down. Today one observes many nuclear-family households composed of a
husband, his wife, and their unmarried children. A related feature of the Amanfro tradi-
tional family system that seems to be changing is the merging of households into a more
inclusive segment. Fortes (1949) termed this the “effective minimal lineage.” In tradi-
tional Ga-Adangbe society, a man had little jural, ritual, and economic authority while
his father was alive, because the household of the man and his children was submerged
in a more inclusive segment with the man’s father as the head. In recent decades, this
full merging of households is rare, although most compounds are still headed by the
fathers and male elders of the patrilineage. Today the authority of the senior male is
more limited in reach, because an increasing fraction of households are organized around
the conjugal pair. In Amanfro as compared to Tubaman, women are more able to exert
authority in their families by virtue of being the main breadwinners of the household. In
contrast to Tubaman, in Amanfro decisionmaking is not monopolized by men, despite
the fact that the people of both communities are patriarchal in their kinship system and
in many practices concerning family matters.
The Akan communities are influenced by an entirely different set of traditional
kinship rules. In Akan society descent is primarily matrilineal, although patrilineal de-
scent is the norm for certain positions, for example the Asafo executioners (members of
the Asafohene) and priests. For this reason, some scholars prefer to classify Akan soci-
ety as based on bilateral descent (e.g., Lesthaeghe 1989: 39). The interplay of these two
systems makes for distinctive spousal roles and decisionmaking authority in Akan com-
munities. Wives are expected to reside with their husbands after marriage. This may be
neolocal residence, but is often virilocal or patrilocal. The residential arrangement and
matrilineal descent combine to balance spousal relations within marriage. The house-
hold organization is centered on the conjugal pair, with the husband normally regarded
as the head. Although in this position the husband has some decisionmaking authority,
he is expected to consult his wife on most family issues, with her views exercising an
important influence on decisions and actions. Hence, while the man is regarded as the
ultimate authority, the norms and rules of Akan society encourage negotiation and bar-
gaining between spouses.
The links between kinship system and postpartum practices should be noted. In
the strict patriarchal system characterizing Tubaman, the new mother does not reside in
her kin’s home during the postpartum period, but rather remains with her husband. In
contrast, in the matrilineal system of the Akans, a woman can return to her parental kin
during the postpartum period, and indeed births may be delivered away from the husband’s
compound. The loose patrilineal system of the Ga-Adangbe in Amanfro does not appear
to include any hard and fast rules about a woman’s residence during pregnancy and the
postpartum period, although by the logic of residential arrangements the new mother
would, more often than not, be found in her kin’s compound.
Other aspects of kinship that bear directly on reproduction do not vary in their
essential features among the six communities. Polygamous marriage is allowed and
practiced in all study communities. In the panel survey data, the proportion of women in
polygynous unions ranges from 41 percent in Tubaman to 13 percent in Frami. The
prevalence of polygamous marriage is somewhat higher in the two Ga-Adangbe com-
munities, which are also somewhat more prosperous, judging by the indicators in Table
1. (An additional factor in the case of Tubaman is that Islamic law permits polygyny.) In
general in these communities the choice of a mate is not a matter for the individual
alone: parents and other senior agnates still exercise some control over matchmaking,
for marriage not only unites the couple, it also establishes a new and important relation-
ship between their respective lineages.
The primary rationale for marriage is the pro-
duction of children, and it is the wish of every member of the family and the descent
group to maintain the strength of the lineage and the clan. Thus infertility in particular
puts women at increased risk of divorce and makes it more likely that her husband will
take another wife. This puts much pressure on women to reproduce, and the community
portraits describe how women consult herbalists, spiritualists, and traditional priests to
assist them in conceiving and bearing children.
From the discussion so far, three distinct types of male–female relationships can
be discerned in the six communities under study. The Tubaman (or Islamized) type of
relationship is characterized by male dominance over females. This relatively extreme
case of patriarchy in West Africa is bolstered by Islamic law. Decisionmaking, both
public and private, lies in the hands of men. Farthest removed from this among the study
communities is the type of male–female relationship found in Amanfro. Although de-
scent is reckoned patrilineally, this is not accompanied by a forceful patriarchy. Women
are not under the strict control of their husbands, and therefore may make independent
decisions. The relative weakness of the conjugal bond is the most salient characteristic
of this system. The third type of relationship is represented by the Akan bilateral system
that is traditional in the other four communities. The Akan relationship may be described
as one of partnership rather than male dominance (Tubaman) or loose relationship
The classification of the six communities in terms of the gender relations that can
be expected to result from the traditional kinship system is as follows:
Community Gender relations
Abuesi Complementary
Amanfro Independence
Brenu-Akyinim Complementary
Frami Complementary
Komfoeku Complementary
Tubaman Male control
Voluntary Organizations
We now investigate the role of voluntary organizations in the six communities, as
measured in the organization inventory. Undoubtedly some of the voluntary organiza-
tions listed in the inventory are actively pro- or antinatalist, and this in itself may influ-
ence members’ reproductive attitudes and behaviors, thereby contributing to differences
among the communities in reproductive behavior. For this reason, we take note of what
the inventory indicates about the fertility-related activities of the organizations, although
we regard this evidence as merely suggestive. Our main purpose here is to assess how
the communities vary in the number of voluntary organizations and their distribution by
type (e.g., religious versus political). Of particular concern is how the voluntary organi-
zations might bear on gender systems, as described in the previous section, and on the
patterns of social interaction that we consider later.
Local-level organizations have captured the attention of many in the field of inter-
national development. Taking off from the concept of “social capital” developed by Bourdieu
and Coleman in the 1980s (Coleman 1988, 1990; Bourdieu and Wacquant 1992), the be-
lief that development initiatives are more likely to succeed in communities that are rich in
voluntary organizations has become popular. The argument is not that voluntary organiza-
tions engage in deliberate activities in support of development initiatives, although indeed
this may occur and contribute to an observed association between community develop-
ment and the prevalence of such organizations. Rather, the hypothesis is that the social
bonds nurtured by voluntary organizations constitute a resource (“social capital”) that
makes community members more receptive to development initiatives and better posi-
tioned to maximize their benefit from such initiatives. This is so for multiple reasons,
according to social capital theorists, including the social cohesion and trust that voluntary
organizations tend to nurture, their capacity to motivate individuals to pursue collective
goals rather than private self-interest, and the channels for social diffusion that these orga-
nizations provide. It is the last mechanism that is of particular relevance to this study. In
testing the social capital argument, empirical studies have related the density of local vol-
untary organizations to the success of development initiatives (e.g., Narayan and Pritchett
1999). The present study adopts the main hypothesis from social capital theory, namely
that a larger number of voluntary organizations in a community should be associated,
ceteris paribus, with more rapid change in health and reproductive behaviors. However,
we are mindful of the opposite argument, that a higher density of organizations can present
an obstacle to change (as pointed out, for example, by Portes and Landolt 1996). Vital and
respected organizations and their leaders can either promote or obstruct the acceptance of
innovative ideas, technologies, and behaviors. By opening or constraining the channels
for communication, they may accelerate or retard the pace of innovation. Because the
social capital effect is not unambiguously unidirectional, it is all the more important to
conduct empirical studies of the association between social capital indicators (such as the
density of voluntary organizations) and community change (in this instance, adoption of
innovative reproductive behaviors).
Our inventory yielded an extensive list of organizations in each community. For
purposes of this paper, these organizations have been classified under seven categories
(see Table 2): cultural, religious, occupational, political, educational (school-related),
philanthropic/welfare, and health/family life education. Several of these major catego-
Table 2 Number of organizations, by community and type of organization
Type of Brenu-
organization Abuesi Amanfro Akyinim Frami Komfoeku Tubaman
Cultural 1————
Christian 10 11 8 8 16
Muslim ————— 3
Farming —3111
Fishing 1 1———
Trading 1————
Others 1 3 1 1
Modern 431223
Traditional 111120
(school-related) 342212
Philanthropic/welfare 2 2 1 2
Health/family life education
Traditional 1
Modern 1—————
Total 23 26 16 17 23 11
Source: Organization inventory.
ries have been further divided. Within the religious category, Christian and Muslim or-
ganizations are distinguished. (Incidentally, no traditional religious organizations were
identified, although traditional religious practices are reported in all communities ex-
cept Tubaman.) The occupational category has been further divided into farming, fish-
ing, trading, and others (hairdressers, drivers, etc.). In addition, modern and traditional
political organizations are distinguished, in recognition of the fact that whereas modern
political organizations have nationwide links and functions, traditional political organi-
zations are local institutions. Finally, a distinction between traditional and modern health
organizations is drawn, on the assumption that their reproductive health values and prac-
tices will often differ, with traditional organizations more likely to be pronatalist and
modern health organizations more likely antinatalist.
Counts of the number of each type of organization, by community, are presented in
Table 2. The number of organizations ranges from a high of 26 in Amanfro to a low of 11
in Tubaman.
Religious organizations represent only about one-quarter of the organiza-
tions reported in Tubaman, but in the other five communities they are by far the most
common type. This should not be interpreted as evidence that religion holds less sway
over individuals’ lives in Tubaman; on the contrary, as noted earlier Tubaman is a Muslim
community, with Islam exerting a dominating influence on values and practices in all
realms of life. Indeed, our community portrait indicates that the people of Tubaman do not
admit of values and practices that are contrary to Islamic law. In short, there is less ratio-
nale for religious organizations per se in Tubaman, because even without such organiza-
tions religion is a pervasive presence in community life. No other community is as reli-
giously homogeneous as Tubaman, as documented in Table 1. In the other five communities,
therefore, a plethora of religious organizations (as many as 16 in Komfoeku) offer com-
munity members a means to observe and celebrate their particular faith with cobelievers.
It is also evident in Table 2 that all the communities except Tubaman have a tradi-
tional political organization as a complement to the chieftancy system and modern politi-
cal organizations (the latter including the unit committee or town development committee,
both common locally throughout Ghana). One traditional political organization, found in
all sites except Tubaman, is the Asafo Company, which acts as both a secular and sacred
organization. Members of the Asafo are traditional village soldiers, guarding and protect-
ing the people both physically and spiritually. The absence of this organization in Tubaman
again reflects the pervasive influence of Islam in that community.
All six communities have educational organizations. The most common are the
parent–teacher association (PTA) and the school management committee. Given our even-
tual aim of explaining variations in reproductive behavior, gender relations are of interest
to us. The information contained in the organization inventory permits classification of the
organizations in terms of the sex composition of their membership. Often this is clear from
the organization’s name: for example, the Jihad Women’s Movement, the Assemblies of
God Women’s Ministry, and the Methodist Women’s Fellowship are exclusively for women,
whereas both men and women participate in the African Faith Tabernacle, the Methodist
Youth Fellowship, and the Catholic Choir. From the membership lists in the organization
inventory, often incomplete, it is also possible to determine whether the membership is of
mixed or single sex. On the basis of this information, Table 3 shows the distribution of
voluntary organizations according to their sex composition.
The majority of organizations are mixed-sex in all communities except Tubaman,
where roughly half (5 out of 11) are mixed. In Abuesi, Amanfro, and Frami, about three-
quarters of the organizations have both men and women as members. Among the single-
sex organizations, female organizations tend to be more common, although once again
Tubaman is the exception.
As indicated above, the organization inventory contains minimal information on
whether or not each organization engages in fertility-related activities. It is plausible
that these organizations exercise influence on the views of their members, and hence
that organizations whose activities, programs, or objectives include reproductive ideas
and behavior would have an impact on the reproductive attitudes and behavior of their
members. This effect might be positive or negative, depending on the messages con-
veyed by the organization. As it happens, organizations reported in the inventory are
primarily antinatalist in their emphasis, which itself may be revealing of a less than
comprehensive measurement of the full range of relevant activities.
For now, we simply report the number of organizations that have fertility-related
programs or objectives. The counts are as follows: Komfoeku, 5 (out of 23); Abuesi, 4
(out of 23); Amanfro, 1 (out of 26); Frami, 1 (out of 17); Brenu-Akyinim, 0 (out of 16);
Table 3 Number of organizations, by community, sex composition,
and type of organization
Type of
Abuesi Amanfro Brenu-Akyinim Frami Komfoeku Tubaman
organization M F X M F X M F X M F X M F X M F X
Cultural –––––1–––––––––––
Christian –37–29–35–26169–––
Muslim –––––––––––––––12
Farming ––––––1–2––11––1––
Fishing 1–––––––1––––––––
Trading ––––1––––––––––––
Others 1––––3–––––11–––––
Modern ––4–12––1––111111
Traditional ––11––1–––111–1–––
(school-related) ––3––4––2––2–11–1
Philanthropic/welfare 1–1––2–––––1––1–1
life education
Traditional ––––––––––1––––––
Modern ––1––––––––––––––
Total 3 3 17 1 4 21 2 3 11 4 13 4 7 12425
“M” denotes male, “F” denotes female, “X” denotes mixed sex.
Source: Organization inventory.
Tubaman, 0 (out of 11). It is, then, a small minority of organizations for which such
activities are reported. Of the four organizations in Abuesi, two are women’s Christian
groups, one is a mixed-sex Christian group, and the fourth is a mixed-sex modern health/
family life education club. All five organizations in Komfoeku are Christian organiza-
tions, with four being women’s groups and the fifth a mixed-sex group. Clearly, the
common belief that religious organizations are less likely to promote antinatalist than
pronatalist attitudes and behaviors is not borne out by the information in the organiza-
tion inventory.
In summary, voluntary organizations are numerous, with only Tubaman lagging
in the density of organizations. Religious organizations are most common, followed by
political organizations; the majority of these organizations have mixed-sex member-
ship; and only a few of the organizations have activities, programs, or objectives that
aim at educating or influencing their members about reproductive behavior.
The second major research tradition informing this paper is the literature on so-
cial diffusion. According to theory and empirical studies, one process through which
social change occurs is the diffusion of ideas, technologies, and behaviors from one
social group to another (Rogers 1983). There are various channels for social diffusion,
including the mass media, agents of organized interventions (school teachers, health
workers, Christian missionaries), and, of most interest in this analysis, informal social
interaction. The hypothesis is that through informal social interaction—with kin, neigh-
bors, friends, workmates—women and men are exposed to new reproductive attitudes
and practices (Montgomery and Casterline 1998; Casterline 2001b). The extent of such
exposure, in turn, is a consequence of individuals’ social interaction patterns, and more
specifically what are commonly termed “social networks.”
In southern Ghana at present, it is plausible that the key issue concerning social
networks, from the standpoint of their efficacy in facilitating (or hindering) the social
diffusion of innovative reproductive attitudes and behaviors, is the extent to which they
expose individuals to persons from a diversity of social groups (as defined by ethnicity,
religion, educational attainment, social class, age, and so forth). Of particular signifi-
cance is whether individuals’ social networks contain persons outside the community,
thereby providing a channel for innovative ideas and behaviors to enter the community.
Following established theory on social diffusion (with roots in Granovetter’s seminal
1973 article), we distinguish “heterogeneous” from “homogeneous” networks, the former
being the more diverse in their membership (in terms of basic social attributes). Homo-
geneous networks are assumed to be more insulated and more likely to support prevail-
ing social norms and practices than to challenge them. In contrast, heterogeneous net-
works are more likely to offer their members an opportunity to consider a wider range of
options for their personal behavior and a reduced pressure to conform to prevailing
community norms. In short, heterogeneous networks are more likely to promote social
diffusion (Watkins 1991; Kohler et al. 2001).
To this basic hypothesis, appealing in its logic and simplicity, must be added an
important qualification (Valente 1995). The hypothesis as stated explains why innovations
might enter some communities more easily, and sooner, than other communities. The in-
troduction of an innovation can be viewed as the first stage of social diffusion, and perhaps
is the primary determinant of the pace of social change. However, the rapidity with which
an innovation spreads within a community is another matter. Within-community diffusion
should be more rapid where the patterns of communication are such that a large fraction of
members soon learn about the innovation, and where social norms can more quickly adapt
to the new ideas, technologies, or behaviors. It is plausible that social interaction patterns
in homogeneous communities are more conducive to within-community social diffusion
(Valente 1995; Casterline 2001a). The more constraining social pressure assumed to char-
acterize homogeneous communities can obstruct social change, but ironically it can also
facilitate such change. A more elaborate hypothesis, then, is that innovations gain a foot-
hold sooner in heterogeneous communities, but once established accumulate adherents
more rapidly in homogeneous communities. If, then, one is examining the prevalence of
innovative attitudes and behaviors—use of modern contraception, for example—it is not
entirely clear whether at any particular date the prevalence should be higher or lower in
heterogeneous as compared to homogeneous communities.
For this analysis, we favor the basic network hypothesis, namely that heteroge-
neous social networks are more conducive to social diffusion. The primary justification
for retaining this hypothesis is that in southern Ghana reproductive decisions are re-
garded as essentially private matters, almost entirely within the province of reproduc-
tive-age couples and their immediate kin. Hence there appears to be less potential for
the positive reinforcement of innovative decisions that homogeneous communities can
more effectively offer. However, social norms and pressures are by no means irrelevant
in this setting, and we will remain mindful of the qualification articulated in the previ-
ous paragraph. This qualification will assist our interpretation of some cross-commu-
nity variation in innovative reproductive behavior that is otherwise puzzling.
The task now is to characterize the six study communities according to patterns
of social interaction, that is, the typical structure of social networks, separately for men
and women. To this end, we consider three clusters of variables: (1) gender systems and
male–female relationships; (2) the nature of economic activity in each of the communi-
ties, on the assumption that economic activity is a primary factor structuring patterns of
social interaction; and (3) individuals’ geographic mobility and social networks.
Beginning with gender systems and their hypothesized effects on network types,
these have far more bearing on the social networks of women than of men, because none
of the systems constrain men to any meaningful extent. Everything else being equal,
women’s social networks should be relatively heterogeneous in communities character-
ized by gender independence, as in the Ga-Adangbe patrilineal system found in Amanfro.
Women in Amanfro need not ask for permission or concurrence of their spouses when
making decisions, even if those decisions have implications for the immediate family.
Less heterogeneous, but still placing only limited constraints on women, is the system
of “complementary” gender relations found in the Akan communities, with their bilat-
eral reckoning of descent. Finally, in Tubaman, with its strongly patriarchal kinship
system and male-controlled gender relations, women are least free to circulate and form
social relationships of their own choosing; hence social networks are expected to be
homogeneous. However, because of the hierarchical decisionmaking power wielded by
men in Tubaman, the qualification to the network hypothesis noted above—that within-
community social diffusion can occur more rapidly in tightly knit communities—may
well apply here. In short, the implications of the gender systems for women’s social
networks are clear: the most heterogeneous networks are in Amanfro, the most homoge-
neous are in Tubaman, and the Akan communities fall somewhere in between.
There is substantial diversity among the six communities in the prevailing types of
economic activities and the extent to which these encourage social mixing. We now de-
scribe the principal economic activities in each community and the implications for social
networks. Because of the predominance of trading in Amanfro, men and women here are
likely to have more contact with persons outside the community than residents of the other
five communities. In particular, women in Amanfro trade at nearby Kasoa and the more
distant Mankesim markets, both of which are on the main Accra–Cape Coast trunk road
and attract people from several regions of Ghana. They also trade in Accra. The daily
commuting of the residents of Amanfro between Kasoa, Mankesim, and Accra exposes
them to diverse groups of people and information. Indeed, the community portrait for
Amanfro notes that traders have received information about modern contraception from
market conversations with other traders and conversations with “self-proclaimed doctors”
and “nurses” who come to the market to sell drugs, including contraceptives.
The residents of Tubaman, too, engage in trade. In contrast to Amanfro, however,
agriculture is the main source of livelihood for households in Tubaman. The norm is for
a husband and wife to farm together, and for the woman to perform the additional task of
selling the produce. Marketing the produce brings the women of Tubaman into contact
with other farmers and residents of the nearby market town of Kasoa, and Accra as well.
For this reason, the extracommunity social exposure of Tubaman women may approach
that of Amanfro women, despite the Islamic norm of social seclusion of women in the
former. If economic activity were the only criterion, then one might posit that social
networks in Tubaman are nearly as heterogeneous as those in Amanfro.
The economic activities of the residents of Frami and Komfoeku demand far less
extracommunity social interaction. As in Tubaman, the main occupation in these com-
munities is farming, but a larger fraction is subsistence farming for household use. Wives
do not travel regularly to markets to sell produce; rather, they process farm produce for
domestic use and for limited sale in the community. Much of this will be traded using
the barter system. There is very little money in the system and, for that matter, there is
very little exchange and contact between the people and the outside world. This system
of economic activity on balance nurtures homogeneous social networks.
The story is much the same for Brenu-Akyinim, although economic activity there
is not as self-contained within the community as is the case in Frami and Komfoeku.
Brenu-Akyinim is a coastal settlement, with farming and salt mining being the leading
economic activities. A minority of Ewe migrants engage in fishing. The farmers engage
in subsistence farming, grow food for home consumption, and also grow cash crops for
marketing. Women engage in food preparation and the sale of salt and pineapple. Re-
cently an effort has been made to promote pineapple production for export. In short,
economic activities in Brenu-Akyinim provide community residents with a limited form
of social interaction with persons from outside the community.
While both Brenu Akyinim and Abuesi are coastal settlements, only in the latter is
fishing the main economic activity. Men catch fish and women are fish smokers and/or
fish mongers. This work brings women into contact with persons from other communities,
especially with women who come to the community to buy fish or with women they meet
in the market in nearby communities or occasionally in more distant cities (Sekondi-Takoradi
and Cape Coast). In addition to the fish industry, Abuesi has been introduced to new eco-
nomic activities and extracommunity social interaction as a result of the construction and
installation of a thermal plant in the adjacent town of Aboadzi. One finds in Abuesi a wide
range of economic activities, including carpentry, blacksmithing, dressmaking, hairdress-
ing, and food preparation. Economic activity in Abuesi, hence, leads to more heteroge-
neous social networks than in the other three Akan communities.
We conclude this section by examining panel survey data on geographic mobility
and social networks (see Table 4). Two indicators of geographic mobility are presented:
the percentage of respondents born in the community and the percentage who visited a
city (Accra, Kumasi, Takoradi, or Cape Coast) at least once per month on average since
the previous interview (six to nine months for most respondents). Roughly half the re-
spondents in the four Akan communities (Abuesi, Brenu-Akyinim, Frami, and Komfoeku)
were born in that community. The fraction is slightly more than one-third in Tubaman,
and only one-eighth in Amanfro. Judging from this indicator, the Akan communities are
far more insulated than the two Ga-Adangbe communities, Amanfro in particular.
Amanfro also is distinctive in the intensity of recent circulation to cities: 80 percent of
Amanfro women visit cities at least once a month, whereas the corresponding figure is
50 percent for Tubaman women and lower yet for women in the four Akan communities
(nearly 50 percent in Frami, which has easy access via a main road to Cape Coast, and as
low as 20 percent in Abuesi, despite its relative proximity to Takoradi). The pattern for
men is essentially the same, with Amanfro men most likely to have visited cities (almost
90 percent) and Abuesi men least likely (about 40 percent).
Table 4 Indicators of geographic mobility, mass media exposure, and social
interaction patterns, by community
Abuesi Amanfro Akyinim Frami Komfoeku Tubaman
Percent born in the
community[R1, W] 50 12 52 51 58 38
Percent who visited cities
once a month or more often
since last interview [R2]
Women 20 80 35 47 40 50
Men 41 89 57 58 77 71
Percent who watched television
twice a week or more often
since last interview [R2]
Women 35 71 42 41 36 37
Men 50 79 46 44 51 53
Percent who had informal
discussions about AIDS
since last interview [R2]
Women 15 66 19 55 16 16
Men 35 83 32 66 35 30
Mean number of network
Discuss important matters
Women 4.1 3.4 3.5 3.5 2.8 4.1
Men 4.8 4.2 4.2 3.9 3.0 4.5
Discuss modern contraception
Women 1.2 2.2 1.4 1.9 0.9 1.7
Men 1.4 2.4 1.7 2.3 1.2 1.5
Discuss HIV/AIDS [R3]
Women 1.5 1.4 0.9 1.6 0.6 1.2
Men 1.9 2.2 1.1 1.7 0.9 1.6
All differentials across communities are adjusted for community differences in age and, for women, marital
status. The adjustments are carried out through fitting logit or poisson regressions. The statistical significance
of variation among communities in each measure has been tested; for all measures shown, the variation among
communities is significant at the .001 level.
R1, R2, R3 denote panel survey rounds 1, 2, and 3. W and M denote samples of women and men.
Accra, Kumasi, Takoradi, Cape Coast.
Source: Panel survey. Sample is limited to women interviewed in round 1. Round 3 estimates, for example, are
based on respondents also interviewed in round 1. The men are partners of women interviewed in round 1.
Frequent television viewing (at least twice per week on average) is substantially
more likely in Amanfro, for both women and men (third panel of Table 4). There is little
variation among the other five communities in this indicator of mass media exposure.
The majority of residents of these communities may have been unaware of HIV/
AIDS when this research project was initiated in the early 1990s. We have no data to
confirm or refute this suspicion, but panel survey data collected in 1998 show that aware-
ness of this disease was almost universal by that date—hardly surprising, given the
proliferation in the mass media (radio, television, billboards) of advertisements warning
of HIV/AIDS and promoting risk-avoidance behaviors. Informal discussion has also
undoubtedly contributed to the spread of awareness about modes of infection, and one
can view the amount of informal discussion about AIDS as an indicator of the extent to
which personal networks are a means for the diffusion of new knowledge and attitudes
about health behaviors (fourth panel of Table 4). Judging from the frequency of discus-
sion about AIDS, once again the conclusion is that Amanfro is most exposed to outside
influences. Frami follows next, with discussion distinctly less common than in Amanfro
but substantially more common than in the other four communities, where discussion is
far less likely. In all six communities, men are far more likely than women to report
having had conversations about AIDS.
Finally, the panel survey contains detailed measurements of social networks (fifth
panel of Table 4). These are conversational networks, defined with reference to different
topics in each of the first three rounds of the panel survey: important matters (round 1),
modern contraception (round 2), and HIV/AIDS (round 3). Respondents in Komfoeku
report the smallest number of conversational partners for all three topics. The communi-
ties that rank high, however, vary according to topic: Abuesi and Tubaman for important
matters, Amanfro for modern contraception, and Amanfro and Abuesi (and, for women,
Frami) for HIV/AIDS. The picture is cloudy, but one might conclude that on average
conversational networks are larger in Amanfro and Abuesi and smaller in Komfoeku.
On the whole these differentials are consistent with other features of the communities,
although it is surprising that conversational networks are not relatively larger in Brenu-
Akyinim, given the patterns of economic activity and the nearby beach resort, quite
popular with Ghanaian and international tourists.
What can we conclude about community differences in typical social networks?
It is difficult to settle on simple characterizations, because the various pieces of infor-
mation examined here imply somewhat different rankings of the communities. More-
over, in some instances these rankings vary by sex. Even allowing for this complexity,
however, the empirical data in hand suggest that social networks are most heteroge-
neous in the market town Amanfro—this is the least ambiguous inference—and on bal-
ance more homogeneous in the farming communities of Frami and Komfoeku. Abuesi,
Brenu-Akyinim, and Tubaman fall somewhere in between. In short, Amanfro stands out
as unique—for women in particular because of their exposure to a diversity of individu-
als in the market and the limited male constraints on their day-to-day mobility.
From this discussion emerges the following classification of the six communities
in terms of the heterogeneity or homogeneity of social networks:
Community Social network
Abuesi Heterogeneous
Amanfro Highly heterogeneous
Brenu-Akyinim Somewhat homogeneous
Frami Homogeneous
Komfoeku Homogeneous
Tubaman Somewhat heterogeneous
We now examine differences among the communities in reproductive behavior,
as calculated from the panel survey (see Table 5). Three aspects of reproduction are
considered: actual fertility, as indicated by the cumulative number of live births; the
main proximate determinants of fertility, specifically nuptiality, postpartum practices
(breastfeeding, abstinence), and contraception; and fertility desires. As we argued in the
first section of this paper, fertility is declining in Ghana, probably indicative of a de-
crease in the number of children desired. The primary direct cause of the fertility decline
to date, in particular the decline during the 1990s, is increased practice of contraception
(Ghana Statistical Service and Macro International 1999; Blanc and Grey 2000).
In thinking about differences in reproductive behavior in the six communities, it
is possible that past differences in marriage and postpartum practices have persisted to
the present and might lead to intercommunity differences in fertility rates. Of more
Table 5 Indicators of reproductive behavior, by community
Variables Abuesi Amanfro Akyinim Frami Komfoeku Tubaman Significance
Children ever born
Women 4.1 3.5 3.9 3.9 4.1 4.2 *
6.5 4.4 5.2 4.6 5.2 6.5 **
Median age at first
Women 19.8 20.2 20.1 20.3 20.4 20.2
Men 22.0 24.8 25.9 25.6 25.6 25.7
percent still breast-
feeding at 24 months
Women 26 20 32 34 31 40 n.s.
Postpartum abstinence:
percent abstaining at
6 months postpartum
Women 67 55 58 36 48 32 **
Percent who want no
more births
Women (at 4
living children) 66 55 68 57 55 41 **
Men (at 5 living
children) 62 44 52 51 56 36 *
Percent currently using
modern method
Women 12 20 16 28 9 22 **
Men 18 28 23 38 19 31 **
Number of cases
Women 232 142 171 190 198 175
Men 153 79 93 109 140 103
*denotes significant at .05 level, **denotes significant at .001 level, n.s. denotes not significant.
Adjusted for community differences in age. The adjustments are carried out through fitting a poisson
Total births fathered, all partners.
Fitted percentages, from logit regressions with living children and living children squared. Average of
percentages for rounds 1–3.
Pill, injection, diaphragm/foam, condom, IUD, ligation, Norplant
Adjusted for community differences in age. The adjustments are carried out through fitting logit regressions.
Average of percentages for rounds 1–3.
Round 1 respondents. Ever-married women; men are current partners of women in sample.
Source: Panel survey (round 1, except where otherwise noted).
interest at this point, however, are differences in fertility desires and in the practice of
contraception (modern methods in particular), because these are indicative of communi-
ties’ participation in Ghana’s fertility transition.
Before examining community differences in fertility desires and contraceptive
practice, we briefly review differentials in fertility, marriage, and postpartum practices
(first four panels of Table 5). The number of children ever born is calculated for both
women and their male partners. Through multiple wives (serially or simultaneously),
men on average have one or two more children than do women. As expected, this sex
difference tends to be larger where polygyny is more common (Tubaman and Abuesi)
and smaller where polygyny is less common (Frami); Amanfro, where polygyny is rela-
tively common, runs counter to this expectation. Nevertheless, the pattern of commu-
nity differences is essentially the same for women and men: fertility is lowest in Amanfro
and highest in Abuesi and Tubaman.
Because the practice of contraception has not been widespread in these com-
munities in the past, the observed differences in fertility can be attributed to differ-
ences in nuptiality and postpartum practices. Interestingly, the timing of entrance to
first marriage hardly differs among the six communities (second panel of Table 5):
women, and especially men, marry earlier in the fishing community of Abuesi, but
residents of the other five communities marry at about the same average age: slightly
over age 20 for women, and around age 25 for men. Hence nuptiality patterns in the
past do not appear to have contributed to community differences in actual fertility.
More intercommunity variation is evident for postpartum practices: breastfeeding (third
panel) and abstinence (fourth panel). The average breastfeeding duration is longest in
Tubaman, followed by the three Akan farming communities of Frami, Brenu-Akyinim,
and Komfoeku, and shortest in Amanfro and Abuesi. Differences in postpartum absti-
nence, which in these communities probably has more impact than breastfeeding on
fertility levels, follow a different pattern. At six months postpartum, roughly two-
thirds of women are still abstaining in Abuesi, nearly 60 percent in Brenu-Akyinim
and Amanfro, roughly one-half in Komfoeku, and only one-third in Frami and Tubaman.
The variation across communities in this indicator is surprisingly large. It is difficult,
on the basis of this quick overview of selected indicators, to determine how the vari-
ous proximate determinants, measured and unmeasured (induced abortion), jointly
account for observed differences among the communities in the average number of
Of more central concern here are community differences in childbearing desires
and contraceptive practice. We have hypothesized that these facets of reproductive be-
havior, which are likely to be especially sensitive to the spread of new ideas and behav-
iors, should be associated with the social organizational features of the communities.
The indicator of fertility desires chosen for this analysis is the percentage of respondents
who wish to terminate childbearing, shown separately for men and women (fifth panel
of Table 5). This desire is evaluated at four living children for women and at five living
children for men. For contraception, the indicator is the percentage using modern meth-
ods, shown separately for men and women (sixth panel of Table 5). Because use of
modern contraception is the clearly innovative behavior in this setting, as well as the
principal impetus for the ongoing fertility transition in Ghana, community differences in
this aspect of reproductive behavior are of particular interest.
The pattern of community differences in the fifth and six panels of Table 5 varies
by sex. Giving more weight here to women, we propose the following ranking of the six
communities according to fertility desires and use of modern contraception:
Desire to Use of modern
Community limit births contraception
Abuesi High Low
Amanfro Moderate Moderate
Brenu-Akyinim High Moderate
Frami Moderate High
Komfoeku Moderate Low
Tubaman Low Moderate
It is surprising that the rankings are so different for fertility desires and contracep-
tive use. Among other things, this may reflect the fact that much of the contraceptive use is
for birth spacing, rather than for terminating childbearing. Even so, it is curious that the
high desire to limit births in Abuesi is accompanied by a low prevalence of modern contra-
ception (suggestive of high unmet need for family planning in that community), and that
contraceptive prevalence is moderately high in Tubaman despite the relatively weak de-
sire to restrict fertility. Contraceptive prevalence is highest in Frami—over one-quarter of
women report that they are currently using a modern method—followed by Tubaman and
then Amanfro. At the other end, the lowest contraceptive prevalence is found in Komfoeku,
followed closely by Abuesi.
We have considered multiple aspects of social organization in southern Ghana
(kinship systems, gender relations, social interaction patterns) and juxtaposed them with
survey estimates of reproductive behaviors, focusing on those behaviors thought to be
driving the ongoing decline in fertility in Ghana: fertility desires and contraceptive use.
As it happens, different aspects of social organization lead to different expectations
about how reproductive behavior should vary among the communities. Adding further
to the challenge of drawing clear conclusions from this analysis, the patterns of repro-
ductive differences vary depending on the reproductive indicator chosen.
Mindful of these complexities, we arrive at the following summary of our sepa-
rate analyses of social organization and reproductive behavior. We show community
differences in patterns of both desired fertility and use of modern contraception, since
the two patterns are markedly different. The fertility indicators are expressed in terms of
their association with the level of fertility; that is, a high prevalence of desires to stop
childbearing and a high prevalence of contraception are both labeled “low” because
they should translate into relatively low levels of fertility.
Expected fertility fertility patterns
levels based on (fertility desires/
Community social organization contraceptive use)
Abuesi Moderate Low/high
Amanfro Low Moderate/moderate
Brenu-Akyinim Moderate Low/moderate
Frami High Moderate/low
Komfoeku High Moderate/high
Tubaman Moderate High/moderate
The levels refer to fertility desires and use of modern contraception, respectively.
They are formulated in terms of effects on fertility levels: “high” refers to low
percentages wanting to stop childbearing or using contraception, whereas “low”
refers to high percentages wanting to stop childbearing or using contraception.
The fit between expected and observed fertility is disappointing. We expected
Amanfro to be most advanced in the fertility transition, as reflected in fertility desires
and contraceptive practice, but instead it falls somewhere in the middle of the six com-
munities. (Interestingly, in Table 5 Amanfro does show the lowest level of actual fertil-
ity, a ranking more consistent with our expectations.) Similarly, Frami and Komfoeku
were expected to be less advanced in the fertility transition, and although Komfoeku
essentially fulfills this expectation, Frami has a high prevalence of use of modern con-
traception. Abuesi and Tubaman further confuse the picture: both were expected to show
moderate levels of fertility, but this expectation is fulfilled only with respect to modern
contraceptive use in Tubaman.
We believe that this unexpected pattern of intercommunity differences in repro-
ductive behavior can be explained, but doing so requires introducing two additional
types of information, namely the actions of community leaders and the provision of
health and family planning services. Neither of these has figured prominently in our
discussion thus far, but they can do much to unravel the mystery of the community
differences in fertility. The major anomalies in the differences are that contraceptive use
is relatively high in relation to our expectations in Frami and Tubaman, and relatively
low in Amanfro. A reading of the community portraits suggests that, for Frami, the
explanation may lie in the greater availability of family planning and health services.
Frami has a well-functioning health clinic and a drugstore that sells medicines, includ-
ing contraceptives. In addition, Frami is visited frequently by staff of the Planned Par-
enthood Association of Ghana (PPAG), as well as government and health personnel
from Cape Coast, only 35 kilometers away on a major road. Finally, community mem-
bers tell of the Frami resident who was sponsored by the village in the early 1990s to
undertake formal training in family planning offered by PPAG. The man in question was
a member of the town development committee. Since then, he has been organizing fam-
ily planning discussions with households in the community. These discussions were
referred to during focus group discussions conducted in the pilot stage of our project. In
short, among the six communities, Frami is perhaps most richly endowed with health
and family planning services, and this might well account for its relatively high preva-
lence of contraceptive use.
The case of Tubaman is also interesting. The notion of practicing family planning
through use of modern contraceptives is relatively new in the community, and is traced
by community residents to a specific development. Contraception had been regarded as
contrary to Islamic teachings. But the son of the present chiefs sister traveled to Saudi
Arabia for his university education, and then returned to Ghana with the news that fam-
ily planning is compatible with Islam. Henceforth family planning could be discussed
openly in Tubaman, and contraceptive practice need not be covert. Given the strongly
hierarchical structure of social life in Tubaman, this sanctioning of family planning by
senior leaders of the community had a larger impact than in most other Ghanaian com-
munities. This fits with our earlier argument that homogeneous communities can be
conducive to rapid diffusion of innovative behaviors, because social pressure can be
exerted more effectively—with the pressure in this instance coming from community
leaders via the strongly hierarchical local power structure. Complementing this factor
has been the fairly effective provision of health and family planning services in Tubaman
by the health nurse, who holds a weekly Friday clinic in the community. In qualitative
interviews, residents tell of women taking their children with them to the Friday clinic,
ostensibly to receive medical care for the children but in fact for the purpose of receiv-
ing a contraceptive injection.
Finally, we consider the case of Amanfro. The relatively low level of contra-
ceptive practice in Amanfro, as compared to our expectations, remains a puzzle. We
note in Table 5 the relative absence of desires to terminate childbearing, suggesting
that children are valued more highly in Amanfro than in some of the other communi-
ties. One can argue that, among the various ethnic groups in Ghana, the Ga are more
pronatalist, with customary practices that encourage women to bear many children
being more strongly maintained than is the case in other ethnic groups. Various initia-
tion rites promote early age at first birth and adolescent pregnancy. While the people
of Amanfro admit the potential economic costs of having a large number of children,
the view persists that large numbers of children are an economic asset in old age. This
attachment to childbearing, despite its recognized economic costs (and despite the
relatively easy and inexpensive availability of contraceptive methods in the commu-
nity), perhaps explains why contraceptive prevalence is lower in Amanfro than we
predicted. This is largely speculation, and clearly further empirical investigation is
required to verify this explanation.
The overarching objective of this research has been to examine the association
between social organization and reproductive behavior. In contrast to most prior re-
search on sub-Saharan African societies with the same objective, we have not focused
on those aspects of social organization thought to account for variation in pretransitional
reproductive regimes. Pretransitional variation in fertility was probably due primarily to
nuptiality and postpartum practices. Rather, our concern lies with how social organiza-
tion affects the diffusion of innovative reproductive ideas and behaviors. Social diffu-
sion is assumed to be strongly affected by the nature of social interaction patterns, and
these in turn are assumed to be determined in part by the social organization of local
communities: gender relations, employment activity, and the prominence of voluntary
organizations. With this theoretical orientation, we are able to develop explicit hypoth-
eses about how the adoption of fertility-control desires and behaviors (e.g., use of mod-
ern contraception) should vary among six communities in southern Ghana.
In the event, the empirical data on fertility obtained in the panel survey does not
clearly confirm the hypotheses. This outcome is a disappointment, but we do not regard
it as a decisive rejection of the social organization and social diffusion arguments. Rather,
we conclude that research which examines a small number of communities must be
sensitive to their particularities and idiosyncrasies. Two of the study communities—
Frami and Tubaman—show relatively high prevalence of contraceptive use, contrary to
expectations, and this can be explained by circumstances specific to the recent history
of each community (the actions of health workers in both communities, the formal dis-
cussions organized by the PPAG-trained man in Frami, and the endorsement of family
planning by the religious leadership in Tubaman). Another community—Amanfro—
shows lower-than-expected prevalence of contraceptive use, and this may be explained
by the continued attachment to pronatalist customary beliefs and practices. A statistical
demographer might view this as a “small-sample” problem: when the sample consists of
merely six communities, historical circumstances specific to one or two communities
can exert a strong influence on the results.
A general conclusion is that an explanatory framework which places social orga-
nization on center stage must be enlarged to incorporate the potentially powerful influ-
ence of community history. In the case of reproductive behavior at times of fertility
transition, the history of health and family planning provision can be especially influen-
tial in determining whether change starts relatively early or relatively late and whether
change proceeds rapidly or slowly. The placement of health and family planning serv-
ices is the outcome of an altogether different set of forces, following a logic that is
primarily political and bureaucratic and may have little to do with community social
organization. Hence when considering the intercommunity pattern of diffusion of new
fertility-regulation practices, one must take account of both local social organization
and the history of service provision. This general conclusion—that explanatory theory
must balance social organization with historical circumstances—emerges from our analy-
sis of variation among small localities. We suspect that the same conclusion applies to
comparative historical research on fertility transition at the national level. Undoubtedly
variables such as mortality decline, schooling, and income are determinants of the tim-
ing and pace of fertility transition; these occupy the place of social organization in this
analysis. But the stance of political leaders and the effectiveness of health and family
planning programs may have substantial bearing on whether transition starts early or
late, and whether it proceeds slowly or rapidly.
This paper draws on materials collected under the project “Diffusion of Fertility
Behavior,” a collaboration between the University of Cape Coast (Ghana) and
the Population Council (United States). The project is supported by awards from
the Rockefeller Foundation and the National Institutes of Health (R01 HD34524).
1 The Asafohene are heads of the traditional army, organized to defend the chiefs
and community members against external aggression.
2Discussion of the age at entry into marriage is deferred until the section on com-
munity variation in reproductive behavior, because the timing of marriage is an
important direct determinant of fertility levels.
3To give some rough notion of the density of organizations, the population sizes
of the six communities according to the census that we carried out in 1998 were
as follows: Abuesi 2,494; Amanfro 858; Brenu 1,211; Frami 1,226; Komfoeku
2,106; Tubaman 1,159. The relatively larger number of organizations in Abuesi
and Komfoeku (Table 2) is to be expected, given their far larger population sizes.
Perhaps most striking is the large number of organizations in Amanfro in relation
to its population size.
Blanc, Ann K. and Steve Grey. 2000. Greater Than Expected Fertility Decline in Ghana:
An Examination of the Evidence. Calverton, MD: Macro International Inc. and
National Population Council Secretariat (Ghana).
Bourdieu, Pierre and Loïc Wacquant. 1992. An Invitation to Reflexive Sociology. Chi-
cago: University of Chicago Press.
Caldwell, John C. 1982. Theory of Fertility Decline. New York: Academic Press.
Casterline, John B. 2001a. “The pace of fertility transition: national patterns in the sec-
ond half of the twentieth century,” in Rodolfo A. Bulatao and John B. Casterline
(eds.), Global Fertility Transition. Supplement to Population and Development
Review 27: 17–52.
———. 2001b. “Diffusion processes and fertility transition: Introduction,” in John B.
Casterline (ed.), Diffusion Processes and Fertility Transition. Washington, DC:
National Academy Press, pp. 1–38.
Coleman, James S. 1988. “Social capital in the creation of human capital,” American
Journal of Sociology 94: S95–S120.
———. 1990. Foundations of Social Theory. Cambridge, MA: Harvard University Press.
Fortes, Meyer. 1949. The Web of Kinship among the Tallensi. London: Oxford Univer-
sity Press.
Ghana Statistical Service (GSS) and Macro International (MI). 1999. 1998 Ghana De-
mographic and Health Survey. Calverton, MD: GSS and MI.
Granovetter, Mark. 1973. “The strength of weak ties,” American Journal of Sociology
78(6): 1360–1380.
Kohler, Hans-Peter, Jere R. Behrman, and Susan Cotts Watkins. 2001. “The density of
social networks and fertility decisions: Evidence from South Nyanza District,
Kenya,” Demography 38(1): 43–58.
Lesthaeghe, Ron J. (ed.). 1989. Reproduction and Social Organization in sub-Saharan
Africa. Berkeley: University of California Press.
Lorimer, Frank. 1954. Culture and Human Fertility—A Study of the Relation of Cultural
Conditions to Fertility in Non-industrial and Transitional Societies. Paris:
Montgomery, Mark R. and John B. Casterline. 1998. “Social networks and the diffusion
of fertility control,” Policy Research Division Working Paper no. 119. New York:
Population Council.
Narayan, Deepa and Lant Pritchett. 1999. “Cents and sociability: Household income
and social capital in rural Tanzania,” Economic Development and Cultural Change
47(4): 871–898.
National Research Council (NRC). 1993. Factors Affecting Contraceptive Use in Sub-
Saharan Africa. Washington, DC: National Academy Press.
Portes, Alejandro and Patricia Landolt. 1996. “The downside of social capital,” Ameri-
can Prospect (May–June): 18–21, 94.
Rogers, Everett M. 1983. Diffusion of Innovation. New York: Free Press.
Valente, Thomas. 1995. Network Models of the Diffusion of Innovations. Cresskill, NJ.:
Hampton Press.
Watkins, Susan Cotts. 1991. “More lessons from the past: Women’s informal networks
and fertility decline,” paper presented at the IUSSP seminar on Fertility in sub-
Saharan Africa, Harare, Zimbabwe, November.
133 Mary Arends-Kuenning and Sajeda
Amin, “The effects of schooling in-
centive programs on household re-
source allocation in Bangladesh.”
134 John Bongaarts and Charles F.
Westoff, “The potential role of con-
traception in reducing abortion.”
135 John B. Casterline and Steven W. Sin-
ding, “Unmet need for family plan-
ning in developing countries and im-
plications for population policy.”
*136 Carol E. Kaufman, Thea de Wet,
and Jonathan Stadler, “Adolescent
pregnancy and parenthood in South
*137 Valerie L. Durrant and Zeba A.
Sathar, “Greater investments in chil-
dren through women’s empower-
ment: A key to demographic change
in Pakistan?”
138 Sajeda Amin, Alaka Malwade Basu,
and Rob Stephenson, “Spatial varia-
tion in contraceptive use in Bangla-
desh: Looking beyond the borders.”
139 Geoffrey McNicoll, “Managing
population–environment systems:
Problems of institutional design.”
140 Barbara S. Mensch, Barbara L. Ibra-
him, Susan M. Lee, and Omaima El-
Gibaly, “Socialization to gender roles
and marriage among Egyptian ado-
141 John Bongaarts and Elof Johansson,
“Future trends in contraception in
the developing world: Prevalence
and method mix.”
*142 Alaka Malwade Basu and Sajeda
Amin, “Some preconditions for fer-
tility decline in Bengal: History, lan-
guage identity, and an openness to
143 Zeba Sathar, Cynthia B. Lloyd, Cem
Mete, and Minhaj ul Haque, “School-
ing opportunities for girls as a stimu-
lus for fertility change in rural Pa-
Recent Back Issues
* No longer available
144 John Bongaarts, “Household size
and composition in the developing
145 John B. Casterline, Zeba A. Sathar,
and Minhaj ul Haque, “Obstacles to
contraceptive use in Pakistan: A
study in Punjab.”
146 Zachary Zimmer, Albert I. Herma-
lin, and Hui-Sheng Lin, “Whose edu-
cation counts? The impact of grown
children’s education on the physi-
cal functioning of their parents in
147 Philomena Nyarko, Brian Pence,
and Cornelius Debpuur, “Immuni-
zation status and child survival in
rural Ghana.”
*148 John Bongaarts and Zachary Zimmer,
Living arrangements of older
adults in the developing world: An
analysis of DHS household surveys.”
149 Markos Ezra, “Ecological degrada-
tion, rural poverty, and migration in
Ethiopia: A contextual analysis.”
150 Cynthia B. Lloyd, Sahar El Tawila,
Wesley H. Clark, and Barbara S.
Mensch, “Determinants of educa-
tional attainment among adoles-
cents in Egypt: Does school quality
make a difference?”
151 Barbara S. Mensch, Paul C. Hewett,
and Annabel Erulkar, “ The report-
ing of sensitive behavior among
adolescents: A methodological ex-
periment in Kenya.”
152 John Bongaarts, “The end of the fer-
tility transition in the developed
153 Mark R. Montgomery, Gebre-
Egziabher Kiros, Dominic Agyeman,
John B. Casterline, Peter Aglobitse,
and Paul Hewett, “Social networks
and contraceptive dynamics in south-
ern Ghana.”
*154 Paul C. Hewett and Mark R. Mont-
gomery, “Poverty and public ser-
vices in developing-country cities.”
* No longer available
155 Zachary Zimmer, Linda G. Martin,
and Ming-Cheng Chang, “Changes
in functional limitations and sur-
vival among the elderly in Taiwan:
1993, 1996, and 1999.”
156 John Bongaarts and Griffith Feeney,
“How long do we live?”
157 Zachary Zimmer and Sovan Kiry
Kim, “Living arrangements and
socio-demographic conditions of
older adults in Cambodia.”
158 Geoffrey McNicoll, “Demographic
factors in East Asian regional inte-
159 Carol E. Kaufman, Shelley Clark,
Ntsiki Manzini, and Julian May,
“How community structures of time
and opportunity shape adolescent
sexual behavior in South Africa.”
160 Julia Dayton and Martha Ainsworth,
“The elderly and AIDS: Coping
strategies and health consequences
in rural Tanzania.”
161 John Bongaarts, “The end of the
fertility transition in the developing
162 Naomi Rutenberg, Carol E. Kauf-
man, Kate Macintyre, Lisanne Brown,
and Ali Karim, “Pregnant or posi-
tive: Adolescent childbearing and
HIV risk in South Africa.”
163 Barbara S. Mensch, Wesley H.
Clark, and Dang Nguyen Anh, “Pre-
marital sex in Vietnam: Is the cur-
rent concern with adolescent repro-
ductive health warranted?”
164 Cynthia B. Lloyd, Cem Mete, and
Zeba A. Sathar, “The effect of gen-
der differences in primary school
access, type, and quality on the de-
cision to enroll in rural Pakistan.”
165 Kelly Hallman, Agnes R. Quisum-
bing, Marie Ruel, and Bénédicte de
la Brière, “Childcare, mothers’ work,
and earnings: Findings from the ur-
ban slums of Guatemala City.”
166 Carol E. Kaufman and Stavros E.
Stavrou, “ ‘Bus fare, please’: The
economics of sex and gifts among
adolescents in urban South Africa.”
167 Dominic K. Agyeman and John B.
Casterline, “ Social organization and
reproductive behavior in southern
Conference Paper
Full-text available
This paper briefly examines evidence on variations in conjugal resources, power, modes of decision-making and domestic task performance among some Ghanaian couples. It speculates about how these dimensions of marital behavior appear to differ widely and to have been changing in Ghana, as a result of social and spatial mobility and individual, gender biased access to jobs and incomes. It uses data from studies carried out at two points in time. The first date back to the sixties and seventies from research carried out by the author. The second set is more recent by the Ghana Statistical Service ( GSS 2004): that is the Ghana Demographic and Health Survey.
Full-text available
Acknowledgments:The approach taken in this paper has drawn on discussions with members,of the NAS Panel on Transitions to Adulthood in Developing Countries. The authors acknowledge helpful comments from Judith Diers, Zeba Sathar, and two anonymous,reviewers. They also acknowledge,the generous support of the Rockefeller Foundation, the Department for International Development, UNICEF, and UNFPA. This material may not be reproduced without written permission from the authors. For a list of Policy Research Division Working Papers, including those available for downloading in PDF format, see © 2004 The Population Council, Inc. ABSTRACT This paper examines ,gender differences in transitions ,to adulthood ,in Pakistan. The analysis is based ,on data ,from the 2001–02 Adolescent and Youth Survey of Pakistan (AYSP), a nationally representative survey of young people aged 15–24 undertaken by the Population Council’s Islamabad office. The ,survey ,covers ,key ,aspects of
We explore young working women's perceptions of marriage and work in contemporary Egypt, when an increase in age at marriage was evident from national survey data. Both working conditions and employment opportunities declined significantly for young women even as their educational attainment increased and marriage was delayed. In-depth interviews were conducted over a 2-year period between 1998 and 2000 with 27 young women between the ages of 15 and 29 who were from relatively poor families and working in a range of salaried jobs in three locations. The qualitative data indicate that young women have high expectations in terms of marital living standards. They seek to achieve this in part by saving intensively before marriage when they work, and otherwise by ensuring substantial monetary support from their families. We conclude that rising material aspirations and family nucleation rather than change in female labor force participation drive marriage change in contemporary Egypt. The driving force behind this conclusion is that there is a reinforcement of the traditional values associated with the institution of marriage rather than its erosion.
Full-text available
Abstract Population policies are deliberately constructed or modified institutional ar- rangements and/or specific programs through which governments influence, directly orindirectly, demographic change. For any given country, the aim of population policy may ,be narrowly ,construed as bringing ,about quantitative changes ,in the membership,of the territorially circumscribed population under the government’s jurisdiction. Governments’ concern with population matters can also extend beyond the borders of their own jurisdictions. Thus, international aspects of population policy have become,increasingly salient in the contemporary,world. Additions to the popu- lation are primarily the result of individual decisions concerning childbearing. Within the constraints of their social milieu, these decisions reflect an implicit calculus by parents about the private costs and benefits of children. But neither costs nor ben- efits of fertility are likely to be,fully internal to the family: they can also impose burdens and advantages on others in the society. Such externalities, negative and positive, represent a legitimate concern for all those affected. The paper briefly discusses how individual and collective interests were reconciled in traditional soci- eties, summarizes the population policy approaches adopted by the classic liberal state, and sketches government responses to the low-fertility demographic regime that emerged,in the West between,the two World Wars. In greater detail it considers international population policies after World War II and contemporary,population
This paper explores the use of an audio computer-as sisted self-interviewing (audio-CASI) methodology i n a household survey of adolescents in two districts of Kenya. Computer software was developed as part of a research project comparing audio-CASI with traditio nal methods of interviewing about sensitive behavio rs, including sexual initiation, risky sexual behavior, coerced sex, and drug and alcohol use. The paper describes the experience of carrying out a househol d-based study using computers and explores the technical challenges faced by the data collection t eams. Few problems emerged with the computer hardware and software, despite the difficult interv iewing conditions. The adolescent respondents easil y adapted to the computerized interview and were able to complete the survey with minimal assistance fro m the interviewing staff. However, the computers were not a completely neutral part of the data collecti on process and added to problems encountered during the fieldwork in one of the districts. Unexpected findings regarding respondents' perceptions of priv acy and confidentiality were also observed.
Full-text available
The detrimental consequences of the AIDS epidemic for older adults in sub-Saharan Africa suggest the need to investigate their characteristics, living situations, and well-being. In this study, we examine the living arrangements of persons aged 60 and older in 16 countries. Data come from the household roster component of recent Demographic and Health Surveys. The focus is on the tendency of the elderly to live with children and grandchildren, and we examine distributions and determinants. Results show that older adults in sub-Saharan Africa live in a variety of household arrangements. Men are more likely to be living in a nuclear household, while women are more likely to be living in extended families. Regressions show that determinants of living with children and grandchildren differ by sex. Taking advantage of survey items on the survival and whereabouts of parents of children, we also examine whether older adults living with grandchildren are involved in fostering or orphaning situations. Those living in countries characterized by high levels of AIDS-related mortality are more likely to be living with grandchildren, with grandchildren but no children in the household, and with orphaned grandchildren. The percent living with one or more double-orphaned grandchildren is strongly associated with AIDS-related mortality, suggesting that the epidemic may be having adverse influences on the living situations of older adults. Knowing about the types of households in which older people live is a first step to understanding their needs in a part of the world with limited resources.
Abstract The interrelationships between ecological degradation, poverty, and rural out- migration in Ethiopia are examined,using data from a Household and Community Survey conducted in 1994‐95. The survey, which covered a sample of 2,000 house- holds, collected retrospective data on changes in household composition, including migration of household members, during the period 1984 to 1994. The study hypothe- sizes that the decision to out-migrate in the impoverished ,rural areas of northern Ethiopia is influenced by a combination of factors based on individual, household, and community,characteristics.A multilevel analysis is applied to determine the role ofthese factors in the decision. The findings show ,that individuals belonging to economically poor households in ecologically vulnerable communities,have a higher propensity to out-migrate for economic reasons, compared with those who belong to wealthier households in ecologically less vulnerable communities. The study pro- vides information relevant to policy formulation in the interrelated areas of environ- mental planning, workers’ mobility, poverty alleviation, and urban development. Ethiopia is a country severely affected by chronic food insecurity and rural pov- erty. With a view to alleviating these problems, the present government, which took power in 1991, has embarked upon a policy that promotes farming and the development of rural infrastructure. In developing countries, government policies dealing with rural development often include migration-related components. In the case of Ethiopia, the government,adopted a national population policy in July 1993. The policy focuses on five major objectives, two of which are directly related to migration: (1) “to reduce the