THE SOCIAL NETWORKS OF IMPOVERISHED
BRAZILIAN WOMEN: WORK PATTERNS AND
HOUSEHOLD STRUCTURE IN AN URBAN
William P. Norris
Working Paper #84
MICHIGAN STATE UNIVERSITY
Copyright 1985, MSU Board of Trustees
THE SOCIAL NETWORKS OF IMPOVERISHED
BRAZILIAN WOMEN: WORK PATTERNS AND
HOUSEHOLD STRUCTURE IN AN URBAN
William P. Norris
Working Paper #84
Abstract: Impoverished women have different networks depending on the
configuration of work patterns, degree of poverty, and household composition.
Drawing on a case study in Salvador, Brazil, network origins and effects are
delineated and related to the larger socio-cultural context. Self-employed
women who work at home and housewives, and single women and couples are
compared, based on a two level measure of poverty. Self-employed women in
couples are found disproportionately below the poverty level, and have more
heterogeneous and complex networks than single female heads of household or
housewives. The cumulative effects of self-employed work and living in a
couple appear to explain these results. Different types are related to more
egalitarian couple relationships, neighborhood participation and location in
politically important neighborhood communication networks.
About the Author: William P. Norris is Associate Professor of Sociology in
the Sociology-Anthropology Department, Oberlin College. This article is part
of a project on household composition, networks and survival strategies on
which he published two articles last year, "Patron-Client Relationships in the
Urban Social Structure: A Brazilian Case Study," Human Organization 43,
1:16-26, and "Coping with Poverty in Urban Brazil: The Contributions of
Patron-Client Relationships," Sociological Focus 17, 4:259-273. His current
research focuses on the changing relationships among households, neighborhood
associations, social service agencies and the state as the basis for
understanding Brazilian urbanization.
THE SOCIAL NETWORKS OF IMPOVERISHED BRAZILIAN WOMEN: WORK PATTERNS AND
HOUSEHOLD STRUCTURE IN AN URBAN SQUATTER SETTLEMENT1
Impoverished women in the cities of Latin America face daily survival
problems. As members of households in which stable jobs or even minimally
adequate incomes are rare, they rely on two means of survival--earning incomes
and using personal relationships. Studies (see Lomnitz 1977) have shown how
these ties with others, conceptualized as social networks, can provide food,
money, information and contacts, but the relationship between these various
network forms and household survival has not been examined. This paper
focuses on the origins of women's networks and their effects on household
survival and other aspects of women's lives.
The key factor shaping the networks of women in this study was their work
patterns; the women in this study are either housewives or self-employed women
working at home. Household structure was also a factor with female-headed
households being differentiated from couple-headed households. Effects of
differences in the forms of women's networks extended beyond survival
abilities to include variation in the equality of relationships with domestic
partners, the control of resources, and their contribution to the neighborhood
The women were imbedded in larger neighborhood and urban/regional social
structures. These larger contexts are incorporated in the study by discussing
the level of poverty of the women's households, analyzing certain important
attitudes, and providing information about the Brazilian region and the
squatter settlement. The level of poverty and the attitudes directly and
indirectly affect network formation.
This study was conducted in a squatter settlement, the Alagodos2, in
Salvador, the capital of the state of Bahia. Located on the northeast coast
with a population of 1.5 million, it is the fifth largest city in Brazil and
the fifteenth largest in Latin America. Now designated one of Brazil's
"development poles," it has had an uneven process of development characterized
by an earlier period of industralization, then stagnation and, recently,
another spurt of industrialization based on oil and petrochemicals. During
this last phase, Salvador has been a magnet for hinterland labor and national
and international capital, with concomitant expansion of demand for
construction workers and people in personal and commercial services (Jelin
1974; Faria 1980; Singer 1980). This expansion has not been accompanied by
much increased demand for industrial workers (Lewin et al., 1977:120). During
this period, real wages in Salvador have declined which means that more
members of a household must earn an income (Carvalho e Souza 1980:84). The
result of this situation for low-income women is a pattern of economic
activities in some ways reminiscent of 19th century Sao Paulo (Kuznesof 1980:
104) or Europe (Tilly and Scott 1978:125). Self-employment is very common.
Prandi (1978:107) found that in Salvador, 23.6% of women over age 14 in the
labor force were self-employed. Women who are not employed work as unpaid
domestic workers in their own households or as self-employed laundresses,
seamstresses, hairdressers, street vendors or small shopkeepers.
The site of the study, their Alagados squatter settlement, is representative
of a characteristic aspect of Latin American urbanization. Squatter settle-
ments develop when people illegally occupy land to build housing. A large
proportion of the population of the largest cities lives in them; for example,
between one-quarter and one-half of the populations of Lima, Mexico City, Rio
de Janeiro and Caracas live lin such settlements. The squatter population
alone of Mexico City would constitute one of the largest cities in Latin
America. Although squatter settlements have long been identified with
shantytowns and poverty, many houses and residents do not fit these stereo-
types--some are even middle class.3 The population of the Alagados was
estimated at 85,000 in 1970 (IURAM 1970) and now substantially over 100,000.
I have argued elsewhere that the larger and older settlements, such as the
Alagados, should no longer be! considered as socially, economically or
culturally separate from the cities of which they are a geographical part (see
also da Silva 1979; cf. Dietz'  for political differences between
Scholars have recently expanded our understanding of the situation of
low-income Latin American, particularly Brazilian, women. Women constitute a
significant proportion of the low-income or marginal workers in Brazil and
Latin America. They play key roles in the social structure of dependent
development because they facilitate the maintenance of very low wages by such
activities as building their own! housing and providing cheap commodities and
services and by their potential as an industrial reserve (Saffioti 1976, 1978;
Faria 1976; Kowarick 1977; Long and Richardson 1978; Arizpe 1977; Jelin
1976). Historically, the labor force participation of Brazilian women has
decreased as capitalist forms of work organization have expanded (Madeira and
Singer 1975; Saffioti 1978, Miranda 1977). Regional differences also shape
their participation (Martine e Peliano 1975:163). Sao Paulo, Rio de Janeiro,
and Belo Horizonte, the three corners of what has been called the "industrial
triangle" have labor markets and income distribution which differ from much of
the rest of the country, especially the Northeast.
The impoverished women face conflicting cultural demands and expect-
ations. Their dilemmas are reflected in contradictory portrayals of them as
powerful (Woortmann 1975) or weak (Saffioti 1976). When a woman earns an
income it may raise questions about the ability of her domestic partner to
support the household (see Aguiar  for rural situation). In a Sao Paulo
state study, more than half of those low-income women working for an income
reported that they would prefer to be housewives (Saffioti 1976). Working for
pay also conflicts with their responsibilities for the maintenance of the
household (Bilac 1978:54-4; da Silva 1979). At the same time, childcare
responsibility ineluctably leads to earning an income.
Cultural attitudes support women's participation in religious activities.
Affiliation with religious groups, in particular the Protestant sects, has
been found to be both a means of overcoming lack of local ties (Roberts 1973:
187-92) and a kind of survival support (Norris 1984). Epstein (1973:136)
suggests that the practice of charity among the Protestant churches attracts
Household composition and structure are related to the labor force
participation of low income women in Brazil.5 Saffioti (1976) found that
increasing poverty was related to more women working for wages and that single
women who were heads of households all worked. In a study of another part of
the Alagados, Woortmann (1975) found that 67% of the women residents earned an
income of some sort. Children are an important determinant of whether a woman
works for wages (da Silva 1979; Bilac 1978); Saffioti (1976) found that women
with young children did not work as often as did those with older children.
Studies of social networks have found them to be more restricted than
those of other classes because of the lack of trust (Roberts 1970, 1973), the
attempt to avoid problems (Eckstein 1977:75), or the scarcity of resources
(Banck 1980); however, Nieves (1979) showed that low-income women have wide
networks. The most extensive application of network analysis to low-income
people in Latin America is the study by Lomnitz (1977) of survival networks
based mainly, but not exclusively, on kinship in an impoverished neighborhood
in Mexico City (see also Lewis 1959; Lobo 1982, Mangin 1970; Peattie 1968;
Woortmann 1975). The important part played by women in neighborhood
communications networks has long been recognized by scholars (Velez-Ibanez
1983) and neighborhood activists.
The following questions shaped this exploratory research:
1. What are the effects on women's networks of different work
patterns, household composition, attitudes, and levels of
2. If there are different networks, how do they enhance or
lessen the survival of a household?
3. What are the other effects of the network differences, either
within the household or the neighborhood?
The unit of analysis is the household, which includes everyone sharing
living space except for tenants. The household may be composed of nuclear
families or an extended kin group. Female-headed nuclear family households
are distinguished from those with couple structures.
The socioeconomic level of the households is determined by total monthly
household income which is calculated by summing the shared quantifiable
monthly income of all household members. The household incomes are divided
into two levels--above US$80 and below US$80 per month--based on the finding
that this amount was necessary to feed all household members for an entire
month (see Norris 1984 for further discussion). Household incomes below
US$80 are grouped in the category "below poverty level" and those above are
referred to as "above poverty level."6
A social network has been defined as a "specific set of linkages among a
defined set of persons" (Mitchell 1969:2). This analysis is concerned with
three aspects of networks: the! nature of specific ties; the number of
people at the center of the network; and the form of the network.7 The
nature of specific ties refers to whether a relationship is based on one or
more role relationships (see Mitchell 1969; Fischer et al. 1977). The
number of people at the center of a household network may be one (the female
head) or two (the couple). Having two people at the center of a network
creates a problem for the characterization of the form of the network. In
this case, I suggest three forms. A network may be interlocking (cf.
Laumann 1973), which means that ties exist between all members, whether or
not it is a one- or two-person center. The third form is the dyad, which is
a tie between one person in the center and only one person in their
network. Interlocking and segmental networks may also have dyads in them.
Two types of work patterns are considered. One is the unpaid domestic
labor of the housewife. The other is the self-employed woman who works in
her own home.
The culturally-based expectations of women's behavior considered here
are attitudes towards earning an income, responsibility for childcare and
home maintenance, relationships with non-kin adult males, and engaging in
This information was collected in 1973, based on systematic observation,
formal and informal interviews, !attendance atj some neighborhood religious
and political association meetings, and secondary sources. I chose this
approach because of the cross-cultural and cross-class nature of the study
and issues of validity (see Valentine  and Warwick and Osherson 
for discussion). Interviews were conducted with 20% of the households,
randomly selected, in one neighborhood of the Alagados. The total sample is
65 households of which 62 have adult women in them. The group of income-
earning women includes four who either have pensions from previous work or
are looking for work.
One-third of the households were extended families and two-thirds were
nuclear families; only two cases included unrelated children. One out of
three of the nuclear family households was female-headed (or 1/6 of the
total sample). All but one of the extended family households were based on
couples.8 All but one female-headed household and two of the nuclear
family households contained children.
Median household income was about US$60 per month, which means that a
large proportion of the sample was below the poverty level. The relation-
ship between household composition and poverty level is complicated by the
two structures possible: female-headed household and couple-headed house-
hold. In Table 1, household Icomposition and income level are cross-
tabulated and the female-headed household is treated as a separate category.
If nuclear family households are considered as a whole, 80% are below the
poverty level. If, however, the female-headed households, all of which are
below the poverty level, are separated out, 73% of the remaining
couple-headed households are below the poverty level. When further
differences between all nuclear family households and extended family house-
holds were investigated, only one difference was discovered. Nuclear family
households below the poverty line tend to receive more aid from others.
Based on this lack of differences, the rest of the discussion is devoted to
differentiating the effects of living in female-headed and couple-centered
households. This is referred to as household structure.
Female-headed households were among the poorest in the sample. The
female heads were also among the oldest women. Many were widows, fewer had
been divorced or abandoned. One effect of age is that relationships with
relatives tend to be with children rather than with siblings or parents. It
was difficult for the children to give aid and what they could give depended
in part on their ages. The different access to resources of men and women
meant that women without men had survival problems.
Women in households with incomes below the poverty level are more likely
to work at income-generating activities. These households also engage in
other activities to supplement their low incomes such as seeking help from
others or from charities and forms of what has been called urban hunting and
gathering. The latter includes catching shellfish, fishing, and collecting
recyclable materials. Some also raise chickens and pigs.
Women below the poverty level received more help, in the form of food
and/or money, from people in their networks than did women above the poverty
line. Those who provided help were usually either patrons or relatives (see
Table 2). About one-sixth of the households received help from patron and
The level of poverty also affected the geographical extent of household
networks. Low income limited physical mobility to the point that impover-
ished women reported that they could not maintain ties with friends and
relatives who lived in other parts of Salvador.
Patterns of work varied among the women. The woman who worked for an
income typically washed clothes, fixed hair, sewed clothes or sold produce
or household goods. In all cases in this study, the women used their homes
as the bases of work. The income-earning women also had to cope with the
demands of maintaining a house and caring for children. As would be
expected, they had a variety of relationships with others because of their
work activities; relationships with customers and suppliers were parti-
cularly important. Some of the women classified as working in commerce for
the purposes of this study also wash clothes to earn some income. Thus,
about two-thirds of the income-earning women wash clothes to earn some
income. Washerwomen who have domestic partners have networks similar to the
other income-earning women. Contrary to expectations, networks of income-
earning women extended beyond customers. Their work took them out into the
neighborhood or brought the neighborhood into their houses. A washerwoman,
for example, not only picked up and delivered 'clothing, she also dried and
sometimes washed it outside. The owner of the small store had a constant
stream of neighbors purchasing small items. The seamstress and hairdresser
served a more specific clientele; however, while working, they also engaged
in conversation and gossip with neighbors as well as customers. The combin-
ation of ties with relatives, customers, neighbors, and others resulted in
heterogeneous and complex networks.
A pattern noted among some income-earning women in couple-headed house-
holds was that these women appeared to have more egalitarian relationships
with their domestic partners than did the housewives. This was based on
reported decision-making participation, and this information was not
collected from all.
Housewives appeared to have work routines similar to the income-earning
women; however, these were important differences. The housewife worked with
household members, i.e., the domestic partner, children and other relatives,
and less importantly, with the people from whom she purchased commodities or
services. To prepare meals, to Icope with the lack of a refrigerator, and,
by necessity, to make purchases in small units, she had to make frequent
trips to neighborhood stores. Even so, her relationships with people
outside the household were predominately with relatives and were shared with
her domestic partner. Other relationships were dyadic and included co-
religionaries as an important group. For women below the poverty level,
there are often patron-client relationships, held by former domestic
servants who were able to call on their former employers for periodic aid.
Patterns of work and household composition were related. This was most
evident in the case of female heads of households, many of whom earned an
income as washerwomen; two-thirds of the female heads of households in this
sample washed clothes and one-half of the washerwomen were female heads of
households. Several related processes contributed to the prevalence of this
kind of work. Some washerwomen were former, domestic servants. Having
children meant that they could' no longer continue in domestic service.
Washing clothes provided a way for them to combine childcare with work with
which they were familiar. Furthermore, they already had some contacts with
potential customers. The start-up costs for washing are low--a tub, soap,
an iron, and space. (Domestic" servants themselves were a heterogeneous
group. One segment consisted of young girls who were brought into the
metropolitan area from the interior in order to be domestics. They had few
relatives in the area and so were isolated and dependent. These in parti-
cular appeared to become washerwomen.)
Women working for an income received more aid than housewives; thus,
patterns of working and support patterns were related. Working for an
income, however, was not related to either age or number of children. There
was no difference between income-earning women and housewives in age and
number of children except that the youngest women with 1-2 children were
somewhat more likely to be housewives.
Attitudes about women's behavior, especially about earning an income and
taking care of the children and the house, were strongly held. Men, in
general, expressed opposition to women working for an income and stated that
the woman belonged in the home, taking care of the children and preparing
meals. As one man said, "I want the soup on the table when I get home."
The household income was also viewed as the man's responsibility. Yet, half
of the women with domestic partners worked for an income. Men and women
explained this normative discrepancy by referring to household survival
problems. The problem of coping with these attitudes emerged most acutely
when women who had been housewives were faced with the necessity of earning
an income. Women with special skills and those who had always earned an
income felt less necessity to explain themselves. Housewives below the
poverty level, on the other hand, justified not earning an income by
referring to their domestic partners' opposition.
Relationships with men who were not kin were potential problems because
of the appearance of sexual relationships. This possibility could be the
subject of extended neighborhood commentary. Balancing the need for support
from males with the desire to maintain the socially sanctioned distance
required great subtlety.
Religion was another aspect of culture that affected networks.
Specifically, women were expected to be more religiously active than men.
While proportionately more income-earning women attended meetings, house-
wives were also active. The four sects or churches with the most parti-
cipants were Umbanda9, Spiritism10, Jehovah's Witnesses, and Catholicism
(indicated by regular attendance at Mass). Each had four members. It
should be noted that both the Witnesses and the Spiritists emphasized
education and attitudes conducive to social mobility. Umbanda, while
decidedly a lower-class sect in Salvador, also represented a synthetic,
dynamic religious movement with an emphasis on female participation as
mediums. Poverty also affected religious participation; some women told me
they were first attracted to certain sects because their members had helped
them when they were desperate.
Three types of women's network patterns emerge from these findings: 1)
the income-earning woman who is part of a couple; 2) the income-earning
female head of household; and 3) the housewife. The separate effects of
patterns of work, household structure, degree of poverty and certain
attitudes have been outlined. The interrelationships of these elements are
evident in the network analysis. The analysis focuses on the nature of the
ties (unidimensional and multidimensional), the number of people in the
center (individual or couple), and the forms of the networks (interlocking,
segmental, dyadic). The female heads of households (11 cases) are combined
with the income-earning category. Differences are noted in the text.
Interlocking networks for couples are much more common among housewives
than among income-earning women (see Table 3). Most of the housewives'
shared ties are with relatives. The income-earning woman, on the other
hand, shares proportionately fewer ties with relatives and more with other
people such as neighbors and co-workers. No female-heads are counted as
having interlocking networks here. One meaning of the interlocking ties is
that resources entering the household over these ties will tend to be shared.
Network segments that have only the woman at the center reveal a
different pattern (see Table 4). Many housewives did not mention groups of
people with whom they had relationships independent of their domestic
partners. Income-earning women, including female heads of households,
mentioned many groups of people they knew. For!both kinds of women, network
segments consisted of neighbors/friends/colleagues or co-religionaries, that
is, multidimensional relationships. Resources entered the household through
the woman, not shared in a couple. Note that, while working for an income
leads to more relationships, it does not result in many relationships with
people in the same work (signified by colleague, above).
Female heads of households reported relationships equally as hetero-
geneous as other income-earning women, but they reported fewer of them.
These women also did not report many relationships with men. Subtracting
from the relationships of other income-earning Iwomen male relationships and
dividing shared relationships does not account for this difference. The
networks of female heads of households are, thus, between those of income-
earning women and the housewives in size. The female head of household
appears to gain the network benefits of work but lose some ties because of
the absence of a male. The activities of these women may be limited by sex-
role prescriptions or by the lack of access afforded by another close tie.
The existence of dyads in these networks is the clearest difference
between the women's networks. While more than half of the housewives report
no ties, almost 3/4 of the income-earners state that they have dyadic ties
with customers, patrons, and colleagues. The latter category is distinct
from neighbors (see Table 5). The majority of dyads are in fact with
customers. These ties tend to be unidimensional, with customers frequently
residing elsewhere and being a different class.| Other ties that appear here
are those with co-religionaries and patrons. Relationships with patrons
were much more frequent among the income-earning women and usually grew out
of past employment as a domestic servant or out of a washerwoman's relation-
ship with her customer. These were multidimensional ties. There were only
a few active godparent relationships, and only two provided any help. Dyads
are exclusive sources of resources, not shared at all.
The variety and complexity of the network patterns may be clarified by
the following examples. Gildete was a self-employed seamstress. She and
her domestic partner had lived together for 13 years and had 6 children.
She was 31 and he 36. He worked as a delivery man for Petrobras. Their
household income included his wage, her earnings, and rent from a small room
and totaled about US$76 per month. She had done piecework for a company,
but the pay became so bad that she decided to work on her own. She had her
sewing machine in the front room and worked surrounded by her children. Her
relationships with her customers' took the form of dyadic ties. She had a
circle of her own female friends in the neighborhood, i.e., a network
segment focused on her, and shared some neighbor-friends with her domestic
partner. A relationship with a doctor had become a patron-client tie where
she received medicines and other help. Because Gildete had been ill, the
oldest daughter, 11 years old, had been doing much of the childcare, and had
to miss school because of this.
An example of the network of a housewife was that of Giselia and her
husband. Married for nine years, they moved to Salvador eight years ago to
join his brother in the Alagados. They constructed a house on recently
filled land using discarded materials from the construction of an apartment
house. She was 29 and he was 35. They had six children, the oldest of whom
was 7. Her husband worked in construction, specializing in concrete
reinforcing work. His wage, plus a family allowance, and rent from an
adjacent room, resulted in a household income of US$92 per month. This
allowed them to have simple food, including some meat, for the entire month
and to make a few purchases of clothing and household items. Her network
had a large joint segment and several dyads. Relationships that she shared
with her husband were all with kin. Her dyadic relationships were with
members of the Spiritist sect. She attended weekly meetings and counted two
of the women members among her friends.
Joselina, a single female head of a household washed clothes for a
living. She was 45 years old. Her former domestic partner had returned to
his home town in the interior because of illness, and she had not seen him
since that time. She had no relatives in Salvador except her 15 year old
son. Her household income was based on clothes washing and the occasional
sale of one of the songbirds or chickens she raised. She shared her small
house with the chickens. Their eggs were a staple of the household diet.
Her house was long and narrow because she had divided the old house and sold
half of it. Her relationships were with women of the neighborhood and some
of their domestic partners. An important dyad was her relationship with a
customer who was also a patron. The patron helped her with food when she
lacked money. Joselina made a point of specifying that she would not let
her son work yet, except in part-time jobs. He was a student, and his
education was of primary importance to her.
To summarize the analysis, sytematic differences were found in three
kinds of networks. Networks of the women who did not have domestic partners
were, by definition, exclusively focused on those women. Among the women
with domestic partners, housewives shared more of their relationships with
their domestic partners than did the self-employed women. There were also
differences in network segments; income-earning women had a greater number
of such segments and they also contained more relationships than did those
of the housewives. More dyads were reported by the self-employed woman.
The nature of the ties also varied. The income-earner had ties with
customers, former customers, and vendors that the housewife did not have.
The income-earning women had more multidimensional relationships than the
housewives. The network segments of the working women were predominately
the role relationships of neighbor and friend. Some also included women in
the same line of work in their networks, e.g. other washwomen or seam-
stresses. Religion was the basis for the formation of some relationships.
Both groups contained women who participated in one of several religious
sects and had friends among them, but the income-earning women had more such
relationships. The nature of the tie differed by income level because
impoverished working women had a greater tendency to use their relationships
as part of their household survival strategy. Patron-client relationships
were found more frequently among the poorest.
The networks of these impoverished women resulted from a complex
interweaving of responses to poverty, attitudes, work patterns, and house-
hold composition. Before essaying an explanation of their particular forms
of networks, several aspects of the findings require discussion.
Patterns of work shaped networks in unanticipated ways. Rather than the
expected expansion of a network segment to encompass relationships acquired
as part of earning an income| (such as relationships with customers,
suppliers and women in similar work), there was a general expansion and
increased heterogeneity. Furthermore, earning an income did not have the
same effects for women in couples and female heads of households. Women in
couples reported more relationships in the neighborhood as well as with
relatives. They also had more relationships based on religious ties. The
female heads of households had a similar heterogeneity, but their networks
were not as extensive, and there were fewer men in them.
The forms of the networks also differed. Income-earning women shared
interlocking networks with their domestic partners less frequently than
housewives. Rather than networks composed largely of relatives, as in the
case of the housewives, the income-earning women reported ties to neighbors,
colleagues and customers as well. Since the same person often had more than
one role, the result was increased multidimensionally in the networks of the
income-earning woman. The dyads were particularly noteworthy because,
whereas the housewives reported very few, the income-earning women had many
such ties. The multidimensional dyads were usually the ones providing
support. Thus, the expansion and heterogeneity of networks by the
income-earning women resulted in increased access to resources. As
indicated above,! the female heads of households experienced the same
changes, but to a less marked degree.
Household structure shaped women's networks through the presence or
absence of a male domestic partner. If only income-earning women are
considered, single women had less extensive networks than women in couples.
Two other aspects of household! structure, the presence of children and
extended families, were much less important. Children were present in most
households so this became something of a constant. Children's age and the
number of them did not seem to have an effect on whether or not the mother
earned an income. The age of children may be important economically because
older children can earn an income also.11
The degree of poverty was found to have direct and indirect effects on
women's networks. Impoverished women turned to others for help, in
particular patrons. This resulted in expanded, more heterogeneous net-
works. The patron-client ties, formed between people of unequal socio-
economic status, were specialized, discrete dyads. They provided survival
resources for clients and some labor for patrons. Poverty had indirect
effects on networks because it appeared to encourage earning an income. It
might also have led to joining religious groups.
Certain attitudes appeared to affect networks indirectly by supporting
some activities, such as taking care of home and children, and negatively
sanctioning others, such as earning an income. Relationships with non-kin
adult men were generally not approved. All women found direct normative
support for religious activities.
The explanation of network differences hinges on the expansion of
activities of the women who attempted to cope with the impoverished state of
their households by earning incomes. Very low household income compelled
women to seek ways to earn money. Women's responsibility for childcare and
home maintenance led to work activities that could be accomplished at home,
e.g. small-scale commerce or personal services. Work-oriented relationships
were formed with customers, suppliers and sometimes with people in the same
line of work. Furthermore, the search for customers and information led to
increased contact with people of the neighborhood and the city in general.
More than one role relationship was formed with some. The result was an
expanded and heterogeneous network. Housewives also worked at home caring
for the children and house, but they did not have the other work that
required them to meet neighborhood people. The result was that housewives
tended to have networks focused on relatives and shared with domestic
partners networks more restricted and homogeneous than those of women who
Some housewives expanded their networks through participation in
religious activities for which there was cultural support. Even so,
proportionately more income-earning women had religious ties. Women who
earned an income and were single heads of households had networks that were
as heterogeneous but smaller than the networks of women in couples who
earned an income. This difference in household structure affected networks
because single women could not form relationships with men as easily as
income-earning women who were part of couples. Female heads of households
were restricted from including men in their networks because of the
potential for compromising one's reputation, sexual exploitation, the
opposition of other women, or the refusal of men to get involved with older
women with children.
Some network patterns had more potential for survival than others. It
seems clear that the female head of household was in the most precarious
situation, reflected by their households being among the poorest. To
distinguish between the two kinds of couple-centered networks in terms of
survival ability, however, is not possible with the data available. Both
survive, albeit poorly. It is apparent that if the addition of the woman's
earnings to household income does not raise the household income above the
poverty line, then the contribution of her domestic partner is indeed
paltry. We may conclude that, based on this evidence, the income-earning
woman in a couple and the housewife have different patterns that organize
survival and look for their effects in other ways.
One result of this network difference may be a more egalitarian
relationship between some of the income-earning women and their domestic
partners than between housewives and their partners. The facts that the
woman controls important resources and that she does so through relation-
ships that are independent of the domestic partner may contribute to this.
Not only does the woman have an income that she can contribute to the
household, but she also has access to food and money through others and to
information about jobs and political activities in the neighborhood.
Also important is the location of the self-employed women in important
junctures in the patterns of flows of information in the Alagados. Some
women, particularly the seamstresses and the women engaged in commerce, have
access to information about other areas of the city that other women do not
have. The washerwomen, by virtue of their ties with women of higher status,
have access to information about jobs and opportunities that are crucial for
others in their networks. In addition, all of the income-earning women, by
virtue of their network ties with others, link networks across the Alagados,
thus enhancing communication. The patron-client relationship is parti-
cularly important because the patron usually lives outside the Alagados and
is of a different social class. Consequently, income-earning women have
access to independent information. They are also important as people who
mobilize opinion around some issues. This is not an absolute distinction;
some housewives also have ties with others that provide information and,
thus, increase their independence. Working for an income appears to lead to
a more heterogeneous set of relationships than those held by the housewife.
In fact, income-earning is related to expanded network ties in all social
arenas. The self-employed woman's work involves her in the neighborhood and
city. The result is extensive and heterogeneous networks of informal
relationships and association membership.
Summary and Conclusions
In the Alagados, women'Is networks were shaped by different work
patterns, household structures, levels of poverty, and certain attitudes.
Women who earned an income as self-employed workers in their own homes had
more complex and heterogeneous networks than housewives. While housewives
tended to share their network relations with ,their domestic partners, the
women earning an income had more independent relations with others. The
nature of the relationships was also different. Housewives had ties
predominately with relatives, while women earning an income had ties not
only with relatives but also with neighbors and others. The difference was
particularly marked when the !two groups were compared in terms of dyads.
Housewives reported few dyads whereas the income-earning women reported many
dyadic ties with their customers, patrons, and others.
Household structure affected networks principally in that networks of
income-earning women in couples differed from those of female heads of
households. The networks of these groups were equally heterogeneous, but
the female heads of households' networks tended to be substantially less
extensive than those of income-earning women with domestic partners.
Explanations for these findings were that women earning incomes were
exposed to a number of contacts beyond those related to work. The result
was an expanded, heterogeneous network. Female heads of households,
although exposed to a variety of people, were limited in their ability to
include men in their networks for a variety of attitudinal and practical
Level of poverty acted indirectly on networks because poorer women were
more likely to earn incomes. A direct effect was that they also were more
likely to form patron-client relationships.
Attitudes that appeared to have an effect on networks were those
specifying that childcare and housework were women's work and that earning
an income was men's work. Relationships with non-kin men were problems for
female heads of households. Religious activities were seen as positive.
The women in this study demonstrated three network patterns: 1) the
housewife; 2) the income-earning woman in a couple; and 3) the female head
of household (always earning an income in this sample). Of the three, the
female head of a household appeared to have the greatest survival problems.
The other two types demonstrated different approaches to survival; one a
sexual division of labor with the women performing the household and
childcare tasks and the man earning an income, and the other a sharing of
income-earning responsibility with a continued sexual division of labor in
childcare and house maintenance. Both provided for the survival of their
households, but the situation of the women in them appeared to be quite
One difference was that relationships in which the woman earned an
income tended to be more egalitarian than those in which she engaged in
household and childcare tasks only. These women were also important modes
in the neighborhood and community information flows. Both their independent
incomes and the information provided them by independent sources contributed
to their more egalitarian relationships with domestic partners.
Further research is required to test the applicability of these findings
to other settings. Given the uneven nature of national economic develop-
ment, it is probable that there are regional, as well as general societal
differences in such patterns. The findings and discussion in this paper
apply most clearly to the less industrialized urban areas of Latin American
countries but may be applicable elsewhere. It is important to note that the
expanded and heterogeneous networks of the self-employed women occurs in the
context of, at times, life-threatening poverty and is facilitated by the
very low income of the women's domestic partners. It is not a model for
increased women's participation in general but is a way of understanding the
importance of the control of resources for urban impoverished women. This
analysis clarifies the contribution by women to household incomes and their
importance in neighborhood and community networks in less developed
1. My thanks to Clayton Koppes, Anthony Leeds, Micaela di Leonardo, Banu
Ozertug, and J. Milton Yinger who have commented on drafts of this
paper. The research on which this article is based was undertaken
through NIMH grant 5-TO1-MH12751 and by grants from Harvard University
and Oberlin College.
2. The Alagados (which means marshes or tidal flats) is composed of several
named areas. Since it has now been filled, some argue that its old name
is not appropriate.
3. The meaning of the squatter settlements for social integration has been
a subject of debate. They were initially conceived as places of
residence for people marginal to the society. More recently, research
has shown that they are integrated, albeit disadvantaged (see Leeds
1977a, 1977b; Perlman 1976; Peattie and Aldrete-Hass 1981). They are
inhabited by long-term urbanites or natives who escape payment of rent
by building their own housing. The residents are quite heterogeneous.
Political activities around issues of squatting appear to be shaped by
the national political system (Leeds and Leeds 1976). For some recent
discussions of settlements, see also Eckstein 1977; Roberts 1979; Lobo
1982. For further information on Salvador, see Jelin 1974 and Faria
4. Salvador is also the center of the most traditional form of Afro-
Brazilian religion, Candomble. This religion has a pervasive impact on
Bahian culture, but appears to only indirectly affect the specific
aspects with which I am concerned. There were no houses of Candomble in
the neighborhood of the Alagados where this research was conducted,
although there are several in the larger area.
5. See also Lomnitz 1977; Chinchilla 1978; Eckstein 1977; Lobo 1982;
Roberts 1973; Safa 1974 on family, household, and other aspects of urban
poverty in Latin America.
6. The maximum household income is US$240. One minimum wage was the
equivalent of US$40 in 1973, the time of this study. Thus, the poverty
line in this study is two times the minimum wage. Cavalcanti (1981)
suggests that all households with incomes below six times the minimum
wage are below the poverty line. Extrapolating his indicator to these
data would mean that all households in the sample were below the poverty
7. The study of networks (see Bott 1957; Fischer 1982; Fischer et al. 1977;
Warren 1981) provides a way "to analyze and describe those social
processes involving links across, rather than within, group and category
limits" (Barnes 1969:54). More technical terms (Mitchell 1969) used for
the network elements on which I focus are: 1) "interactive multi-
plexity" referring to the nature of ties; 2) "density," a morphological
element referring to the form of the network; and 3) "anchorage,"
another morphological element referring to the number of people at the
center of the network.
8. "Extended families" here includes both partially extended families
(e.g., a couple, their children and a sibling of one of the couple) and
fully extended (e.g., a couple, their children, and a parent of one of
the couple). iEvery extended case is different.
9. Umbanda is a syncretic religion with elements of Candomble, Spiritism
and Roman Catholicism. It has grown rapidly in recent years,
particularly in South Central Brazil, and is now gaining adherents in
10. Spiritism is a Brazilian offshoot of Kardec's Spiritism. A central
belief is that one can communicate with the dead through mediums.
11. There is some controversy in' the literature on Latin America about why
households expand and contract, i.e., change from nuclear to extended
and sometimes back. Tienda (1980) has argued, based on Peruvian data,
that expansion while children are young 'allows for more income and
potentially more childcare. In contrast, Bilac (1978) argues that,
although other providers may enhance family consumption, they them-
selves become burdens as they age. Furthermore, she found that the
poorest families can send children to live with other relatives or find
live-in employment for young girls as maids. On the other hand, da
Silva (1979) found that having a large number of children was the most
viable survival strategy for the urban poor because wage controls led
to the need for many workers in a household.
TABLE 1. HOUSEHOLD AND
WORK BY HOUSEHOLD INCOME LEVEL
and Structure Women's Work
Extended wife Earning
12 16 31
(70.6) (69.6) (81.6)
* Nuclear and
** The poverty
Female head of household
level is US$80.
are presented separately in this table.
TABLE 2. THE EXISTENCE OF HELPING! RELATIONSHIPS BY WORK, IN PERCENTAGES
Income-Earning Women Housewife
No Patron Client No Patron Client
Relatives 20 6 19 3
(71.4) (60.0) (95.0) (100.0)
No Relatives 8 4 1* 0
Assist (28.5) (40.0) (5.0) (0)
Relatives 28 10 20 3
Assist (99.9) (100.0) (100.0) (100.0)
* Only assisted household at the subsistence level.
TABLE 3. DOMINANT ROLE RELATIONSHIPS
PARTNER, BY WORK
IN NETWORK SEGMENTS SHARED WITH DOMESTIC
* Other includes colleagues, friends, fellow members
religious groups and combinations of these.
of associations and
** Neighbors includes neighbor and combinations of friend and colleague.
*** Includes female-heads
TABLE 4. DOMINANT ROLE RELATIONSHIPS IN WOMEN'S INDIVIDUAL NETWORK SEGMENTS,
* See explanation, Table 3
** See explanation Table 3
TABLE 5. DOMINANT ROLE RELATIONSHIPS IN WOMEN'S
DYADS IN THEIR NETWORKS, BY
Income Earning Housewife Total
No Dyads 6 13 19
Relatives 4 2 6
Others* 27 6 33
Neighbors** 1 2 3
38 23 61
* Others refers to patrons, customers,
combinations of these
creditors, colleagues, friends and
** Neighbors refers to neighbors, and combinations of friend, colleague and
fellow religious association members.
1976 The Impact of Industrialization on Women's Work Roles in Northeast
Brazil. In June Nash and Helen Icken Safa (eds.), Sex and Class
in Latin America. New York: Praeger. Pp. 110-128.
1977 Women in the Informal Labor Sector: The Case of Mexico City. In
Wellesley Editorial' Committee (eds.), Women and National
Development: The Complexities of Change. Chicago: Unversity of
Chicago Press. Pp. 25-37.
Banck, Geertz A.
1980 Survival Strategies of Low-Income Urban Households in Brazil.
Urban Anthropology 9(2): 227-242.
Barnes, J. A.
1969 Networks and Politioal Process. In J.
Social Networks inI Urban Situations.
University Press. Pp. 51-76.
Clyde Mitchell (ed.),
Bilac, Elisabete D6ria
1968 Familias de Trabalhadores: Estrategias de Sobrevivencia. Sao
Paula: Edicbes Sfmbolo.
1957 Family and Social Network. London: Tavistock Publications.
Inaia M. Moreira de and Guarici A. Alves de Souza
A Producgo Nao-Capitalista de Desenvolvimento do Capitalismo em
Salvador. In Guaraci Alves de Souza e Vilmar Faria (eds.), Bahia
de Todos os Pobres. Petropolis: Editora Vozes. Pp. 71-101.
1981 Employment, Production and Income Distribution
Urban Sector of the Northeast: The Case of
Luso-Brazilian Review XVIII,(1): 139-154.
in the Informal
1978 Familia, Economia y Trabajo de la Mujer en Guatemala. Demografia
y Economia 12(1): 99-112.
Lower Class Life Strategies:
Recife's Metropolitan Area.
A Case Study of Working Families in
Unpublished Ph.D. dissertation,
Dietz, Henry A.
1980 Poverty and Problem-Solving under Military Rule: The Urban Poor
in Lima, Peru. Austin: University of Texas Press.
1977 The Poverty of Revolution. Princeton: Princeton University
1973 Brasilia, Plan and Reality:
Urban Settlement. Berkeley:
A Study of Planned and Spontaneous
University of California Press.
1980 Divisao Inter-regional do Trabalho e Pobreza Urbana: 0 Caso de
Salvador. In Guaraci Alves de Souza e Vilmar Faria (eds.), Bahia
de Todos os Pobres. Petrop6lis: Editora Vozes. Pp. 23-40.
1976 Occupational Marginality, Employment and Poverty in Urban
Brazil. Unpublished Ph.D. dissertation, Harvard University.
1982 To Dwell Among Friends.
Chicago: University of Chicago Press.
Fischer, Claude, Robert Max Jackson, C. Ann Stueve, Kathleen Gerson, Lynn
McCallister Jones with Mark Baldassare
1977 Networks and Places. New York: The Free Press.
e Administracdo Municipal (IURAM)
Recuperacao de Alagados: Estudo
Bahia: Imprensa Oficial da Bahia.
1976 The Bahiana in the Labor Force in Salvador, Brazil. In June Nash
and Helen Icken Safa (eds.), Sex and Class in Latin America. New
York: Praeger Publishers. Pp. 1291-146.
1974 Formas de Organizacao de Atividade Econ6mica e
Ocupacional. Estudos CEBRAP 9 (julho-setembro): 51-78.
Capitalismo e Marginalidade na America Latina. Rio de Janeiro:
Paz e Terra.
Household Composition and Headship as Related to Changes in Mode
of Production: SAo Paulo 1765 to 1836. Comparative Studies in
Society and History 22(1): 78-108.
Laumann, Edward 0.
1973 Bonds of Pluralism: The Form and Substance of Urban Social
Networks. New York: John Wiley and Sons.
1977a The Metropole, The Squatment and the Slum: Some Thoughts on
Capitalism and Dependency. Paper delivered at the Burg
Wartenstein Symposium No. 73. Wenner-Gren Foundation for
1977b Low-Income Urban Settlement Types: Processes, Structures,
Policies. Unpublished manuscript.
Leeds, Anthony and Elizabeth Leeds
1976 Accounting for Behavioral Differences: Three Political Systems
and the Responses of Squatters to Them in Brazil, Peru and
Chile. In John Walton and Louis Masotti (eds.), The City in
Comparative Perspective. Beverly Hills: Sage. Pp. 193-248.
Lewin, Helena, Jacqueline Pitahguy and Carlos Manoel Romani
1977 Mao-de-Obra no Brasil.1 Petr6p.olis: lEditora Vozes.
1959 Five Families. New York: Basic Books.
1982 A House of My Own. Tuscon: University of Arizona Press.
1977 Networks and Marginality. New York:' Academic Press.
Long, Norman and Paul Richardson
1978 Informal Sector, Petty Commodity Production, and the Social
Relations of Small-scale Enterprise. In John Clammer (ed.), The
New Economic Anthropology. New York: St. Martin's Press. Pp.
Madeira, Felicia R. and Paul Singer
1975 Structure of Female iEmployment and Work in Brazil, 1920-1970.
Journal of Interamerican Studies and World Affairs 17(4): 490-496.
1970 Urbanization Case History in Peru. In William Mangin (ed.),
Peasants in Cities. Boston: Houghton Mifflin. Pp. 47-54.
Martine, George and Jose Carlos Peliano
1975 Migracao, Estrutura Ocupacional e Renda nas Areas Metropolitanas
in Manoel Augusto Costa (ed.), Estudos de Demografia Urbana. Rio
de Janeiro: Instituto de Planajamento Economico e Social/
Institute de Pesquisas. Pp. 161-196
Miranda, Glaura Vasques de
1977 Women's Labor Force Participation in a Developing Society: The
Case of Brazil. In Wellesley Editorial Committee (eds.), Women
and National Development. Chicago: University of Chicago
Press. Pp. 261-274
Mitchell, J. Clyde
1969 The Concept and Use of Social Networks. In J. Clyde Mitchel
(ed.), Social Networks in Urban Situations. Manchester:
Manchester University Press. Pp. 1-50
1979 Household Arrangements and Multiple Jobs in San Salvador. Signs
Norris, William P.
1984 Patron-Client Relationships in the Urban Social Structure: A
Brazilian Case Study. Human Organization 43(1): 16-26.
Peattie, Lisa R.
1968 The View from the Barrio. Ann Arbor: University of Michigan
Peattie, Lisa R. and Jose A. Aldrete-Haas
1981 'Marginal' Settlements in Developing Countries. Annual Review of
Sociology 7: 157-175.
1976 The Myth of Marginality. Berkeley: University of California
Prandi, Jos4 Reginaldo
1978 0 Trabalhador por Conta Pr6pria sob o Capital. Sao Paulo:
1979 Cities of Peasants. Beverly Hills: Sage.
1973 Organizing Strangers: Poor Families in Guatemala City. Austin:
University of Texas Press.
1970 The Social Organization of Low-Income Urban Families. In Richard
N. Adams, Crucifixion by Power. Austin: University of Texas
Safa, Helen Icken
1974 The Urban Poor of Puerto Rico. New York: Holt, Rinehart and
Heleieth I. B.
Relationships of Sex and Social Class in Brazil. In June Nash
and Helen Icken Safa (eds.), Sex andiClass in Latin America. New
York: Praeger. Pp. 147-159.
1978 Women in Class Society.
New York: Monthly Review Press.
Singer, Paul I.
1980 A Economia Urbana de un Ponto de Vista Estrutural: 0 Caso de
Salvador. In Guaraci Alves de Souzale Vilmar Faria (eds.), Bahia
de Todos os Pobres. Petropolis: Editora Vozes. Pp. 41-69.
1980 Age and Economic
Dependency in Peru:
of Marriage and the Fami
A Family Life-Cycle
ly 42(3): 639-652.
Tilly, Louise and Joan W. Scott
1978 Women, Work and Family. New York: Holt, Rinehart and Winston.
1978 Hustling and Other Hard Work. New York: Free Press.
V6lez-Ibanez, Carlos G.
1983 Rituals of
Warren, Donald I.
1981 Helping Networks. Notre Dame: University of Notre Dame Press.
Warwick, Donald P. and Samuel Osherson (eds.)
1973 Comparative Research Methods. Englewood Cliffs: Prentice-Hall.
Marginal Men and Dominant Women: Kinship and Sex-Roles among the
Bahian Poor. Unpublished Ph.D. dissertation, Harvard University.
OFFICE OF WOMEN IN INTERNATIONAL DEVELOPMENT
MICHIGAN STATE UNIVERSITY
The WOMEN IN INTERNATIONAL DEVELOPMENT PUBLICATION SERIES were founded in
1981 to disseminate information rapidly to national and international
specialists in universities, government, and private institutions concerned
with development issues affecting women. The two series, WORKING PAPERS ON
WOMEN IN INTERNATIONAL DEVELOPMENT and the WID FORUM, publish reports of
empirical studies and projects, theoretical analyses, and policy discussions
that illuminate the processes of change in the broadest sense and encourage
manuscripts that bridge the gap between research, policy, and practice.
Publications in the series address women's historical and changing
participation in economic, political, and religious spheres, intra- and
inter-family role relationships, gender identity, women's health and health
care, and the sexual division of labor.
EDITOR: Rita S. Gallin
MANAGING EDITOR: Pat Hubbard-Garcia
EDITORIAL ASSOCIATE: Patricia Whittier
EDITORIAL BOARD: Marilyn Aronoff, Sociology; Anne Ferguson, Anthro-
pology; Ada Finifter, Political Science; Peter
Gladhart, Family & Child Ecology; John Hinnant, Anthro-
pology; Susan Irwin, Anthropology; Akbar Mahdi, Soci-
ology; Anne Meyering, History; Ann Millard, Anthro-
pology; Nalini Malhotra Quraeshi, Sociology; Barbara
Rylko-Bauer, Anthropology; Judith Stallmann, Agri-
cultural Economics; Paul Strassmann, Economics
NOTICE TO CONTRIBUTORS: To provide an opportunity for the work of those
concerned with development issues affecting women to be critiqued and refined,
all manuscripts submitted to the series are peer-reviewed. The review process
averages two months and accepted manuscripts are published within five to six
weeks. Authors receive ten free copies, retain copyrights to their works, and
are encouraged to submit them to the journal of their choice.
Manuscripts submitted should be double-spaced, sent in duplicate, and
include the following: (1) title page bearing the name, address and
institutional affiliation of the author; (2) one-paragraph abstract; (3) text;
(4) notes; (5) references cited; and (6) tables and figures. The format of
the article may follow any journal of the author's choice. Submit manuscripts
to Rita Gallin, Editor, WID Publication Series, Office of WID, 202
International Center, Michigan State University, East Lansing, MI 48824-1035.
TO ORDER PUBLICATIONS: Publications are available at a nominal cost and
cost-equivalent exchange relationships are encouraged. To order publications
or receive a listing of them, write to Office of WID, 202 International
Center, Michigan State University, East Lansing, MI 48824-1035, USA.
ACKNOWLEDGMENT: The Women in International Development Publication Series are
partially funded by a Title XII Strengthening Grant.
MSU is an Affirmative Action/Equal Opportunity Institution