{tknnmI-Wmm

n 1,990

n...

‘mrminm

.. “MM

..

Aida

T nu",-

 

‘ZFhafit‘

'3

_ (”“13“):‘y'b

an

r.
a? «’3': ”i
“A? "Ti

mafiatfiw
“$6515.
v fif‘fi 1' V???“
7, ~;:':v.: =. if" +'x'3"}‘v::.$1 ‘13:
:51“... 25%;“;ng ‘7?- ”:3 affix.
I’fii-

'6.. 5:62:11 "

:23;-

_ 6,: ~6-

6:35 . . .- ‘ 7’51526‘4- é$§$

313:3; :u 6- , firm» :‘ w}: .%-= ' Trim}: Jig _ “a; '

@646 99:1”sz was;
a W???

2".¢<
“ON

3111:“

a“!
”:65 3:53?

€914"
. av-w~~6‘~'~1- 6 '
ifi‘fiifi . .

 

 

 

 

2'3

. ‘3‘

 

~ ( n _:’.';’:’...:." 7—}. ' '
J . mm .5
()1, __

 

llllllllllllIIIHIIIHUJlllllllllllllllU!Ullllllllllllllll

193 00779 6208

__

r \

LIBRARY
Michigan State
1 University

M I

 

 

This is to certify that the

dissertation entitled

WOMEN'S STATUS AND FERTILITY
IN EGYPT AND BANGLADESH

presented by

Abdullah Al-Said

has been accepted towards fulfillment
of the requirements for

Ph.D. Sociology

degree in

 

 

a' rofessor

Date /5 WW

MS U is an Affirmative Action/[q ual Opportunity Institution 0- 12771

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before dds duo.

 

DATE DUE DATE DUE DATE DUE

oer 30 mo
‘éa 2894

MAGIC g

 

 

 

 

 

 

35

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunlty Inflation
C.

 

u r - ‘ ‘

   
 
  

STATUS AND FERTILITY IN EGYPT AND BANGLADESH
BY

Abdullah Al-Said ‘ ,-

.I
bu

 

“gajfifigiruiidg

 
 
 

'A DISSERTATION

 

   

Submitted to
Michigan State University
partial fulfillment of the requirements
for the degree of .

    
   
 
   
 
 

DOCTOR OF PHILOSOPHY

  

Department of Sociology

é77- 756.5"

ABSTRACT
WOMEN'S STATUS AND FERTILITY IN EGYPT AND BANGLADESH
BY

Abdullah Al-Said

The purpose of this study' was to investigate the
effects of women's economic and gender status on the
number of children ever born. Four indicators of women's
status were chosen. Educational attainment and gainful
employment indicated women's economic status, and
religious affiliation and son preference indicated women's
gender status.‘

The data were obtained from the Bangladesh Fertility
Survey (1976) and the Egyptian Demographic and Health
Survey (1988). A total of 5,128 Bangladeshi women and
7,774 Egyptian women currently married and still in their
first marriage were included in this study.

The major findings of the study indicate that women's
economic status is associated with the number of children
ever born when age at first marriage, age at interview,
and urban-rural residence were controlled. Egyptian women
with higher levels of education had lower fertility. The
effect of gainful employment on fertility was found to be
weak in Egypt. Bangladeshi women with gainful employment
had lower fertility than those who had no work; however,

education had a weak effect on the number of children ever

Abdullah Al-Said
born in Bangladesh. The effect of education was highly
related to women's age at interview: younger women had
fewer children but also had Inore years of schooling.
Women's gender status was also related to the number of
children ever born in both Egypt and Bangladesh. Muslim
women in both countries had a higher number of children
ever born than did non-Muslhn women. In addition, in both
countries, women with strong son preference had more
children ever born than did women with no son preference.

Two interaction terms were observed in this study.
The first interaction term was between women's religion
and gainful employment. Gainful employment after marriage
was associated with lower marital fertility, but it
depressed fertility more strongly for non-Muslims than for
Muslim wives. The second interaction term was found
between level of education and country of residence. The
strength of the depressant effect of women's education on
their fertility differed between Egypt and Bangladesh.
Egyptian women with no education had a higher number of
children ever born than did the same group in Bangladesh,
but Bangladeshi women with more than a primary level of
education had higher fertility than did the same group in
Egypt.

The implications for research and policy are

explored.

    
     
   

DEDICATION

. I dedicate this work to my wife Hesse and my 1 ‘
\ children Shatha and Khalid for their emotional g.) .~
“ support, encouragement, and endless love. ‘ “

ACK NOW LEDGMENTS

My sincere thanks and deep gratitude are extended to
Professor Nan E. Johnson, the chairperson of my doctoral
dissertation committee and my major academic advisor
during my Ph.D. program. Without. her unlimited help,
support, patience, and trust, this work would have been
very difficult to accomplish. Dr. Johnson was very
understanding and always helpful whenever I was in need.
It was a great pleasure learning front and working with
her.

I am especially grateful to other members of my
dissertation committee: Professor Jay Artis, for his
understanding and helpful comments; and Professor Daniel
Clay, whose valuable comments, suggestions, and
constructive criticisms were very useful to me. I am also
grateful to Professor Thomas Conner, whmn I had the
pleasure of working with during my master's degree. His
understanding, suggestions, and comments, especially about
the methodology portion of this dissertation, were quite
helpful.

My thanks and appreciation go to Imam Muhammad Ibn

Saud Islamic University for giving me the opportunity to

maintairl my graduate study and for its financial help
during my study in the United States.

I am indebted to the Egyptian National Population
Council, Institute for Resource Development, and Professor
Nan Johnson for their help in obtaining the data used in
this study' and for permitting me to use data from the
Egyptian Demographic and Health Survey and the Bangladesh
Fertility Survey for this study.

My special thanks go to Mohamed Kamali at Michigan
State University for his help and assistance with the
statistical analysis used in this study.

I would also like to thank my father, my mother, and
all my family members for their love and encouragement.

Finally, my wife Hessa AlHasmy, my daughter Shatha,
and my son Khalid deserve special thanks, not only for
their patience, support, and encouragement during my
study, but also for their endurance and toleration of

being far from their families and home country.

vi

TABLE OF CONTENTS

Page
LIST OF TABLES I I I I I I I I I I I I I I I I I I I V iii
LIST OF FIGURES I I I I I I I I I I I I I I I I I I I x
Chapter
II INTRODUCTION I I I I I I I I I I I I I I I I I I 1
Egypt and Bangladesh: A Demographic Overview . 9
The Status of Women: Definition and
Theoretical Background . . . . . . . . . . . 13
Organization of the Study . . . . . . . . . . . 15
II. RELATED RESEARCH AND LITERATURE . . . . . . . . . 17
Women's Economic Status and Fertility . . . . . 19
Gender Status and Fertility in Bangladesh
and mypt I I I I I I I I I I I I I I I I I I 24
Socioeconomic Development and Fertility:
Macro-Level Analysis . . . . . . . . . . . . 39
Research Hypotheses . . . . . . . . . . . . . . 44

III. RESEARCH METHODOLOGY . . . . . . . . . . . . . . 47

The Data . . . . . . . . . . . . . . . . . . . 48
Dependent, Independent, and Control Variables . 54
Dependent variable . . . . . . . . . . . . . 55
Independent Variables . . . . . . . 57
Control Variables . . . . . . . . . . . . . . 60
Statistical Techniques . . . . . . . . . . . . 62

IV. DATA ANALYSIS I I I I I I I I I I I I I I I I I I 67
Univariate Comparisions Between Egypt and

Bangladesh I I I I I I I I I I I I I I I I I 67

Testing the Hypotheses . . . . . . . . . . . . 74

92

V. SUMMARY, CONCLUSION AND RECOMMENDATIONS
Suggestions for/Further Research . . . . . . . 100

REFERENCES I e I 0 o e I e e I e e e e e e e e e e O 104

LIST OF TABLES

Page

1. Fertility rates and behavior in Egypt and
Bangladesh, CI 1990 I I I I I I I I I I I I I I 10

2. Descriptive statistics for selected variables in
Egypt and Bangladesh . . . . . . . . . . . . . . 68

3. Average number of children ever born in Egypt and
Bangladesh in each age group as a result of the
analysis of variance . . . . . . . . . . . . . . 71

4. Educational level of currently married women in
their first marriage by children ever born in
Egypt and Bangladesh . . . . . . . . . . . . . . 72

5. Currently married women who worked for cash
after marriage, distributed by number of
children ever born . . . . . . . . . . . . . . . 73

6. Son preference, distributed by number of children
ever born, in Egypt and Bangladesh . . . . . . . 74

7. Results of regression of the number of children
ever born on educational attainment and gainful
employment, controlling for age, age at first
marriage, and rural-urban residence in Egypt . . 76

8. Results of regression of the number of children
ever born on educational attainment and gainful
employment, controlling for age, age at first
marriage, and rural-urban residence in
Bangladesh . . . . . . . . . . . . . . . . . . . 77

9. Results of regression of the number of children
ever born on religion and son preference,
controlling for age, age at first marriage, and
rural—urban residence in Egypt . . . . . . . . . 79

10. Results of regression of the number of children
ever born on religion and son preference,
controlling for age, age at first marriage,
and rural-urban residence in Bangladesh . . . . 80

viii

A

11.

12

l3.

14.

15.

16.

Page

Summary table for MANOCOVA of the number of
children ever born on educational attainment,
gainful employment, religion, son preference,
and country with urban-rural residence, age,
and age at first marriage as covariates and
the interaction terms . . . . . . . . . . . . . 82

Interaction between religion and gainful
employment, controlling for other variables . . 83

Combined adjusted means of children ever born
for the interaction of work and religion, in
Egypt and Bangladesh . . . . . . . . . . . . . . 85

Contrast procedure in MANOCOVA of the effect of
level of education and country of origin on
children ever born, in Egypt and Bangladesh,
controlling for other variables . . . . . . . . 88
Coefficients for the contrasts in Table 14 . . . . 88

Adjusted means for the effect of education on
children ever born in Egypt and Bangladesh . . . 89

ix

 

  
  
 
 
   

LIST OF FIGURES

Lu ‘ Page
" I}{§I. ‘Interaction between work and religion . . . . ;-. .85
' .V ..-' ' ~

;§:3.\ Interaction between education and country . . .2., 89

CHAPTER I
INTRODUCT ION

Many demographers and policy makers are concerned
about high fertility rates in the less developed countries
(LDCs). While higher levels of social and economic
development are causally associated with lower fertility,
it is often less clear which aspects of economic
development should be promoted first so as to accelerate
fertility decline. Arguments are often made that the
higher status of women in the more developed countries
(MDCs) is an important reason for the lower fertility
there. Higher rates of employment in the professions and
of completion of secondary and tertiary levels of
education place most women into higher economic statuses
in MDCs than in LDCs. Because employment in the
professions usually occurs away from the homesite, it
usually creates incompatibilities between the roles of
mother and workera Moreover, since the completion of
higher levels of education postpones marriage, it also
delays the timings of first and later births so as to
reduce the overall number. The message to policymakers is
that schemes to promote the education of women, especially
for the professions, should hasten fertility decline.

1

2

Women have status based not only on their
socioeconomic class but also on their sex m E. The
biological potentialities unique to females is that they
alone can give birth to children and breastfeed them until
the infants can cut teeth and begin to chew semi-solid
foods. In poor LDCs, where children must begin to work as
youngsters and where there are no pensions or health care
programs for the elderly (such as the Social Security and
Medicare plans in the United States), parents must rely
chiefly on their offspring for financial, social, and
political support, particularly in old age. This need
pressures women in poor LDCs to put a premium on their
social roles as childbearers and breastfeeders, even if
this emphasis threatens their own physical health. Since
men cannot bear and breastfeed children, their mandate to
become parents does not pose the same physical risks as
faced by women. These realities mean that the gender
inequality between females and males is stronger in many
of the LDCs, where offspring are subjectively perceived as
important sources of economic support to parents and where
alternative sources are less available. This root of
gender inequality has been described as follows:

Wives in developing countries are often said to

suffer from frequent and closely spaced births and

to be exploited by the male member of the

household. This has been the motivation for

women's groups and family planning agencies to

stress the need for a woman to have control over

her body and the reproductive process. (Anker,
1982, p. 37)

3

Indeed, the World Fertility Survey (WFS) found that
wives in many developing countries desired to control the
timing as well as number of their births (Lighthourne,
1985). Nonetheless, this generalization may not be true
for most Muslim women. While the WFS included a number of
Islamic nations from North Africa, the Middle East, and
South Asia, few studies have systematically explored how
Islan1 might temper the antinatalist effects of economic
development.

Most Muslim societies, like other developing
countries, have very high levels of fertility, and most,
if not all, couples in the Muslim world prefer boys over
girls (Williamson, 1976; Kirk, 1966). This sex preference
most likely stems from the greater control of material and
nonmaterial resources by men than women. The economic
_dependence of women on men means that women will typically
want a large number of sons so as to increase the number
of male kin on whom to lean for help. The ecomwdc
advantages of having son's include financial help to the
family and security in old age (Cain, 1981; Nugent,
1985). Moreover, sons can physically defend the family
and carry on the family name. Evidence suggests that this
strong desire for sons increases the number of children
born per wife and discourages her use of contraception
(for example, see Johnson and Sufian, 1991). Therefore, a
higher level of economic development may provide greater

educational and employment opportunities for women, reduce

4

their reliance on men for advice and material support, and
increase women's access to health care. These factors
ought to strengthen incentives to reduce fertility and
provide means to (h) so. As such, any examination of
factors producing variation in the fertility rates of
Islamic nations at different levels of economic
development should investigate how the relationships
between women's education, women's work, and fertility
change across those levels.

Economic development should not be seen as the
all-powerful determinant of women's improved economic and
gender statuses nor as the only pathway to fertility

decline. For instance, The World Bank Atlas 1989 showed

 

that the gross national product per capita (GNP/c) in 1988
U.S. dollars was much lower in Indonesia than in Egypt
($430 and $650, respectively) but that the total fertility
rate (TFR) was also much lower in Indonesia than in Egypt
(3.3 and 4.7 children, respectively, expected to be born
per' woman. by age 49 if current age-specific fertility
rates hold)(Haub, Kent, and Yanagishita, 1990).

Noting that the Indonesian TFR was the lowest for the
40 Islamic nations of the world, Weeks (1988) attributed
this to the relatively high status of women in Indonesia.
He noted that Islam was introduced into Indonesia by Arab
traders in the fifteenth century, where the religion
became blended with Hinduisn, which had preceded it

(Weeks, 1988:11-12). Caldwell (1980:245-246; 1986:202)

‘

5

concluded that in nations where the predominant religious
culture emphasizes enlightment (the stress of Buddhimn,
Hinduism, and most Christian cultures) over strict
adherence to the literal "word of the book" (Islam and
fundamentalist Christianity), there is less fear of
educating girls. In poor nations, where most women are
unschooled, female education is not synonymous with female
autonomy from males but does help to bring this on
(Caldwell, 1980, 1986). These considerations highlight
the fact that a religious emphasis on education can
promote female autonomy in very poor countries (such as
Indonesia). But in nations where Islam has not been
grafted onto an earlier religious culture that emphasized
enlightenment, conservative Islamic cultures might slow
the pace at which economic development can elevate the
status of women.

I have just argued that economic development is not a
necessary condition for women to gain high status, but
neither is it always a sufficient condition. Even when
new wealth is created through economic development,
women's status may not rise if women are culturally seen
as inferior to men or if men retain control of the
wealth. For example, Caldwell (1986), in his study of low
mortality in poor countries, argued that women's position
is related to infant and child mortality. When women's
position is relatively good, the health of the children is

usually good as well. This was found to be true in the

6
poor countries. However, infant and child mortality was
found to be high in most of the rich Arab Gulf countries,
where the position of women is considered to be low.
Thus, the social structure of a society may play a
powerful role determining the status of women
independently of the wealth of the society.

Bangladesh and Egypt differ in their level of
economic development. This study uses data from these two
Muslim countries to investigate the relationship between
the status of women and fertility in the contexhs of
different levels of economic development. Although there
are many measures of economic development, per capita GNP,
life expectancy at birth, and the infant mortality rate
will be used here as its indicators. The per capita GNP
and life expectancy at birth are higher in Egypt than in
Bangladesh ($650 and 60 years in Egypt; $170 and 54 years
in Bangladesh), and the infant mortality rate is lower in
Egypt than in Bangladesh (90 and 120 per 1000,
respectively)(Haub et al., 1990). Hence, Egypt appears to
be more advanced than Bangladesh in terms of economic
development.

A comparison of some of the social and economic
differences between Egypt and Bangladesh reinforces this
conclusion. Egypt, for example, is much more urbanized
than Bangladesh (45 percent and 13 percent, respectively;
see Haub, Kent and Yanagishita, 1990). Egypt also enjoys

a substantially higher literacy rate. In 1980, Egypt's

7
literacy rate was 28 percent for women and 58 percent for
men over age 14 (Farid, 1987:341), while in 1981 in
Bangladesh it was 13.2 percent for all women and 26
percent for all men (Heitzman and Worden, 1989).
Moreover, this national inequality persisted even in 1988,
when the overall literacy rates were 44 percent in Egypt
and 33 percent in Bangladesh (Hoffman, 1991). Likewise,
the healthcare delivery system of Egypt is more extensive
than that of Bangladesh. In Egypt, the number of hospital
beds was one per 505 persons; and physicians: one per 616
persons. In the same period, the number of hospital beds
in Bangladesh was one per 3,187 persons; physicians: one
per 6,219 persons (Hoffman, 1991). It is plausible that
the expansion of health-care delivery as part of Egypt's
greater economic development has reduced Egyptian
fertility rates below Bangladeshi levels by making
contraception and postnatal care more accessible. This
antinatalist trend in Egypt may have been reinforced by
more literate wives who can gain health care information
for themselves and communicate more effectively with
doctors and nurses. Finally, Bangladesh seems to have
stronger cultural restrictions on female employment
outside the home than does Egypt. Zurayk (1985), for
example, writes:

In Bangladesh, the underemployment of women and

poor women in general relative to men is primarily

a consequence of highly segregated markets for

male and female labor, with the employment

opportunities on the female side far fewer than on
the male side. The degree of labor market

8

segregation is not simply a neutral accommodation

of women' s childbearing role. Rather market

segregation is both a consequence of and a means

for perpetuating the system of patriarchy. (p. 52)

These social and economic differences between Egypt
and Bangladesh determine the status of women. Egypt has
better educational and health systems and fewer cultural
restrictions on women's employment than does Bangladesh.
In other words, the economic status of women is thought to

be greater in Egypt.

The main objective of this study is to examine and
compare the status of women in Egypt with that of women in
Bangladesh and to explore how the status of women is
_re1ated to fertility in both countries. The advance of
economic development in Egypt (compared to Bangladesh)
probably explains why an Egyptian wife can expect to have
fewer children in a lifetime than a Bangladeshi wife (TFR
= 4.7 and 4.9, respectively), desires fewer children (3.6
and 4.6, respectively), and is more likely to use
contraception currently (38 percent and 33 percent,
respectively) (Haub, Kent and Yanagishita, 1990;
Lightbourne, 1985: 184). The reduction in actual and
wanted fertility and the willingness to employ
contraception over the course of economic development is
probably an outcome of economic development and change in

the status of women.

9

Egypt and Baggladesh:
A Demographic Overview

Egypt and Bangladesh, as two developing nations,
share many social and demographic characteristics. In
both countries, the majority of the population are Muslim:
91 percent in Egypt and 85 percent in Bangladesh (Weeks,
1988). Agriculture is predominant in both societies and
most <3f the two populations live in rural areas (urban
residency is 45 percent of the population in Egypt and 13
percent in Bangladesh)(Hoffman, 1990). Culturally, Egypt
and Bangladesh are male-dominated societies and the status
of women is objectively low in both countries.

The size and growth rate of the population are
serious problems facing both Egypt and Bangladesh.
According to the 1990 World Population Data Sheet (Haub et
al., 1990), the Egyptian population was 54.7 million and
its natural increase was 2.9 percent. In the same year,
the crude birth rate was 38 per thousand (see Table l) and
the death rate was 9 per thousand. The population of
Egypt is expected to reach 69 million by the year 2000,
yet its birth rate is considered one of the lowest in the
region. In fact, the birth rate in Egypt has not risen
above 44 per thousand since the 19205. The density of its
population and the settled nature of the society are
important reasons for the lower birth rate in Egypt
(Caldwell, 1981).

According to the Egyptian Demographic and Health

Survey Report (Sayed et al., 1989), the average number of

10

Table l--Fertility rates and behavior in Egypt and
Bangladesh, circa 1990.

 

 

Fertility Variables Egypt Bangladesh
Crude birth rate 38 39
TFR l 4.7 4.9
Mean age at marriage 21.3 16.3
Current usage of contraception 38% 33%
Mean number of 1

children ever born* 6.3 6.9
Sex preference ratio** 1.5 3.3

 

 

 

 

 

*mean for age group 40-49

**ratio of number of women who preferred a son to those
who preferred a daughter

lFertility Behavior in the Context of Development:
Evidence from the World Fertility Survey. United
Nations, 1987.

All other data in the table are from Haub (1990).

births per household in urban areas was 3.5 compared to
5.4 in rural areas. In the same report, the median age at
first marriage in Egypt was 18.5 years, and urban women on
average appeared to marry three years later than rural
women. Women in Egypt breastfeed their children for an
average of 17 months.

The Egyptian Family Planning Program was established
in 1965 but did not prove effective until the 19703.
Realisation In! Egyptian policymakers that socioeconomic
development was important for solving population problems
and for increasing the effectiveness of family' planning
programs led to the introduction of the population and
development project 111 the late 19705. This project

included about 1,500 villages across the country and had

11
the following objectives: (1) lower the birth rate;
(2) improve population distribution: and (3) improve
health, education, and training in rural areas. By 1982,
the project covered 2,915 villages (92 percent of all
Egyptian villages)(Stycos et al., 1988).

The population and development program has
successfully supported and encouraged the use of family
planning 111 Egypt. By 1988, 38 percent of currently
married women were using contraception (52 percent in
urban areas and 24 percent in rural areas) and 98 percent
knew at least one method. Of those who knew of a contra-
ceptive method, 87 percent approved of family planning,
and, of those, 70 percent believed their husbands approved
of using contraception (Sayed et al., 1989). These
figures were much higher than those reported in the 1979
Egyptian Rural Fertility Survey (ERFS). According to the
ERFS, only 215 percent of currently married women were
using family planning and only 57.4 percent of females and
47 percent of males approved of the use of contraception
(Kelley, Khalifa and El-Khorazaty, 1982).

Bangladesh, although smaller than Egypt in areal
terms (55,598 square miles and 386,650 square miles,
respectively), has a ‘population more than twice that of
Egypt. In 1990, the World Population Data Sheet indicated
that the population of Bangladesh was 114.8 million. Its
natural increase was 2.5 percent and the birth rate was as

high as 39 per thousand. The death rate in Bangladesh was

12
also high (14 per thousand) and the infant mortality rate
was 120 per thousand in 1990. Extrapolating these
figures, the population in Bangladesh will be about 146.6
million by the year 2000.

Bangladesh is considered one of the poorest countries
in the world and faces serious problems from rapid
population growth. Realizing this, Bangladeshi officials
introduced a family planning program in the early 19605.
As a result, knowledge of family planning increased so
that 64 percent of married women knew at least one method
of contracepton by the late 1960s, 82 percent by the
mid—19705, and 95 percent by 1980. The practice of family
planning, however, has remained very low. In the late
19605, only 6.4 percent of women in Bangladesh had ever
used contraception, and. only .3.7 percent. were currently
using some type of fertility control. In 1975, the rate
of ever use of contraception increased to 13.6 percent and
the current usage reached 9.6 percent. In 1980, 19.6
percent of women had ever used contraception, while
current usage increased to 14 percent in the same period
(Ahmed, 1987). The Bangladesh Contraceptive Prevalence
Surveys of 1983 and 1989 indicate that current users now
comprise 33 percent of the total population of married
women.

The great discrepancy between knowledge and use of
contraception in both Egypt and Bangladesh probably arises

from the fact that the majority of women in both countries

13

have never been to school (60 percent of ever-married
Egyptian women and 78 percent of ever-married Bangladeshi
women)(United Nations, 1987:218). However, I shall argue
that the higher average number of school years completed
by ever-married women in Egypt than Bangladesh (2.1 years
and 1.0 years, respectively) and the higher percentage of
ever-married Egyptian than Bangladeshi women employed in
professional or clerical jobs (4.9 percent and 0.5
percent, respectively) symbolize the higher status of
women 1J1 Egypt (United Nations, 1987:218, 264). These
social and economic factors perhaps explain the
differences in fertility and fertility behavior in both
countries (see Table 1). In chapters to follow, I shall
explore whether Egyptian wives have lower fertility than
their Bangladeshi counterparts not only because the
composition of Egyptian wives in higher strata of
educaticni and work is greater but also because the
antinatalist effect of women's education and work is
greater in Egypt.

The Status of Women: Definition and
Theoretical Background

 

 

There is no agreement on the definition of women's
status in the demographic literature. Female autonomy,
female dependency, and gender inequality are used in the
literature as synonyms of women's status. According to
Mason (1985), some authors believe that it is the social

institutions of gender that influence fertility rather

14

than the economic status of individual women as shaped by
economic institutions, while others argue that it is only
their socioeconomic positions that influence women's
fertility. Mason (1985) states that "Most terms and
definitions refer at least in part to gender inequality,
and most specifically focus on one of three basic
dimensions of gender inequality, namely (1) inequality in
prestige, (2) inequality in power, or (3) inequality in
access to or control over resources“ (p. 8). "Prestige"
is epiphenomenal, according to Mason, and thus unworthy of
theoretical debate. Instead, she emphasizes that people
have status because they belong to a certain socioeconomic
class (economic status) and tn) a certain biological sex
(gender status), and that both statuses should influence
fertility.

In this dissertation, female education and female
labor force participation are used as indicators of
women's economic status. Religion and son preference are
used as indicators of gender status. Religion, as an
indicator of gender status, plays an important role in
determining the status of men and women in a society.
Islam, for example, considers women inferior to men in
many social and political aspects: the testimony of a
woman is worth half that of a man: women inherit half of
what men inherit: the wife is in the service of her
husband, who has full authority over her: and women do not

have the same right of divorce as men (Stowasser, 1984).

15
All of these and other religious restrictions on women,
such as sex segregation, assign different social roles
(gender statuses) to men and women. In general, women are
constrained to work in the home and to care for children,
and men are responsible for providing resources for the
home and its residents. Son preference also reflects the
gender status of women. Women who are more independent
and can support themselves have less need for children as
sources of social and economic support (Mason, 1985).
Finally, women with higher gender status are more likely
to have access to and use fertility regulation than are
those with lower gender status.

However, for reasons to be outlined in Chapter II, I
shall examine whether women who have a lower ascribed or
achieved gender status (the Muslim religion or a
preference for sons over daughters) have higher fertility
even when they might have more favorable economic statuses
(some schooling or gainful employment). Evidence in favor
of this relationship will show that even when economic
development can increase female employment and education,
the antinatalist effects can be tempered by a patriarchal

religion requiring gender stratification.

Organization of the Study
This research on the effect of women's status on
fertility begins with this introductory chapter which
has presented a statement of the problem, the

definitions of and theoretical base for women's status

16
and fertility, and a brief overview of the demographic
characteristics of Egypt and Bangladesh. Chapter II
reviews some related literature regarding women's
status and fertility and concludes with a statement of
my hypotheses. The third chapter describes the data
used for this study, presents the measurement of the
variables of interest, and discusses the appropriate
statistical methods. Chapter IV is devoted to an
analysis of the data and a test of the hypodumes
stated in Chapter 11. Finally, Chapter V contains a
summary, conclusions based on the findings of this

study, and recommendations for future research.

CHAPTER II

RELATED RESEARCH AND LITERATURE

Many fertility studies consider the status of women
an important determinant of fertility behavior. Weeks
(1988) used worldwide data to show that even in nations
where important gains have been made in Gross National
Product per capita (e.g., most Arab Gulf states), high
fertility persists if the status of women is low; and even
in countries where little economic development has
occurred, fertility has declined if the status of women is
relatively high. Weeks used the sex ratio of school
enrollment and of life expectancy as indicators of the
status of women.

According to Anker (1982), the role and status of
women may influence fertility in two ways:

1 - It can affect fertility by affecting the

overall level of the household's well—being (which

is stressed in the literature on socioeconomic
determinants of fertility).

2 - It can affect fertility through the degree to

which the interests of households and wives

diverge and the relative ability of each spouse in
determining the number of children they should

have. (p. 38)

Mason (1987) has also examined the impact of women's

status and social position on fertility in developing

17

A

18
societies. She suggests that women's positions can
influence and depress fertility in five situations through
child supply, child demand, and child costs. These five
situations are:

(1) women's autonomy is likely to be inversely
related to the importance of urgency of
marriage for daughters, something that should
influence age at marriage and hence the supply
of children.

(2) In situations where women want fewer children
than men do or have more positive attitudes
toward fertility control, the greater is their
autonomy, the more power they will have in
fertility decisions, and the more likely they
will be to engage in innovative behaviors such
as using modern contraceptives. These should
in turn reduce fertility by reducing either
the demand for children or the psychic costs
of fertility regulation.

The greater the woman's economic dependency,

the more important children will be as

security assets and the stronger will be
preferences for sons.

(4) The seclusion of women is likely to lower the
opportunity costs of children. Insofar as
female autonomy leads to higher levels of
female education, it should increase the
opportunity costs of children. Both of these
should affect the demand for children.

(5) The greater the equality of sex, the more the
wife's health and well-being are likely to
enter into fertility decisions. This means
that where health considerations dictate
limiting childbearing or lengthening birth
intervals, higher female status may be
associated with lower fertility. (p. 738)

(3

v

To illustrate these points, a study of female
autonomy and fertility in India by Dyson and [bore (1983)
showed that fertility was lower in South India, where
women had higher gender status. In North India, where
fertility was observed to be high, women moved away from

their own lineage groups at marriage. Brides were

19
socialized to believe that their own wishes and interests
were subordinate to those of the husband's family group,
who put them under strong pronatalist pressures. On the
other hand, women in South India typically married lateral
kin (uncles or cousins) and did not become physically
separated from consanguineal relatives at marriage. This
meant that South Indian women had more social support.
Also, Southern women were more economically active outside
the home and had greater education than Northern women.
The higher status of Southern women was consistent with

their lower fertility.

Women's Economic Status and Fertility

As discussed in the preceding chapter, I shall use
two dimensions to represent the status of women. The
first dimension, women's economic status, shall be
measured in this dissertation research by female education
and female participation in the labor force. Caldwell
(1986) argued that the education and employment of women
are not the same as female autonomy, but they contribute
much to it. Education by itself, however, is considered
the most effective factor for improving women's status.
As Noman (1983) explains:

Educated women have more possibility to have an

equal share, more authority and equal participa-

tion in decision making processes both in the

family and in the society. This leads to more

personal autonomy of women, helping her to develop

definite opinions and views regarding many
important aspects of her own life. (p. 31)

20

Since education is one of the important indicators of
female status, Caldwell also sees female education as
playing the major role in fertility transition.

The primary determinant of the timing of the onset

of the fertility transition is the effect of mass

education on the family economy. The direction of

the wealth flow between generations is changed
with the introduction of mass education, at least
partly because the relationships between members

of the family are transformed as the nmrality

governing these relationships changes. (Caldwell,

1980, p. 225)

The greatest impact of education is not direct, says
Caldwell, but through the restructuring of family
relationships and family economies, and through the
changed direction of wealth flow.

Most other researchers agree that there is an inverse
relationship between female education and fertility.
Farques (1989) delineates three factors that cause
education to influence fertility. The first factor is
demographic, where longer schooling contributes toward
delaying the age at first marriage. The second is
economic, where the spread of education, especially
secondary education, is a sign of a transformation in the
role of children. The balance between the gain of having
a large family and the cost they require is reversed.
Finally, there is a psychological factor, where education
projects girls into :3 world outside their family circle.

With education, they may gain occupational ambitions other

than traditional home work.

21

Much empirical evidence supports the influence of
women's economic status (on fertility and contraceptive
behavior. Using level of education and work for cash or
kind to measure female status, Syed (1978) examined the
influence of female status on fertility in Pakistan. His
analysis shows that education played a powerful role in
determining fertility. hi both urban and rural areas,
education postponed age at first marriage, while the
effect of work status on age at first marriage was weak
for both urban and rural women. Furthermore, an inverse
relationship between education and children ever born was
observed. Primary education was found to be stronger in
rural areas with regard to children ever born. Also,
women's education influenced the desired number of
children in both urban and rural areas. Women with some
years of schooling desired fewer children, whereas the
effect of women's work status on desired number of
children was not significant. Finally, women with
schooling knew more about and used more contraception.
Syed argued that "education broadens the outlook of
females, increases their ability to plan their life more
meaningfully, and enables them to find security through
means other than excessive childbearing" (p. 428).

Following the same procedures, Shathar et a1. (1988),
in their study of women's status and fertility in
Pakistan, found that women's education (as a Ineasure of

women's status) played a major role in lowering fertility.

A

22
In addition, women with higher levels of education desired
fewer children and were less likely to have a preference
for boys.

Noman (1983), in his investigation of women's status
and fertility in Bangladesh, argues that fertility
declines as education level increases. Women with higher
levels of education were more likely to have contraceptive
knowledge, to practice family planning, and to have a
smaller family size than men. Noman also indicates that,
in general, employed women in Bangladesh had somewhat
lower fertility than non—employed women.

Thus, while most studies dealing with women's status
and fertility use female education and labor force
participation as measures of female status, most evidence
shows that. female education is stronger than female
employment as a fertility determinant. When Speare et a1.
(1973), for example, examined the effect of urbanization,
female employment, and female education on fertility
behavior in Taiwan, they found that female education was
the major factor affecting desired number of children, age
at marriage, contraceptive use, and son preference.
Residence background and work experience both had small
effects on fertility attitudes and behavior. In TUrkey,
Stycos and Weller (1967) also found fertility
differentials were associated with education and

urban-rural residence rather than with female employment.

 

23

While women's education appears stronger than
employment as a fertility determinant, other studies do
indicate the influence of female employment on fertility.
A cross-cultural study by Hirschman and Guest (1990), for
example, examines a multilevel model of fertility
determination with microlevel data from four Southeast
Asian countries. They assumed that community attributes,
including the status of women, children's economic roles,
and the level of infant mortality, affect fertility
behavior directly and indirectly through individual
statuses. Their analysis showed that the economic context
of women's status, especially modern sector employment,
had the highest (and most consistent effect on lowered
fertility in all four countries: Indonesia, Malaysia, the
Philippines, and Thailand.

Kasarda (1971) used data from 50 countries to examine
the effect of female labor force participation on
fertility. He found that high labor-force participation
of females as wage or salaried employees in non-
agricultural activities depressed a society's fertility
level. In another study, Rosen and Simmons (1971)
interviewed 726 Brazilian women in order to examine the
relationship between women's work, their participation in
family decisions, their fertility preference, and actual
fertility. They found that work status of women,
husband-wife decision making, and women's role attitudes

all appeared to negatively influence fertility. More

24
specifically, smaller family size was associated with
generally higher education, greater social contacts, more
sl<i].leed en1plc>yn1er1t, arid ecguallitzy in faniily
decision-making.

Gender Status and Fertility in
Bangladesh and Egypt

 

 

The second dimension of women's status is the social
position assigned to them by virtue of their biologically
based roles as breeders and breastfeeders. I shall use
two measures of gender status. The degree to which
individual women have not become psychologically
emancipated from traditional gender roles will be tapped
by their preference for the birth of sons. Oyeka (1989),
for example, examined the relationship between the number
of living sons and the use of contraception among female
teachers in urban Nigeria and found that women with no
lliving sons were least likely' to have ever used modern
contraceptives. Contraceptive use increased with the
number of living sons, and women with only one son and no
daughters are less likely to have ever used contraception.

Most studies have emphasized that son preference
raises fertility behavior. In her theoretical model
regarding the influence of sex preference of children on
fertility, Sheps (1963) states that

It is common knowledge that many parents do have

such preference and some empirical studies have

been made of the relationship between sex
distribution and number of children. If parents

really want three children regardless of sex, they
might reasonably be expected to try not to have

25

any more after the three have been born. Should
they want three boys, however, and persist in the
effort to reach this goal, they may have to have
more than three children in order to have the
three boys they desire. In fact, they will have
an average of about six children by the time they
succeed in having the three boys. (p. 66)

In other words, Sheps argues that if a certain number of
sons is strongly desired, then the total number of
children wanted by, or acceptable to, a couple can be
unstable and subject to upward revision until the desired
number of sons have been born.

Mannan (1989), in his investigation of the
rationality of large families in Bangladesh, illustrates
how children, especially' male children, are valuable in
Bangladesh.

The persistence of high fertility in Bangladesh

appears to be deep-rooted in the structure of

society and its culture. Children are regarded as

a good for' which there is run substitute. The

rural child provides valuable labor services to

the parental household during childhood.

Dependent children, especially sons, who engage in

wage labor usually give their earnings to their

parents. Grown children living away from the
parental household usually send regular remittance

to their aging parents. They are also the most

dependable source of old age security or of

insurance against temporary disability or
unemployment. Again, grown children contribute to
the security of their families in a violent world

by providing physical protection. (p. 67)

Mannan further illuminates the important role of male
children compared to female children by arguing that
"because of prevailing social custom, only male children
work outside the home. Thus daughters cannot help in most

of these activities, and to make matters worse, they cost

26
a large amount at the time of marriage by way of dowry,
etc." (p. 67).

In Egypt, as in Bangladesh, a very strong preference
for sons exists. Gadalla (1978) has argued that "It is
generally recognized that Egyptian culture, especially the
rural subculture, places a much higher value on male than
on female children and emphasizes the necessity of each
family's having at least one son" (p. 99). In his
investigation of Egyptian fertility, Gadalla found that
most respondents indicated they would continue to have
more children if they had more than two daughters and no
sons. There are two main reasons behind the importance of
having male children in Egypt. The first is to continue
the family and carry on the family name. The second is to
support the family financially, help on the farm, and
provide security for parents in their old age.

Since the son preference phenomenon in both Egypt and
Bangladesh is well documented in the literature, it should
influence fertility and fertility behavior in both
societies. With regard to contraceptive use, son
preference is found to play an important role. For
example, Koening et al. (1987), in their study of son
preference in Bangladesh, show that the use of
contraceptives among currently married women increased
with the number of living children and with the number of
living sons. For instance, among women with 0, 1, 2, 3,

or 4+ male children, the use of modern contraceptives

27
corresponded tr) 7.2, 17.4, 27.2, 33.1, and 34.4 percent,
respectively.

In a similar but more recent study, Rahman, Akbar and
Phillips (1990) used longitudinal data to examine the
influence of sex composition of surviving children on the
acceptance and continuation of contraceptive use among
3,145 women during 60 months in Matlab, Bangladesh. Their
findings show a direct and strong relationship between the
number of surviving sons and acceptance and continuation
of contraceptive use. The acceptance rate increased with
the number of living children, but more strongly with the
number of surviving sons. The authors indicate that women
with no living sons were less likely to accept contracep—
tion, while those with three or four sons were most likely
to accept and continue to use contraception. The authors
suggest that: "Couples make decisions regarding fertility
regulation based on the achieved number of sons and
daughters. Permanent contraception is more acceptable to
couples with more sons than daughters" (p. 15). Further—
more, "It also appears that the effect of son preference
is stronger" on the acceptance of terminal contraceptive
methods since nearly every couple who accepted
sterilization had at least one son, and son composition
predominates among children of couples who have terminated
contraception" (p. 25). In this study, the use of data

from only rural Bangladesh and the ever-usage of

28
contraception limited the generalizability of this
investigation.

Somewhat surprisingly, Sufian and Johnson (1989) did
not find that the death of a son in Bangladesh led to a
speedier subsequent birth than did the death of a
daughter, as would be expected from attempts by parents to
replace dying sons. However, the failure of their
analysis of the 1975-76 Bangladesh Fertility Survey to
find shorter median birth intervals after the death of a
son may have reflected a parental "strategy of hoarding"
sons (having as many as possible as soon as possible)
rather than a parental "strategy of replacing“ sons
(Preston, 1978:10).

Following this work, Johnson and Sufian (1990)
examined the relationship between son preference, son
mortality, and the adoption of contraception. Their
analysis of the 1975-76 Bangladesh Fertility Survey showed
a relationship between the mortality of sons, but not
daughters, and a lower rate of contraceptive initiation
(as indicated by proportions ever using contraception).
In addition, their analysis indicated that the experience
of son mortality was associated with a higher rate of
discontinuation of contraceptive use (as indicated by
proportions ever using but not currently using
contraception) .

In Egypt, Cochrane, Khan, and Osheba (1990) examined

the determinants of desired family size using the 1980

29

Egyptian Fertility Survey, which was a component ofthe
World Fertility Survey. Their analysis indicated that
wealth or lifetime income was positively associated with
the demand for children, and that education for both
husband and wife was negatively related to desired family
size. The couple's residence was found to be an important
variable determining their desired family size. The most
important determinant of desired family size was the
husband's/wife's knowledge of family planning, which
showed a negative effect on desired family size.
Surprisingly, their analysis showed that the wife's
desires for family size influenced the husband's desires.
Husbands did not appear to determine much use of family
planning. On the average, husbands in Egypt did not
prefer a larger number of children than did their wives.
These findings made it appear that women occupy greater
decision-making power over their reproduction in Egypt
than in Bangladesh.

Rizk, Stokes and Nelson (1982) have used data from
rural Egypt to examine the influence of child mortality on
birth spacing, additional children, and contraceptive
use. They find no relationship between child mortality
and length of birth intervals. However, child mortality
was found to be associated with both contraceptive use and
desired number of additional children. Women who had lost
one or more children desired a larger number of additional

children and were less likely to use contraception than

30
those with no experience of child mortality. This study
had a limited sample size of rural residents only and
failed to control for the sex of dying children, as was
done by Sufian and Johnson (1989).

Gadalla, McCarthy and Campbell (1985) have examined
the influence of the number of living sons on contracep-
tive use with data from 22,799 currently married women in
38 Egyptian villages. Their results show that, in
general, contraceptive use increased regularly as the
number of living children increased. Within each parity,
women with no male children were less likely to be using
contraception. However, the use of contraception
increased with tflua increase in living sons in each
parity. Ear example, 15.5 percent of all women with two
living sons practiced contraception, compared to 7.4
percent of women with two daughters. Also, 13.0 percent
of those with one son and one daughter were using
contraception at a higher parity. Among women. with four
children, the use of contraception was the lowest when
there were no sons. Among this group, the rate of
contraceptive use increased with an increased number of
living sons (14.9 percent with one son, 25.8 percent with
two sons, 30.0 percent with three sons, and 32.0 percent
with four sons). In this study, the authors indicate that
contraceptive continuation rates increased regularly with

the number of living sons.

31
In the same study, there was a contraceptive
distribution program. Four different oral contraceptives
were offered to eligible women. In a follow-up interview
nine months after the visit, the results remained the
same: women who had no male children or only one son were
still less likely to be using contraception, compared to
those with the same number of living children but with two
or more living sons. At the first visit, among those
women who did not use contraception, those with more sons
were more likely to accept the offered contraceptives. A
very low rate of contraceptive use was also found among
women with a large number of female children but with no
sons. The authors conclude that:
Despite theefforts of the household distribution
program to make contraceptive supplies and
information readily available to all married
women, and to remove the access barriers between
the method and the potential users, the program
did not achieve appreciable results among women
with few or no sons. In addition, it did nothing
to weaken the strong association between the
number of living sons and the use of contracep-
tion. (Gadalla, McCarthy and Campbell, 1985:168)
According to Williamson (1978), sex preference
influences fertility behavior through: (1) the strength of
sex preference; (2) the amount by which the preferred sex
ratio deviates from the natural sex ratio: (3) a
small-to-moderate desired number of children; and (4) the
availability of effective birth control. Because the
higher level of economic development in Egypt than

Bangladesh has put medical care (including contraception),

primary schoOling, and professional or clerical employment

32

more within the reach of Egyptian women, I maintain that
the latter will have a weaker dependency on male children
and thus a weaker preference for sons. In other words,
while son preference prevails in Muslim countries, I
surmise that it is stronger in those at the lower end of
the scale of economic development (Pakistan, Bangladesh)
than in those higher up the scale (Egypt).

The second measure of women's gender status in this
dissertation is membership in the Muslim religion. Many
western studies that deal with the status of Muslim women
have seen women in Islamic societies as subservient,
submissive, and powerless. Although most of these
religious obstacles to an equitable division of labor by
sex were true, especially in the past, women in most
Muslim societies are now overcoming these barriers
(Allman, 1978). It seems that most of these obstacles are
found in the more conservative Muslim societies, such as
in the Arabian peninsula societies, where restrictions on
women's emancipation still exist.

Weeks (1988) argued that Islam itself does not impose
any particular restrictions on labor-force activity by
women, but tradition has tended to impose itself in many
Islamic nations. In virtually every Islamic nation, there
are increasing numbers of women who are well-educated
and/or are engaged in non-domestic, non-agricultural
labor-force activities. Weeks pointed out that Muslim

women are faced with the lowest status in Southern Asia,

33
particularly in Bangladesh, and with the highest status in
Southeast Asian Islamic nations. This could explain the
high level of fertility in Southern Asia and the lower
fertility rate in Southeast Asia.

The low gender status of Muslim women probably causes

a higher rate of fertility in order for them to maintain
social and economic supports, especially in old age. The
higher fertility rates of Muslims have often been
empirically documented. Kirk (1967:568), for example,
argued that "Muslim institutions, more than those of other
world religions, favor a generally high natality.
Religion and high natality are more closely correlated for
Muslims than for any other religious group." In his study
of Muslim countries, Kirk drew the following conclusions:

1 - Islamic countries uniformly have high birth
rates.

2. These are supported by distinctive Islamic
attitudes and practices in family life rather
than by political or religious doctrines.

3 - The "normal” diffusion of birth control to and
within Muslim countries on the European pattern
has been inhibited by cultural discontinuity
between Muslim peoples and their neighbors.

4 - The continuing high birth rates in Muslim
countries, matched with encouraging progress in

reducing deaths, now lead to rapid population

5-

34
growth and its especially high visibility as a
handicap to economic and social progress.
As a result, several Muslim countries have

adopted measures to introduce birth control.

Kirk gave several reasons explaining the high fertility

rates of the Muslim population:

1 _

The high degree of tenacity with which old
beliefs and practices are maintained by Muslims.
Conformity to religious and social practices
which are so closely interwoven in Muslim life.

The strongly patrilinear and patrilocal quality
of the Muslim family, with male dominance and
responsibility prescribed by the Koran (the
Muslims' holy book).

Religious precepts that are favorable to early
remarriage of the widowed and divorced.

The belief that pleasures of flesh, especially
sexual intercourse, are God-given virtues to be
enjoyed and a conjugal obligation to be
fulfilled (celibacy is considered abnormal for
men and unthinkable for able-bodied women).

The unusually subordinate place of women in

Muslim society.

Kirk considered the Islamic religion a major obstacle

to the diffusion of family planning, fertility control,

and modern demographic behavior. However, it seems that

while the Islamic religion may encourage people to have

35

large families, cultures, belief systems, and low
socioeconomic development have more influence on Muslim
fertility behavior other than Islamic theology itself.
Allman (1978) argued that Islam is more favorable to
family planning than other religions and that the
obstacles to economic and social development are the major
determinants of Muslim traditional demographic behavior.

In the last two decades, many studies have been
devoted to examining and investigating Muslim fertility.
In Lebanon, for example, Chamie (1981) found that Sunni
and Shiite Muslims had higher birth rates than Catholic
and non-Catholic Christians, and within each group Shiites
had higher fertility than Sunnis, and Catholics had higher
fertility than non-Catholics (5.7, 4.8, 3.9, and 3.6
children ever born per wife in each respective religious
group). Nagi (1984) conducted a comprehensive study of 33
Muslim countries to examine the fertility trends and
differentials between 1960 and 1978. The main conclusions
of his analysis can be summarized as follows:

1 - Muslim fertility remains universally high and is
generally higher than in non-Muslim countries in
the same region:

2 - Very few Muslim countries have succeeded in
bringing down their level of fertility to
justify a search for the predictors of Muslim

fertility levels:

36

3 - In spite of a sufficient range of variations in
the economic and social correlates of fertility,
the. corresponding fertility variables in these
countries do not suggest that the reproductive
behavior of Muslim women has reached such
variations;

4 - Efforts directed toward stronger family planning
programs are clearly related to fertility
decline.

Regardless of these conclusions, fertility rates and
behavior have changed in many Muslim countries. Recent
investigations of Muslims' fertility indicate that
fertility rates dropped substantially in the Islamic
countries of Southeast Asia and North Africa (Weeks, 1988;
Farques, 1989).

Most studies of Muslim fertility compare Muslim
countries with-other countries rather than comparing the
fertility of Muslims with others within the country. In
an early investigation of social and psychological factors
affecting fertility in Egypt, Rizk (1963) feund a
significantly higher.ferti1ity rate for Muslhn women than
for non-Muslim women. In Bangladesh, Kabir, Uddin,
Chowdhury, and Ahmed (1986) observed that non-Muslim women
were at higher levels of contraceptive knowledge than
Muslim women. In addition, the use rates of modern as
well as traditional methods were found to be higher for

non-Muslims than for Muslim women. In a more recent study

37
in Bangladesh, Rahman et a1. (1990) found that
contraceptive use was higher among Hindu women (25 percent
higher) than among Muslims and that Hindus had a higher
continuation rate than Muslims.

Ahmed (1981) also examined the influence of religious
affiliation on cumulative fertility in Bangladesh. He
found that Muslim women in both urban and rural areas had
higher fertility rates than non-Muslim women, at least
partly because Muslim women had less education, a younger
age at marriage, and husbands with less education. While
Islamic values enshrining domestic roles as the proper
sphere for women probably influence the low levels of
female education and their youthful entry into marriage,
Ahmed suggested that religion can influence fertility
through other variables such as perception of sexuality,
son preference, subordination of women, contraceptive use,
and ritual and postpartum abstinence.

The higher fertility rate of Muslim than non-Muslim
women probably is at least partly an outcome of the Muslim
women's lower gender status. Although some scholars (for
example, Farques, 1989) have argued that women's status in
Islamic countries is improving, Muslim women still have a
low status in most Muslim societies because of the
religious restrictions on women's social and economic
activities. In Bangladesh, for example, it is noted that:

Rural women lack control over their own bodies and

over the decision to have children. Child-bearing

patterns depend on social status, which is often
dictated by important family members, by child

38

mortality rate, the number of and sex of previous

children, and by the family's economic status.

(Islam, 1982, p. 86)

Yet, other empirical evidence questions the inverse
effect of gender status on fertility in Bangladesh. For
example, Harbison, Khalegue and Robinson (1989) examine
female autonomy and fertility in the Garo villages in
Bangladesh, where female status is considered higher than
in the rest of the country. In Garo villages, women
control all property, do not have to wear purdah, and can
freely speak to other men. Moreover, Garo women play
major roles in familial decision-making, which is unusual
in Bangladesh. I-bwever, fertility of Garo‘women is high,
regardless of their high status. The authors argue that,
”Women's autonomy and high status are neither necessary
nor sufficient conditions for fertility decline. The
generally favorable economic and ecological settings made
fertility limitation unnecessary" (p. 1006). This
conclusion may be valid only in Garo villages.

Cain, Khanam and Nahar (1979) believe that women in
Bangladesh depend more on men because of the powerful
system of male dominance. In their analysis of women's
roles in rural Bangladesh, they explain the influence of
the patriarchal system on women and high fertility in two
points:

First, women of all economic classes face not only

the risks common to all in rural Bangladesh, such

as floods, political disruption, epidemics, and
crop failure, but also the special risks

associated with the patriarchal system. As the
bonds of obligation between kin erode under the

39

pressure of poverty, the risks of precipitous

decline in status for women increases. Second,

women who, for whatever reason, have no male on
whom to depend, and have to sell their labor, face

a market that is highly restricted both spatially

and functionally, entailing low wages and a high

rate of unemployment. The risk and insecurity

that partriarchy imposes on women represents a

powerful systemic incentive for high fertility.

(p. 432)

In summary, although a study by Cain et a1. (1979)
suggested that poverty may, in and of itself, reduce
the gender status of women by making them more
dependent on men for their physical survival, the work
of Harbison et a1. (1989) implied that greater material
wealth may, in fact, be a precondition for greater
female autonomy. As such, an innovation of this
dissertation research will be to explore how the
effects of women's gender status on fertility might be
conditioned by their economic status.

Socioeconomic Development and Fertility:
Macro-Level Analysis

To examine the relationship between fertility and
socioeconomic development, many fertility research studies
have focused on the micro- rather than the macro-level of
analysis. Researchers have argued that some particular
elements of economic development tend to reduce fertility,
such as by postponement of the age of marriage because of
women's educational attainment, while other elements tend
to increase the rate of fertility, such as by the

reduction of postpartum amenorrhea associated with

shortening of the breast-feeding period (Tabah, 1989),

40

which in turn can be associated with women's labor-force
participation. However, in a recent macro-analysis of the
association between economic development and fertility in
lesser developed countries, Raymont (1989) argued that the
economic welfare of a population, as measured by equality
in the distribution of income, is a powerful indicator of
lower fertility rates across the developing countries.
Raymont suggested that the mechanisms by which poverty
induces high fertility are more important than those by
which high fertility might cause poverty.

Raymond argued that rich people in most countries can
achieve a low rate of infant mortality whereas the poor
cannot. It is observed that high rates of infant and
child mortality resulted in more births being necessary to
achieve the desired family size. In addition, wealthy
people are more likely to seek higher education and become
involved in the labor force, which is negatively related
to fertility. Hence, equity of income distribution may
promote lower fertility by giving people equal access to
education, employment, health systems, and changes for a
better lifestyle. 1

Differences in socioeconomic factors do indeed seem
to play an important role in influencing human fertility,
as evidenced by numerous studies reported in the
literature. In a study by the United Nations (1989) that
included 27 developing countries, for example, child

survival and female education were found. to be strongly

41

correlated with lower fertility. Moreover, gross national
product (GNP) was found to play a modest role in
predicting not only the level of fertility but also those
very factors of child survival and educational attainment
as well as others. In other words, the level of the GNP
per capita sets limits on how much the government can
spend on building schools, hospitals, and roads and hiring
teachers, doctors, nurses, and laborers, and providing
other kinds of social services which alleviate poverty and
illness and make families (especially women) less
dependent on their children for their welfare.

In a similar study, Entwisle and Mason (1985) used a
multilevel analysis to examine the effects of
socioeconomic development and family planning programs on
fertility in 15 developing countries. Specifically, they
investigated the effect of socioeconomic characteristics
of individuals (women's education and rural-urban
childhood residence) on children ever born; and the effect
of per capita GNP and official family planning program
efforts on micro-level relationships between socioeconomic
characteristics and fertility. Their results showed that
the effect of micro socioeconomic characteristics on
fertility changes from positive to negative as economic
development proceeds. In addition, a positive micro
socioeconomic-fertility difference was observed in
countries at low level of economic development and with no

family planning programs. Moreover, per capita GNP and

42

family' planning programs influence women's fertility
within each country (for example, a positive effect of GNP
on fertility was foumd for women with a rural background
and with no education, but the relationship was negative
for women from an urban background with 10 years of
education). Although this study shows interesting
results, it failed to control for variables such as
women's age at marriage.

Kelley, Khalifa, and El-Khorazaty (1982) analyzed
community-level data from 32 villages in Egypt to examine
the effect of socioeconomic development on fertility.
Their findings show that, in general, a higher level of
socioeconomic development is positively related to
children ever born and negatively to the number of child
deaths. Agricultural mechanization reduced both the
desired number of children and children ever born: it was
also associated with an increase in contraceptive use in
rural areas. The authors concluded that:

1 - children ever born may increase in the short run
with socioeconomic development, in general, and
decrease with provision of mechanized
agricultureq. urban transportation, and
communication services:

2 - child deaths will decrease with socioeconomic
development and the provision of educational

services:

43
3 - desired number of children will decrease with
the provision of educational services and the
expansion of mechanized agriculture: and
4 - contraceptive use will increase with provision
of agricultural mechanization and educational
services. (p. 201)

Findings such as these lend support to Hess' claim
that "family planning program efforts depend on the level
of socioeconomic development, which, in turn, stimulates
lower fertility rates. This dependence decreases over
time and with further advances in socioeconomic
development” (1988:138). Recently, Faour (1989) has
analyzed data from 11 Arab countries to examine the
influence of family planning programs and socioeconomic
conditions on contraceptive prevalence. He found that
much of the variation (63 percent) in contraceptive
prevalence was explained by socioeconomic setting, which
is measured by urbanization, gross enrollment ratio, life
expectancy at birth, infant mortality, and per capita GNP.
Faour concluded that increase of contraceptive use was the
outcome of improved socioeconomic conditions as well as
stronger family planning program efforts.

In another study, Entwisle, Casterline and Sayed
(1989) examined the effects of village characteristics on
contraceptive behavior in rural Egypt. They concluded
that "contraceptive behavior in rural Egypt varies

systematically with village characteristics. The village

44

economy, modernization of agriculture, level of school
participation, and family planning service environment
positively influence contraceptive use and intenthxmfl
(p. 1030). 'Finally, Hoque (1988), examining the effects
of rural electrification and village-level development on
fertility behavior :hi Bangladesh, found significant
differences 1J1 fertility-related behavior among the
populations in electrified and nonelectrified villages.
He argued that the expansion of education, family planning
program, and communication through electrification
provided essential changes in fertility behavior and
shifted the functions of familial institutions to
nonfamilial ones.

Most of these macro-level analyses indicated that a
relationship between socioeconomic development and lower
fertility exists within and between countries. In other
words, a region or a country at a higher level of economic

development is associated with a lower fertility rate.

Research Hypotheses

The purpose of this study is to investigate the

influence of women's economic and gender status on the

number of children ever born per wife in Egypt and

Bangladesh. The first two hypotheses are deduced from the
literature review:

1. - Wives who have greater economic status (i.e.,

higher levels of education or gainful

employment) have fewer children ever born alive.

45

2. - Wives who have greater gender status (i.e., a
non-Muslim religion or a weak preference for
sons) have fewer children ever born alive.

An important goal of this comparative study is to
explore how the effects on fertility of these two
dimensions of women's status differ between Egypt and
Bangladesh. The literature cited above suggested that the
rising affluence in Egypt--as seen through the more
extensive education of women to the point of literacy and
the greater access per person to hospitals and doctors--
provides Egyptian women with greater access to information
about the world outside their homes, more employable
skills, and greater control over their own bodies, their
illnesses, and their pregnancies. Simultaneously, we have
seen that fertility remains high in Egypt, that son
preference continues strong, and that the resistance of
Egyptian women to family planning wears down gradually
with the successive births of sons. These considerations
lead to Hypothesis 3:

3.- The relationship put forth in Hypothesis 1 will
be much weaker in Egypt for Muslims than for
non-Muslims and for women who prefer many sons
than for women who prefer few sons: but this
relationship should be true in Bangladesh, too.

The final two hypotheses venture that the
relationships in Hypotheses 1 and 2 are not uniform in the

two Muslim countries under analysis because these

46
countries are at different levels of economic development.
These hypotheses may be stated as follows:

4. - The relationship proposed in Hypothesis 1 will
be stronger in Egypt, where economic development
has allowed a more pervasive rise in women's
economic status.

5. - The relationship proposed in Hypothesis 2 will
be much stronger in Egypt, where economic
development has lessened the dependency of women

upon men.

CHAPTER III
RESEARCH METHODOLOGY

Economic status and gender status are hypothesized
in this study to be two dimensions of women's status that
should depress fertility, and this depression should be
stronger at higher levels of economic development.
Therefore, tests of these proposals required data from two
countries at different levels of economic development.
From criteria discussed in Chapter 11, Bangladesh was
chosen from countries at the lowest level of economic
development, and Egypt was selected from countries at the
middle-low level of economic development.

I argued in the last chapter that economic
development may allow a more pervasive increase in the
status of women. However, the Islamic religion supports
family role segmentation. Conservative Muslims call for
the subordination of females to males in both public and
private life as being in harmony with the will of Allah
and with biologically governed human capacities. Thus,
strict observance of the Islamic (religion will reduce the
priority a society will place on education for girls in
public primary schools, will limit their exposure to other

(non-Islamic) cultures, and will relegate them to family

47

48

roles in the private sphere. These restrictions will
lower the status of women by reserving access to the
public sphere to men. Therefore, an alternative
hypothesis can be stated as follows: if men retain greater
control over the new wealth created by economic
development, then the status of women will not rise and
fertility will not fall (Caldwell, 1980). To choose
between my research hypothesis and its alternative, I
shall compare the two Muslim countries at different levels
of economic development.

In this chapter, I will first describe the data
sources from Bangladesh and Egypt. Second, the
measurement of variables of interest to this study will be
explained. Finally, the appropriate statistical

techniques for testing the hypotheses will be discussed.

The Data

 

Data for this study were obtained from the
Bangladesh Fertility Survey (BFS)(1976) and the Egyptian
Demographic and Health Survey (EDHS)(1988). The
Bangladesh Fertility Survey (BFS), conducted in 1975-76 as
part of the World Fertility Survey, had the primary
objective of providing a reliable estimate of and data on
fertility, mortality, and contraceptive use. It was
carried out by the Population Control and Family Planning

Division of the Health Ministry in Bangladesh.

49

In the BFS, a three-stage sampling procedure was
followed. In the first two stages, urban and rural areas
were selected; households were selected in the last stage.
The primary sampling units (PSUs) were census circles.
Bangladesh was divided into 4,949 census circles (4,241 in
rural areas and 708 in urban areas). The census circle is
an area determined by the boundaries of the Bangladeshi
administrative divisions. As a sampling frame, the census
office in Bangladesh provided two lists for) urban and
rural census circles, along with the population size and
number of households in each census circle. 160 PSUs were
selected from the 4,241 rural census circles, and 80 PSUs
were selected from the 708 urban census circles. The
probability that a given PSU would be randomly chosen was
proportional to the number of households within it. In
urban areas, one PSU consists of a few blocks, but in
rural areas it consists of a few villages. Lists of all
villages and urban blocks in the selected PSUs, along with
(the household numbers, were also obtained from the census
office. Each village or census block with less than 50
households was added to the closest village or census
block so that all units had 50 or more households. These
units were used as intermediate sampling units (ISUs).
From each PSU, one intermediate sampling, unit was sampled
with a probability proportional to size.

The ultimate sampling unit was a household which is

defined as "a group of persons usually living and eating

50

together in a structure or dwelling" (Sufian, 1984:34).
The ultimate sampling unit (i.e., a household) was sampled
with a probability inversely proportional to the size that
was used in sampling the PSU and ISU. From the total of
4,626 sampled rural households, 4,437 (95.9 percent) were
successfully interviewed. In the urban areas, out of
1,519 households selected, 1,418 were successfully
interviewed (93.4 percent).

The BFS sample covered 6,648 ever-married women
under the age of 50 in both urban and rural areas. A
total of 6,513 respondents were successfully interviewed
(5,024 of the eligible 5,123 respondents in rural areas
and 1,489 of the eligible 1,525 respondents in urban
areas).

Data collection for the BFS occurred during a major
rice harvest season (data collection began in December
1975 and ended in March 1976). During this season, men as
well as women in rural areas spend long hours working
out-of-doors. Failure to achieve a 100 percent response
rate in urban Bangladesh may be because of the mobility of
people where transportation facilities are available.
Other threats to the response rate included the social
distance that usually existed between older, illiterate
ever-married women and younger, literate, never-married
female interviewers. Despite these problems, the overall

completion rate was 94.1 percent (.959 X .981 = .931) in

51
rural areas and 91.2 percent (.934 X .976 = .912) in urban
areas.

Severalstrategies were followed in Bangladesh to
increase the response rate to the BFS. For example,
publicity about BFS objectives was broadcast in national
newspapers and on public radio. In addition, expecting
women's unwillingness to respond to male interviewers, a
team of female interviewers was trained in order to
minimize the non-response rate (Sufian, 1984).

It should be noted thaturban women were oversampled
in the BFS in order to obtain a larger sample of urban
women to facilitate statistical control of a number of
social factors that are related to fertility in urban
areas (Sufian, 1984). This study was not affected by the
oversampling of urban women since urban-rural residence
was used as a control variable.

The Egyptian Demographic and Health Survey (EDHS)
was conducted in Egypt in 1988 by the National Population
Council, with support from the United States Agency for
International Development (USAID) and the Institute for
Resource Development. (IRD). Its main objective was to
provide current and reliable information on fertility,
mortality, family planning, and maternal and child health
measures in rural and urban areas. The EDHS was carried
out in 21 of Egypt's 26 governates. Five governates were

excluded because disproportionately high resources were

52

needed to survey the dispersed population in these
governates.

Three stages of the EDHS sample design were 'used.
In the EDHS, the primary sampling units (PSUs) were towns
in urban areas and villages in rural areas. In the first
stage, 228 PSUs were selected (108 towns and 120
villages). Within each governate, the sample of P805 was
systematically selected with probabilities proportional to
sizes indicated in the 1986 Egyptian census. A list of
all towns and villages grouped by governate was prepared.
Any primary unit with less than 270 people, as reported in
the 1986 census, was combined with the neighboring unit in
the same governate. At the second stage, two segments
(i.e., intermediate sampling units) were selected from
each PSU, giving a total of 456 segments. In order to
select segments, a detailed map for each selected PSU was
obtained and divided into equal-sized segments. Each
selected segment has well-defined boundaries. During the
second stage, a complete listing of all the households
living in the selected 456 segments was prepared by a
well-trained staff. The third and final stage of the EDHS
sampling involved the selection of households (i.e.,
ultimate sampling units). A systematic sampling procedure
was followed to select households using the lists of all
households prepared during the second stage. The

household is defined as a dwelling which contains all

53
usual household members and any visitors who were present
during the night prior to the interviewer's visit.

A total of 10,528 households were selected from the
EDHS sample, of which 661 were ineligible because no
person was in the household the night before the
interview, the household was vacant, or the dwelling could
not be contacted for other reasons. A total of 9,805
households (93.13 percent) were successfully interviewed
(5,370 in urban areas and 4,435 in rural areas).

A person eligible for interview was defined as "an
ever-married woman between the ages of 15 and 49 years who
was present in a sampled household during the night before
the household interview" (Sayed et al., 1989). Out of a
total sample of 9,095 individuals, 8,911 (98 percent) were
successfully interviewed (4,409 in urban areas and 4,502
in rural areas). Thus the overall completion rate was 91
percent (.931 X .98 = .91), the same as for Bangladesh.

As in the BFS, the target sample of the EDHS was
divided among the governates in proportion to their size.
However, there as oversampling in five governates (Port
Said, Suez, Ismailia, Damietta, and Aswan). The reason
for oversampling in these disproportionately urbanized
governates was to obtain a sufficient household sample to
allow for governate-level contraceptive prevalence
estimates (Sayed et al., 1989).

The fieldwork for the EDHS began on October 29, 1988

and ended on January 27, 1989. All the field editors and

54
interviewers were females who were recent university
graduates, while the supervisors were well-trained males
(Sayed et al., 1989).

In Muslim countries, women's economic status is
usually altered by divorce. In addition, it would be
insulting to ask a post-married (widowed or divorced)
woman the question used in the Bangladesh Fertility Survey
to identify a preference for sons over daughters, namely,
"Do you prefer your next child to be a boy?" Because
widowhood or divorce could be expected to at least halt
the pace of childbearing until remarriage, I chose to
control the effects of marital disruption on fertility by
limiting my study to currently married women in their
first marriages. According to these criteria of sample
selection, for this study 7,774 women were selected from
the total sample of 8,911 in the EDHS and 5,128 currently
married women still in their first marriage were chosen
from the BFS sample (about 79 percent of the entire BFS
sample). These samples are not only large but offer

measures of all variables of interest to this study.

Dependent, Independent, and Control Variables
This study has six major variables--one dependent and
five independent--and three control variables. The
dependent variable is the number of children ever born
alive to the respondent. Two independent variables--
educational attainment and gainful employment--are used to

measure women's economic status. Two other independent

I.

55
variables--religion and son preference—~are used to
measure women's gender status. Country of residence is
the fifth independent variable. Since the data for this
study were obtained from two different samples (EDHS and
BFS), every attempt has been made to obtain an identical

measurement of the variables.

Dependent Variable

 

In the EDHS, three questions were used to measure the
dependent variable and to obtain information similar to
the BFS: (1) "How many sons live with you, and how many
daughters live with you?": (2) "How many sons are alive
but do not live with you, and how many daughters are alive
but do not live with you?"; and (3) "How many boys have
died, and how many girls have died?” The sum of the
responses to these questions was used as the total number
of children ever born alive to the Egyptian women. This
approach would be more likely to discourage underreporting
of live births in Egypt.

In the Bangladesh Fertility Survey, the interviewer
said: "Now I want to ask you some questions about each of
your pregnancies; that is, each live birth, each
stillbirth, each miscarriage, and each abortion. If you
have had any of these or any children who have died or who
live away from home, I would like to know about them.”
Then the interviewer asked the ever-married women to
report in order of occurrence: each of their pregnancies:

the name, date of birth, and sex of the child; if the

56

pregnancy ended in a live birth: survival status of the
child (if it was born alive); and its age at death (if it
died later after birth). It will be noticed that unlike
the Egyptian Survey, the Bangladeshi questionnaire did not
pose separate questions about sons who died, daughters who
died, sons now living away, and daughters now living
away. Thus, I wondered whether a cultural focus on sons
would result in underreporting of daughters not now alive
or living with the respondent, if she were not prompted by
questions specifically on daughters. Selective omission
of daughters' births would bias downwardly the total
number of children reported as ever born alive to women in
Bangladesh and would understate the differences in
fertility rates between Bangladesh and Egypt.

One procedure to test the relative underreporting of
sons' and daughters' live births is to use the sex ratio
of children ever born in both countries. In the BFS, the
sex ratio was 106 male births per 100 female births, which
appears to be normal (Sufian, 1984). In the EDHS, the sex
ratio was 106.5 (i.e., 106 roles for every 100 females),
which is also a normal sex ratio, since sex ratios of live
births vary between 104 and 107 in countries where birth
registration systems are complete (Shryock and Siegel,
1976:109). Thus there is no evidence that the
sex-selective underreports of live births were more severe

in Bangladesh.

57
Independent variables

There are five major independent variables in this
study. Two independent variables measure a woman's
economic status, of which the first is her level of formal
education. Both surveys have‘a very similar question
about the respondent's level of education. In the BFS,
the question related to educational level was, "Have you
ever attended-school?" Respondents answering "Yes" were
then asked, "What was the highest level of school you
attended? - 1) primary incomplete, 2) primary, 3) middle,
4) metric, 5) university." hi the EDHS, the educational
attainment question was, ”Have you attended school in the
past or are you currently attending school? - 1) yes,
attended 1J1 the past” :2) yes, attending currently,
3) never attended." Respondents answering "1” or "2” were
then asked, "What was the highest grade which you
successfully completed? -- 1) primary, 2) preparatory,
3) secondaryn 4) upper intermediate, 5) university,
6) more than university."

To promote comparability in measuring the wife's
educational level in the two countries, I have collapsed
the measure into three categories because only a very
small percentage of respondents achieved more than middle
school, especially in the case of Bangladesh. These three
categories of education are: (0) no formal education;
(1) up to completed level of primary schooling; and

(2) more than a primary level of education.

58

The second index of women's economic status is their
work since marriage. This variable .was measured in the
BFS by asking the respondent, "Have you ever worked since
you were married?" Respondents answering "Yes" were then
asked, "Do/did you work for a member of your family, or on
a family farm, or for someone else, or are/were you
self-employed? - 1) family member, 2) work on family farm,
3) someone else, 4) self-employed." In the EDHS, the work
question was more direct, asking, "Since you were first
married, have you ever done any work for cash? - 0) Yes,
1) No."

Tb, obtain a similar measurement of work status from
both samples, the four categories in the second question
of the BFS work status (work for family member, work on a
family farm, work for someone else, self-employed) were
recoded as working, scored "0." Those who responded
negatively to the first question of work status in the BFS
were scored as not working, scored "1."

Women's gender status. in this study is measured by
on) independent variables, one ascribed and one achieved.
The ascribed variable is religion, i.e., whether or not a
woman is Muslim (0 = yes, 1 = no). The religion question
is stated in the BFS as follows: "What is your religion? -
1) Islam, 2) Hindu, 3) Christian, 4) Buddhist, 5) other."
The same question was asked in the EDHS but with different
categories: "What is your religion? - 1) Muslim,

2) Christian, 3) other." Since the interest of this study

59
is whether the respondent is Muslim or not, and to make
this variable comparable in both surveys, the religion
variables was scored 0 = Muslim and 1 = others.

The achieved variable is whether women's preference
for sons is strong or weak. From the Egyptian data, son
preference is measured as the ratio of the desired number
of sons to that of daughters. For the Bangladesh data,
the procedure for arriving at strength of son preference
is more complicated because there is no direct question
about the desired number of male and female children. In
Bangladesh, a mother preferring more children was asked
whether she wanted her next birth to be a boy. If she
said "yes,” "1" is added to her number of sons already
born alive, and the sum is divided by) the number of
daughters already born. If she said "no," ”1" is added to
the number of daughters already born, and the sum is
divided into the number of sons already born. If the
mother desired no additional children, the sex ratio of
her children ever born is computed. This dimension of
women's gender status is operationalized in such a way
that higher scores mean lower status. The son preference
variable in both Egypt and Bangladesh was scored 0 = no
son preference (if the desired sex ratio was not greater
than unity) and l = son preference (if the desired sex
ratio was greater than unity).

Finally, because this dissertation is a comparative

study, country of residence is used as the fifth predictor

60
variable. The survey data are pooled and each female
respondent is scored by country (0 = Bangladesh, 1 =
Egypt). The direction of this score should be inverse to

fertility.

Control Variables

 

Testing the proposed negative effect of women's
status on fertility is the basic objective of this study.
Most of the literature on fertility in Egypt and
Bangladesh reports that women's age and rural residence
are positively related to their number of children ever
born, while their age at first marriage is negatively
related. Thus it is necessary to hold constant the
effects on fertility of these factors that might confound
the associations hypothesized between the independent
variables and the dependent one. Therefore, three control
variables are used in this study: women's age at
interview, women's age at first marriage, and women's
residence.

Women's age at interview was measured similarly in
both the BFS and the EDHS. In the BFS the question of
respondent's age was phrased as, ”How old are you?" In
the EDHS, the question was asked, "How old were you at
your last birthday?" The responses were recorded in
completed years of life.

Measurement of women's age at first marriage in the
Bangladesh data was asked by a direct question, "In what

month and year were you and your husband married? How old

61
were you then?" In the Egyptian data, age at first
marriage was measured by two questions: "In what month and
year were you born? In what month and year did you first
enter into a marriage contract?" The Egyptian women's age
at first marriage was computed by subtracting the year of
birth from the year of first marriage.

It should be noted that in both Bangladesh and Egypt,
the age at first marriage is not necessarily the same age
at which the bride goes to live with the groom. This is
because most marriages in both countries are arranged by
the parents CHE the bride and groom, and the formal
contracts may be drawn while both are still minor
children. The marriage of little girls happens more often
in Bangladesh. Forty-seven percent of the women in the
BFS reported being married before age 13, while only 4
percent of the women in the EDHS reported being married at
that age. While age at first marriage cannot accurately
indicate the age at which the risk of pregnancy began, it
does nonetheless suggest inverseLy the status that a
particular woman occupied in her family of origin and thus
should be inversely related to fertility.

Finally, women's place of residence was similarly
measured in both surveys by scoring urban =70 and rural =
1, according to both the Egyptian and Bangladeshi

administrative definitions of urban-rural residence.

62
Statistical Techniques
Tests of the first two hypotheses require a
multivariate statistical technique because each
independent variable has been operationalized in two ways
and because three control variables have been identified.
For this reason, a multiple regression analysis was
performed. The basic assumptions of regression analysis
are:
l - the samples are randomly selected from the
population of interest;
2 - y is conceptually interval-scaled;
3 - y is a linear function of the independent
variable xi;
4 - with each population j, the distribution of y
about xij is normal:
5 - within each population j, the variances Wzij
are equal:
6 - the error terms eij are independent and normally
distributed within each group (Johnson, 1977).
The statistical equations of Hypotheses l and 2 are:
“1‘ Y1 = a + B1x11 + B2x12 + B3x13 + B4x14 +
Bsxis + ei, where
y1 = children ever born alive,
a = the intercept,

B1,BZ,B3,B4,B5 = the unstandardized regre551on

coefficients,

63
= level of education,
xi2 = gainful employment,

xi3 = age at first marriage

xi4 = age at interview
xi5 = urban-rural residence,
e = the error term.

H ‘ Y1 = a + B1x11 + B2x12 + B3x13 + B4x14 +

' . + .
85x15 e1, where

yi = children ever born alive,

a = the intercept,

Bl,B2,B3,B4,BS = the unstandardized regression

coefficients,

xi1 = religion,

xi2 = son preference,

xi3 = age at first marriage
xi4 = age at interview

xis = urban-rural residence,
e. = the error term.

1
In order to test Hypotheses 3, 4 and 5 to examine the

effect of the independent variables on the dependent
variable while controlling the effect of other variables,
multivariate analysis of covariance (MANCOVA) is required.
It has been stated that, "If one or more variables are
collected to statistically control for sources of
variation with multiple criterion variables, then
multivariate analysis of covariance (MANCOVA) is the

appropriate method of analysis” (Bray and Maxwell, 1985).

64

Moreover, Hypothesis 3 expects a statistical interaction
between women's economic status (educational attainment
and gainful employment) and women's gender status
(religion and son preference). Hypothesis 4 ventures a
statistical interaction between women's economic status
and country of residence, while Hypothesis 5 anticipates a
statistical interaction between women's gender status and
country of residence. As such, a multivariate technique
was needed to test for interactive relationships between
the independent variables while holding constant the
effect of the control variables (women's age at interview,
women's age at first marriage, and urban-rural residence).
These two important requirements led to the choice of
multivariate analysis of covariance (MANCOVA) as
appropriate for testing Hypotheses 3, 4, and 5.

The structural equations for Hypotheses 3, 4, and 5
are as follows:

“3‘ Y1 = u + Bxi‘h'uxi) + Bx2("2‘ux2) +
Bx3 k + 8L +
(ER)ik + (WR)jk + (ES)iL + (WS)jL, where

(x3-ux3) + E:i + Wj +~R

y1 = children ever born alive,
Bl,B2,B3 = the unstandardized regression

coefficients,

x the first covariate, age at interview,

11
x12 = the second covariate, age at first

marriage,

ER.

H

4

1k

Y1

65
x13 = the third covariate, urban-rural

residence,

E1 = level of education,

Wj = gainful employment,
= religion,

SL = son preference,

interaction term between level of
education and religion,

WR. interaction term between work and

JR:
religion

ES. = the interaction term between level of

1L
education and son preference,
WSjL = the interaction term between work and
son preference,
= u + Bxl(xl-uxl) + Bx2(x2-ux2)
+ Bx3(x3-ux3) + Ei + wj + Cm + (EC)im
+ (WC)jm, where
y1 = children ever born alive,
81,82,83 = the unstandardized regression
coefficients,
xil‘= the first covariate, age at interview,

x12 = the second covariate, age at first

marriage,

x13 = the third covariate, urban-rural
residence,

E1 = educational attainment,

W = work for cash,

66
Cm = country of residence,
ECim = interaction term between level of
education and country of residence,
Wij = interaction term between work for cash
and country of residence.
“5‘ Y1 = u + Bxl‘xi‘uxi) + Bx2‘x2'“x2)
+ Bx3(x3-ux3) + Rk + SL + Cm + (RC)km
+ (SC)Lm, where
y1 = children ever born alive,

Bl,B2,B3 = the unstandardized regression

coefficients,

xi1 the first covariate, age at interview,

xi2 = the second covariate, age at first
marriage,

xi3 = the third covariate, urban-rural
residence,

religion,

,3"

S = son preference,
Cm = country of residence,

interaction term between religion

w
$0
a

and country of residence,

SCLm = interaction term between son preference

and country of residence.

CHAPTER IV

DATA ANALYSIS

The aim of this research is to investigate the
influence of women's status on fertility. In Chapter II,
I argued that the advance of socioeconomic development
should stimulate a higher level of women's status, which I
hypothesized to depress fertility. In Chapter III, I
described the national surveys of two Muslim cultures at
different levels of economic development. This chapter
presents the findings based on a total of 12,902 currently
married women respondents, still in their first marriage
(N = 7,774 cases from Egypt and 5,128 from Bangladesh).
Below, I first describe the differences between Egypt and
Bangladesh on the levels of the independent, dependent,
and control variables of my five hypotheses. Second, some
important cross-tabulations are presented. Finally,
multiple regression analysis and multiple analysis of
covariance are used to test the hypotheses established in
Chapter II.

Univariate Comparisons Between
Egypt and Bangladesh

 

Table 2 shows that the majority of the respondents in

Bangladesh lived in rural areas (76.8 percent), whereas in

67

Table 2--Descriptive statistics for selected variables in

68

Egypt and Bangladesh.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Egypt Bangladesh
Variable Frequency Percent Frequency Percent
Residence
Urban 3902 50.2 1191 23.2
Rural 3872 49.8 3937 76.8
Total 7774 I00.0 5128 100.0
Religion
Muslim 7236 93.1 4182 81.6
Others 535 6.9 944 18.4
Total 7771 100.0 5126 100.0
Education
No school 3720 47.8 3591 70.3
Through
primary 2150 27.7 1061 20.8
More than
primary 1904 24.5 455 8.9
Total 7774 100.0 51 7 100.0
Work status
Working
for cash 1190 15.3 488 9.5
Not working 6581 84.7 4640 90.5
Total 7771 100.0 5128 100.0
Son preference
Yes 1572 20.2 2116 41.3
No 6199 79.8 3012 58.7
Total 7771 100.0 5128 100.0
Variable Mean Std.Dev. Mean Std.Dev.
Age at
interview 31.834 8.409 27.137 9.372
Age at first
marriage 18.911 4.139 12.848 2.638
Children
ever born 3.975 2.847 3.982 3.117
Living children 3.295 2.190 3.040 2.395
Living sons 1.707 1.434 1.565 1.529
Living daughters 1.588 1.401 1.475 1.409

 

 

69
Egypt the distribution between urban and rural areas was
nearly equal (50.2 and 49.8 percent, respectively) because
of the sampling procedures. The higher level of urbanism
did not imply greater religious diversity in Egypt, where
the percentage of Muslim respondents was higher than in
Bangladesh (93.1 and 81.6 percent, respectively).

Table 2 also shows that Egyptian women were more
educated and more likely to work for cash than Bangladeshi
women. For the Bangladeshi women, 29.7 percent had some
education and 8.9 percent had more than a primary
education. For the Egyptian women, 52.2 percent of the
currently married women had some schooling; 24.5 percent
had more than primary schooling. Only 9.5 percent of the
Bangladeshi women worked for cash compared to 15.3 percent
of the Egyptian women. These relationships are consistent
with my thesis that the economic status of married women
of childbearing age is greater in Egypt than in
Bangladesh.

Mason (1985) pointed out that more highly educated
females may have a greater voice in the planning of their
marriages in cultures where marriages are arranged and
where their families may be expected to raise larger
dowries. Perhaps for this reason the mean age at first
marriage was much later in Egypt (18.911 years) than in
Bangladesh (12.848 years). In addition, the higher mean
age of respondents at time of interview in Egypt (31.834

years) compared to respondents in Bangladesh (27.137

70

years) may reflect the shorter life expectancy in
Bangladesh, a fact which occurs partly from an unusually
high rate of maternal mortality during or shortly after
childbirth. Mason (1985) noted that maternal mortality in
Bangladesh often results from inadequate nutrition during
pregnancy. But this in itself is a product of the low
status of women in Bangladesh, where men eat meals first
and women get the leftovers.

Another indicator of women's much lower gender status
in Bangladesh compared to Egypt is in the preference for
the births of sons instead of daughters. Indeed, this
preference was expressed nearly twice as often by the
Bangladeshi respondents than by the Egyptian respondents
(41.3 percent and 20.2 percent, respectively). However,
wives' preferences may have little impact on actual
fertility if husbands dominate fertility decision-making.

Surprisingly, the fertility rate was almost the same
in Egypt and Bangladesh. The average number of children
ever born was 3.97 in Egypt and 3.98 in Bangladesh.
However, this apparent similarity may be distorted by a
lack. of control for the age of the respondent. The
Bangladeshi women were, on average, younger than the
Egyptian women and would then be expected to average
fewer, not similar, numbers of children ever born. In
fact, within all but one age group, Bangladeshi women had
on average a higher number of children ever born than did

Egyptian women (see Table 3). The exception occurs only

71

Table 3--Average number of children ever born in Egypt and
Bangladesh in each age group as a result of the
analysis of variance.

 

 

 

Average number of children ever born
Age group Egypt Bangladesh
less than 15 .291 .213
16-19 .687 1.012
20-24 1.694 2.502
25-29 2.971 4.412
30-34 4.046 5.928
35 or more 5.858 7.423

 

 

 

 

 

in the first age category, where the number of children is
almost the same, probably due to the small number of
Egyptian respondents ix! that category. However, the
average numbers of living children, living sons, and
living daughters were somewhat higher in Egypt than in
Bangladesh (3.29, 1.7, and 1.58; and 3.04, 1.56, and 1.47,
respectively), perhaps because of the longer life
expectancy at birth in Egypt.

Table 4 shows the relationship between educational
attainment and number of children ever born to currently
married women in Egypt and Bangladesh. It is clear from
the table that the higher the level of education, the
lower the fertility rate for both countries. For example,
only 18.1 percent of Egyptian women and 24.6 percent of
Bangladeshi women who had more than a primary education

had 4-to-8 live tflrths. However, the educational factor

72

Table 4 -Educational level of currently married women in
their first marriage by children ever born in
Egypt and Bangladesh.

 

Number of children ever born

 

Level Egypt Bangladesh
of
Education 0-3 4-8 9+ Total 0-3 4-8 9+ Total

 

 

No school 1426 1837 457 3720 1661 1551 379 3591
38.3% 49.4 12.3 100.0% 46.3% 43.2 10.6 100.1%

Through 940 1071 139 2150 562 403 96 1061
primary 43.7% 49.8 6.5 100.0% 53.0% 38.0 9.0 100.0%

More than 1546 345 13 1904 328 112 15 455
primary 81.2% 18.1 0.7 100.0% 72.1% 24.6 3.3 100.0%

Total 3912 3253 609 7774 2551 2066 490 5107
50.3% 41.8 7.8 100.0% 50.0% 40.5 9.6 100.0%

 

 

 

 

 

Note: Percentages may not add to 100.0% due to rounding
error.
appears to be stronger in Egypt than in Bangladesh when
respondents had more than a primary level of education.
Table 5 presents a cross-tabulation of currently
married women who worked for cash since they entered into
the marriage union by the number of children ever born.
The table indicates that working for cash, like education,
depresses fertility in both Egypt and Bangladesh.
However, Egyptian women who worked for cash show lower
fertility than their counterparts in Bangladesh. For
example, of those who had worked for cash, only 25.8
percent of Egyptian women had 4-8 live births, compared to
51.2 percent of the Bangladeshi women. However, the World
Fertility Survey showed that the percentage of

ever-married women aged 15-49 whose current or most recent

 

73

Table 5--Currently married women who worked for cash
after marriage, distributed by number of children
ever born.

 

 

 

 

 

 

 

 

Number of children ever born
Egypt Bangladesh

Work for
cash 0-3 4-8 9+ Total 0-3 4-8 9+ Total
Yes 843 307 40 1190 192 250 46 488

70.8% 25.8 3.4 100.CB 39.3% 51.2 9.4 99.9%
No 3067 2945 569 6581 2370 1825 445 4640

46.6% 44.8 8.6 100.0% 51.1% 39.3 9.6 100.0%
Total 3910 3252 609 7771 2562 2075 491 5128

50.3% 41.8 7.8 99.9% 50.0% 40.5 9.6 100.1%

 

Note: Percentages may not add to 100.0% due to rounding
errors.

job since marriage was clerical or professional was much
higher in Egypt than in Bangladesh (25.3 percent and 3.6
percent, respectively; United Nations, 1987:264). As
such, a higher proportion of Egyptian women worked away
from home for nonrelatives and likely faced incompati-
bilities between their roles as mothers and workers.

In Table 6, the relationship between son preference
and children ever born is shown. Although the majority of
women 1J1 both countries had 1“) son preference, this
majority was larger in Egypt (79.8 percent) than in
Bangladesh (58.7 percent), a fact consistent with our
argument that women's gender status improves with economic
development. In both countries, women preferring sons
were more likely to have borne four or more children than

were women not preferring sons.

74

Table 6--Son preference, distributed by number of children
ever born, in Egypt and Bangladesh.

 

Number of children ever born

 

Egypt Bangladesh

 

Son
Preference 0-3 4-8 9+ Total 0-3 4-8 9+ Total

 

Yes 747 697 128 1572 995 898 223 2116
47.5% 44.3 8.1 99.9% 47.0% 42.4 10.5 99.9%

No 3162 2556 481 6199 1567 1177 268 3012
51.0% 41.2 7.8 100.0% 52.0% 39.1 8.9 100.0%

 

Total 3909 3253 609 7771 2562 2075 491 5128
50.3% 41.9 7.8 100.0% 50.0% 40.5 9.6 100.1%

 

 

 

 

Note: Percentages may not add to 100% due to rounding
error.

Testinggthe Hypotheses

 

The remaining section of this chapter is devoted to
testing the hypotheses stated in Chapter 11. Two
statistical techniques are used to examine the hypotheses
under investigation. The first two hypotheses are tested
by using multiple regression analysis, and the remaining
three hypotheses are tested using multivariate analysis of
covariance.

Hypothesis 1: The first hypothesis stated that wives

 

who have greater economic status (i.e., higher levels of
education or gainful employment) have fewer children ever
born alive. I first regressed the number of children ever
born to Egyptian respondents upon their education level,
gainful employment, and the three control variables (their

age at first marriage and their age and type of residence

75

at interview: see Table 7). Most of the variance in the
dependent variable is explained by the wife's age (beta =
0.679) and age at first marriage (beta = -0.378). While
the beta coefficient for wife's educational attainment is
smaller (-0.098), it is statistically significant at the
0.001 level and is similar in absolute value to the beta
weight for urban-rural residence (beta = (L126). {All of
these predictor variables together determined 60.04
percent of the variance in number of children ever born.
Gainful employment in Egypt makes no difference in the
number (ME children ever born, as shown in Table 7.
However, this variable has a small effect on the number of
children ever born when tested separately from educational
attainment. The reason it loses its significance is
probably due tx> the high correlation between gainful
employment and education (r = .433). These results for
Egypt partially support Hypothesis 1.

Hypothesis 1 was also tested for Bangladesh
(Table 8). The results in Table 8 are somewhat similar to
the regression analysis for Egypt in Table 7 except that
education is not significant and employment is
statistically significant at the 0.05 level. Apparently,
higher education in Bangladesh was associated with a much
younger age in Bangladesh than Egypt (r = -0.14 and -0.07,
respectively); thus, when the latter variable was present
in the multiple regression, it absorbed all the inverse

effect of education upon the number of children ever born

76

Table 7--Results of regression of the number of children
ever born on educational attainment and gainful
employment, controlling for age, age at first
marriage, and rural-urban residence in Egypt.

 

 

 

 

 

Independent variable B Beta T Sig. T
Age at first marriage -.248 -.378 -40.65 .000
Age at interview .223 .679 80.98 .000
Urban-rural residence .691 .126 14.18 .000
0 = urban
l = rural
Gainful employment -.041 -.006 -.63 .530
0 = work for cash
1 = do not work
Educational attainment -.318 -.098 -9.90 .000
O = no education
1 = up to primary
2 = more than primary
Intercept 1.367 8.08 .000
Coefficient of determination, R2 = 0.6004*
*significant at the 0.001 level
to Bangladeshi women. In fact, much of the variance in

the number of children ever born was explained by age at
first marriage and age at interview (beta = -0.l34 and
0.779, respectively). However, we cannot completely
reject Hypothesis 1 for Bangladesh since gainful
employment in Bangladesh explains some of the variance in
children ever born.

In addition, the urban-rural residence is not
statistically significant (see Table 8) in Bangladesh, but
it was very significant in Egypt (see Table 7). It is

generally expected that fertility is higher in rural than

77

Table 8--Results of-regression of the number of children
ever born on educational attainment and gainful
employment, controlling for age, age at first
marriage, and rural-urban residence in
Bangladesh.

 

 

Independent variable B Beta T Sig. T
Age at first marriage -.158 -.134 -15.31 .000
Age at interview .259 .779 91.73 .000
Urban-rural residence -.112 -.015 -l.75 .080
O = urban
1 = rural
Gainful employment .180 .017 2.02 .043
0 work for cash

1 do not work
Educational attainment -.069 -.014 -1.60 .109
0 = no education
1 = up to primary
2 = more than primary

2

Coefficient of determination, R = 0.642*

 

 

 

*significant at the 0.001 level

in urban areas. 'However, women in urban Bangladesh show
similar fertility rates to those in rural areas. This
similarity might be caused by the large number of urban
women who migrated from rural areas in Bangladesh (65.24
percent of women who lived in urban areas at the time of
the survey were originally from rural areas), but this was
not the case in Egypt, where 15 percent of the urbanites
had originated in rural areas. Those rural-to-urban women
migrants probably carried with them the strong values for
children, since children are an asset to those who depend

economically on agriculture.

78

Mpothesis 2:' The second hypothesis stated that
wives who have greater gender status (i.e., a non-Muslim
religion or £1 weak preference for sons) have fewer
children ever born alive. Two models have been used in
order to test Hypothesis 2: one for the Egyptian sample
and one for the Bangladeshi sample.

Table 9 illustrates the results of regression
analysis of number of children ever born on religion and
son preference in Egypt, controlling for age, age at first
marriage, and urban-rural residence. It is clear from the
table that although time beta weights for both son
preference and religion are small (0.019 and -0.024,
respectively), they are statistically significant at the
0.05 and 0.01 levels, respectively. The regression
analysis of Hypothesis 2 for Egypt indicates that Muslim
women in Egypt have a higher number of children ever born
than (M) non-Muslim women. Also, Egyptian women who
preferred sons had higher fertility than those with no son
preference. This result supports Hypothesis 2, that
Egyptian women with greater gender status have fewer
children ever born.

Table 10 shows the regression analysis of number of
children ever born on religion and son preference,
controlling for age, age at first marriage, and
urban—rural residence :h1 Bangladesh. As reported for
Egypt, son preference and religion play an important role

in predicting the number of children ever born. The beta

79

Table 9--Results of regression of the number of children
ever born on religion and son preference,
controlling for age, age at first marriage, and
rural-urban residence in Egypt.

 

 

Independent variable B Beta T Sig. T
Age at first marriage -.268 -.407 -47.94 .000
Age at interview .229 .697 84.63 .000
Urban-rural residence .832 .152 17.76 .000
0 = urban
1 = rural
Son preference .130 .019 2.34 .019
0 = no son preference
1 = son preference
0 = Muslim
1 = other
Intercept 1.171 8.39 .000

2

Coefficient of determination, R = 0.595*

 

 

 

*significant at the 0.001 level

weight for son preference is statistically significant at
the 0.05 level, and the beta weight for religion is
significant at the 0.001 level. This means that Muslim
women in Bangladesh had higher fertility than non-Muslim
women, and those with son preference had a higher number
of children ever born. As in the case for Egypt, much of
the variance in the dependent variable (children ever
born) is explained by the wife's age at interview and age
at first marriage. I-bwever, Hypothesis 2 should not be
rejected. Hence, the greater the gender status of women
in both Egypt and Bangladesh, the fewer the number of

children ever born alive.

80

Table 10--Results of regression of the number of children
ever born on religion and son preference,
controlling for age, age at first marriage, and
rural-urban residence in Bangladesh.

 

 

Independent variable 8 Beta T Sig. T
Age at first marriage -.l60 -.135 -15.94 .000
Age at interview .259 .782 93.34 .000
Urban-rural residence -.075 -.010 -1.20 .229

0 = urban
l = rural

Son preference .125 .019 2.36 .018
0 = no son preference

1 = son preference
0 = Muslim
1 = other
Intercept -.960 -5.62 .000

Coefficient of determination, R2 = 0.644*

 

 

 

*significant at the 0.001 level

Hypothesis 3: The third hypothesis stated that the

 

relationship posited in Hypothesis 1 will be much weaker
in Egypt for Muslims than for non-Muslims and for women
who prefer many sons than for women who prefer few sons,
but this relationship should also be true in Bangladesh.

By using multivariate analysis of covariance
(MANOCOVA), we can test Hypothesis 3 by means of the
statistical significance of four interaction terms:
education by religion, education by son preference, work
by religion, and work by son preference. In addition, the

covariates (urban-rural residence, age at interview, and

81‘
age at first marriage) are included in the model (Table

11). The only one of the interaction effects to attain

statistical significance is work by religion (F 18.97,
significant at the 0.001 level). This interaction
indicates that the effect of women's employment on number
of children ever born depends on religious affiliation in
both Egypt and Bangladesh.

In order to examine the significant interaction term
(work by religion) shown in Table 11, we disaggregated the
variance in fertility jointly explained by work status and
religion into three orthogonal vectors. Although Table 12
shows statistical significance for all three contrasts,
the differences are clear when we look at the optimal
Scheffe contrast coefficient and adjusted means for the
contrasts. The coefficient for the first contrast (i.e.,
working women vs. non-working women regardless of
religion) is -0.748 and is significant at the 0.001 level.
Working women in both countries had fewer children ever
born alive. The coefficient for second contrast (Muslim
women who work versus non-Muslim women who work) was
0.7733 and it is statistically significant at the 0.001
level. This indicates that Muslim women who worked for
cash had more children than non-Muslim women who also
worked for cash in both Egypt and Bangladesh. Finally,
the coefficient for Muslim women and non-Muslim women who
did not work was 0.1839: it is significant at the 0.002

level. In other words, among women who did not work for

 

 

82

Table ll—Surrmary table for MANOCOVA of the number of children ever
born on educational attainment, gainful employment,
religion, son preference, and country with urban-rural
residence, age, and age at first marriage as covariates and

the interaction terms .

 

 

 

Source of Variation SS 11‘ MS F Sig. of F
Within cells 36849.55 11309 3.26
Regression 50990. 80 3 16996. 93 5216 . 30 . 000
Constant 32. 65 1 32 . 65 10. 02 . 002
Education 40 . 55 2 20. 28 6. 22 . 002
0 = no education
1 = up to primary
2 = more than primary
Religion 129.46 1 129. 46 39. 73 .000
0 = Muslin
1 = other
Employment 48.63 1 48.63 14.92 .000
0 = work for cash
1 = do not work
Son preference 17.40 1 17.40 5.34 .021
0 = no son preference
1 = son preference
country ' 2.08 1 2.08 .64 .424
0 = Bangladesh
1 = Egypt
Education by religion 4.31 2 2.15 .66 .516
Work by religion 61.80 1 61.80 18.97 .000
Education by country 186.34 2 93.17 28.59 .000
Work by country 1.95 l 1.95 .60 .439
Religion by country .04 1 .04 .01 .909
Son preference by country .00 1 .00 .00 .992
Son preference by education 2.43 2 1.22 .37 .688
Son preference by employment .15 l .15 .05 .829

 

 

Table 11, continued.

Controlling for:

83

 

Covariate 8 Beta Std. Error tvvalue Sig.'of t
Urban-rural -

residence .39715 .06195 .040 10.007 .000
Age .23965 .74190 .002 120.449 .000
Age at first .

marriage -.22472 -.25346 .005 -41.l99 .000

 

 

 

Table 12--Interaction between religion and gainful enplowent,
controlling for other variables.

 

 

 

 

Source of Variation SS DF MS F Sig. of F
Within cells 36845.55 11309 3.26
mgression 50990. 80 3 16996. 93 5216. 30 . 000
Contrast 1 79.31 1 79.31 24.34 .000
Contrast 2 109.55 1 109.55 33.62 .000
Contrast 3 31.13 1 31.13 9.55 .002
Education 291.41 2 145. 71 44. 72 .000
Son preference 40.37 1 40.37 12.39 .000
Country 125. 56 1 125. 56 38. 53 . 000

I Contrast 1 = working vs. non-working women, regardless of religion

(bntrast 2 = Muslin women vs. non-Muslim women who work for cash

Contrast 3

Muslin mmen vs. non-Muslim women who did not work

 

84
cash, Muslim women had more children ever born than
non-Muslim women: but the difference was smaller than that
between Muslim and non-Muslim women who worked for cash.

To illustrate this work-religion interaction more
clearly, Table 13 shows the adjusted means for religion
and work. The highest number of children ever born was
for Muslim women who did not work for cash (x3 = 3.876),
and the lowest number of children ever born was for
non-Muslim women who did work for cash (x2 = 3.023).
Graphing these adjusted means (Figure 1) illustrates that
the interaction between gainful employment and religion
arises because work for gain is more sharply antinatalist
for non-Muslim than Muslim wives. This relationship may
signify that their higher fertility propels Muslim women
more often to work from need or that Muslim women less
often retain control over their wages. If either
possibility were true, then Muslim women would have fewer
work-induced incentives to keep their fertility in check.
These results provide partial support for Hypothesis 3.

Hypothesis 4: The fourth hypothesis stated that the

 

relationship proposed in Hypothesis 1 will be stronger in
Egypt, where economic development has allowed a more
pervasive rise in women's economic status. The above
hypothesis anticipates interactions between women's
education, employment and country. In other words, the

effect of educational attainment or working for cash on

Table 13--Combined adjusted* means of children ever born
for the interaction of work and religion,

85

Egypt and Bangladesh.

 

Work Status

 

 

 

Religion Working for cash Not working for cash
Muslim x1 = 3.7965 x3 =
Others x2 = 3.023 x4 =

 

 

 

*These means have been adjusted for age at marriage,
at interview,

lhmbercfi
ohikken
ewerlxxn

Figure l--Interaction between work and religion.

4.0
3.9
3.8
3.7
3.6
3.5
3.4
3.3
3.2
3.1
3.0
2.9

 

and urban-rural residence.

X

bkmlhn 3
x
l-/Beta=
.1839
erMwflJm
Ram?
.7733
x2

age

 

wofldhgimmen

marwofldng1mmen

86
the number of children ever born should depend on country
of origin.

The multivariate analysis of covariance shown in
Table 11 enables us to test the interaction terms stated
in Hypothesis 4 (education-by-country, work-by-country).
Table 11 shows that the work by country interaction term
is not statistically significant. This indicates that the
depressant effect of women's work for cash on fertility is
about equally strong in Egypt and Bangladesh. As stated
earlier, Islam is dominant in both Egypt and Bangladesh.
the Islamic religion places many restrictions on women's
work outside the home. For example, Muslim women must be
veiled when they are outside the home and they cannot work
together with men. These restrictions mean that a
majority of gainfully employed women in both Egypt and
Bangladesh work in sex-segregated occupational structures:
for examples, women trade with other women; women teachers
instruct female students; and women physicians deliver
health care to female patients. These realities may
account for the similar effect of women's work on
fertility in Egypt and Bangladesh.

On the other hand, the strength of the depressant
effect of women's education on their number of children
ever born differed between Egypt and Bangladesh
(F-statistic for education by country interaction = 28.59,
which is significant at the 0.001 level). To understand

the form of this statistical interaction, we computed an

87
analysis of covariance in which the covariates were
women's age at interview, urban-rural residence, _and age
at first marriage. The main effects were work, religion,
son preference, and five orthogonally coded vectors which
disaggregated the variance in fertility jointly explained
by women's education and country (Table 14).

Table 14 shows that all orthogonal contrasts were
statistically significant except contrast 4 (Egyptian
women versus Bangladeshi women, with up to a primary level
of schooling completed). Table 15 shows the coefficients
for the contrasts. Contrast 3 (women in Bangladesh with
no education versus the same group in Egypt) and contrast
5 (Bangladeshi women with more than a primary level of
schooling versus the same group in Egypt) appeared to be
the most important elements in the interaction term
between education and country factors. In other words,
Egyptian women with no education had a higher adjusted
mean number of children ever born (4.149) than the same
group in Bangladesh (3.625): but Bangladeshi women with
more than a primary level of schooling had a higher
adjusted mean (3.656) than the same group in Egypt
(3.367)(see TabLe 16). The graph of these adjusted means
(Figure 2) illustrates that education more sharply
depresses fertility in Egypt than in Bangladesh, as was
posited in Hypothesis 4.

The adjusted mean for the education by country

interaction term provides additional clarification of the

88

Table 14--Contrast procedure in MANOCOVA of the effect of level of
education and country of origin on children ever born, in
Egypt and Bangladesh, controlling for other variables.

 

 

 

 

 

Source of Variation SS 11‘ _ ‘ MS F Sig. of F
Wfithin cells 36849.55 11309 3.26

Regression 50990.80 3 16996.93 5216.30 .000
Contrast 1 12.20 1 12.20 3.74 .053
Contrast 2 112.80 1 112.80 34.62 .000
Contrast 3 288.61 1 288.61 88.57 .000
Contrast 4 3.95 1 3.95 1.21 .271
Cbntrast 5 25.73 1 25.73 7.90 .005
work 5.56 2 5.56 1.71 .192
Son preference 39.33 1 39.33 12.07 .001
Religion 79.72 1 79.72 24.46 .000
Contrast 1 = no education vs. up to ccmpleted primary level
Contrast 2 = no education vs. more than a primary education level
Contrast 3 = No education: Bangladesh vs. Egypt

Contrast 4 = Up to completed primary: Bangladesh vs. Egypt

Contrast 5 = More than primary level of education: Bangladesh vs. Egypt

Table 15--Coefficients for the contrasts in Table 14.

 

 

Contrast Coefficient T Sig. T
Contrast 1 0.1653 1.935 0.053
Contrast 2 1.335 5.883 0.000
Contrast 3 -.524 -9.411 0.000
Contrast 4 -.085 -1.101 0.271
Contrast 5 .289 2.809 0.005

 

 

 

lhmbercfi
chihhen

ever born

4.2
4.1
440
3.9
3.8
3.7
3.6
3.5
3.4
3.3
3.2

3.1

 

89

 

 

x6
no up to conpleted more than
ahxsthx1 prhmmy prhmmy

Figure 2-—Interaction between education and country.

Table 16--Adjusted* means for the effect of education on
children ever born in Egypt and Bangladesh.

 

 

 

 

Country
Education level Bangladesh Egypt
No education x1 = 3.625 x2 = 4.149
Up to completed
primary x3 = 3.762 x4 = 3.847
More than primary x5 = 3.656 x6 = 3.367

 

 

 

 

*These means have been adjusted for age at marriage, age

at interview,

and urban-rural residence.

90
effect of education on fertility, which depends on the
country of origin.

Table 16 illustrates how the effect of educational
attainment on children ever born depends on country of
origin. The table shows that Egyptian women with no
education have higher fertility than Bangladeshi women.
The difference in number Of children ever born, on
average, to Bangladeshi and Egyptian women with more than
a primary level of education (3.656 and 3.367,
respectively), is statistically significant at the 0.01
level. This indicates that Egyptian women with more than
a primary level of schooling have lower fertility than the
same group in Bangladesh. In this case antinatalist
conditions in Egypt (1 assert that these conditions are
created by economic development) combine with the effects
of having secondary or tertiary schooling to reinforce the
antinatalist consequences of the latter. Figure 2 shows
the interaction between educational attainment and country
of origin.

An interpretation of this interaction term can be
ventured. Egyptian women with no education probably have
more children ever born than the Bangladeshi women because
fertility usually rises in the early stages of economic
development due to improved health conditions which reduce
subfecundity and intrauterine mortality. Although
Egyptian women with some primary education appear to have

higher fertility than the same'group in Bangladesh, the

91

difference is very small and not statistically
significant. (M1 the other hand, Egyptian women with more
than a primary level of education have lower fertility
than their counterparts in Bangladesh probably because
this level of education in Egypt is more apt to lead women
into employment as clerks or professionals (United
Nations, 1987:264). Education beyond the primary level
does not have the same payoff in Bangladesh, since its
lower level of economic development means less opportunity
for both men and women to work in the modern sector.

Hypothesis 5: The final hypothesis stated that a
non-Muslim religion or a weak preference for sons' would
more strongly depress fertility in Egypt, where economic
development has lessened the dependency of women upon men.
This hypothesis assumes two interactions: one between the
country variable and religion, and one between country and
son preference. When the design of multivariate analysis
of covariance was created (see Table 11) to test for
interaction in the above relationship, no such
interactions were found. Although Table 11 shows a
significant effect of both religion and son preference on
the number of children ever born alive, the strengths of
these effects were about the same for the two Muslim

countries. These findings refute Hypothesis 5.

CHAPTER V
SUMMARY, CONCLUSION AND RECOMMENDATIONS

Egypt and Bangladesh have several demographic
similarities. The majority of both populations are
Muslims. In both countries policymakers are concerned
about population problems: birth rates as well as death
rates are high in both Egypt and Bangladesh. Several
measures have been taken in both countries to reduce the
high birth rates. In Egypt, for example, as early as 1965
family planning programs were started to improve the
population's health and reduce the high birth rate. After
several years, because these programs were not successful,
the Egyptian government inaugurated the population and
development program project, with similar goals of family
planning. This project enjoyed some success in improving
the health conditions of the population and increasing the
number of contraceptive users in Egypt. Bangladesh
employed a strategy similar to Egypt's for reducing its
high birth rate. The national family planning program was
established in the early 19605. About a decade later, as
in the Egyptian case, very little success had been
achieved in reducing the fertility rate or increasing the

prevalence of family planning use. In fact, only eight

92

93
percent of married women were using contraception ten
years after the establishment of the family planning
program in Bangladesh (Sufian, 1984).

From the experience of family planning programs, both
Egypt and Bangladesh have had very limited success in
reducing high birth rates. This study turns the focus
away from programs as such and toward the important issue
of the status of women and its relation to fertility and
fertility behavior. Previous research studies (see, for
example, Anker, 1982; Mason, 1987) support the argument
that the investment in raising the status of women can
reduce the fertility rate more effectively than many
techniques of fertility control.

Egypt and Bangladesh are well-chosen for comparing
the effects of two dimensions of women's status on
fertility, since these countries are at different levels
of economic development. The two dimensions were women's
economic status and their gender status. I posited that
the level of a nation's economic development would place
limits on the amount of economic status that women (or
men) could achieve. Since schooling imparts marketable
skills, I hypothesized that women's education would more
strongly depress fertility in Egypt than in Bangladesh,
because Egypt is at a higher level of economic
development.

Indeed, education was found to be an important factor

reducing fertility in the Egyptian case. Although the

94

educational factor was a weak indicator of lower fertility
in Bangladesh, its effect was highly related to women's
ages at interview: younger women had fewer children but
also more years of schooling. This finding may portend a
future emergence of an inverse relationship between
education and fertility for younger cohorts of Bangladeshi
women. These results are consistent with the finding from
38 nations participating in the World Fertility Survey:
that the relationship of education to fertility becomes
more sharply inverse as the level of economic development
becomes higher (United Nations, 1987). An implication for
policymakers is that a retention of girls in school
systems beyond the primary level is necessary for an
appreciable impact on fertility in countries as poor as
Bangladesh or Egypt. However, the Muslim religion does
not buffer the growing negative impact of education on
fertility at higher levels of economic development, as
Caldwell (1980) had speculated that it might.

Gainful employment, 'the second indicator of women's
economic status, was related to fertility in Bangladesh,
but not in Egypt. Women in Bangladesh who worked after
marriage had fewer children ever born alive than those who
did not work, probably because working gave women more
material resources and made them less economically
dependent on men and children. Dixon (1985) contends that
women's involvement in the labor force (especially when

they work for nonrelatives) gives them an independent

95

power base, provides a direct source of income over which
they have control, and reduces their dependency on male
kinfolk for economic and social support. Dixon suggests
that expansion of employment opportunities for females in
low-income households is not only a valuable end in itself
but also a means to enable women to transcend the inherent
pro-natalism of a patriarchal kin-based institution
(p. 16).

An important finding of this research was that
independently of women's economic characteristics, their
membership in a religious grouping assigning lower
statuses to women (Muslims) or their psychological
commitment to the higher status of males over females (as
indicated by a preference for sons over daughters) was
associated with higher fertility. In fact, membership in
the Muslim religion dampened the negative effect that
women's work had on their fertility. In other words,
employment of women reduced the fertility of Muslims much
less than that of non-Muslims. Perhaps the social
distance between men and women required by Islam restricts
the occupational choices of Muslim women wishing to enter
the labor force (since they cannot have male coworkers or
colleagues) and relegates them to low-prestige jobs
inviting weaker career commitments. 'Or perhaps Muslim
women wishing to keep purdah are more reluctant than

non-Muslim women to work far away from home, and the

96
proximity to the homesite reduces the role conflicts
between parenthood and employment.

It is probable that Muslim women have a lower gender
status than do others because they are given limited
social rights and because males are dominant in most
Islamic nations. However, in general, the principles of
Islam neither regard women as inferior to men nor place
restrictions on women's work or education. Rather,
particular Muslim cultures and traditions impose these
restrictions and resist gender equality. weeks (1988), in
a discussion of women in the history of Islam, informs us
that "In a world in which women were often little more
than chattel or slaves, Islam accorded to women secure
legal rights, including independent legal status, the
right to own and dispose of property, and the right to
receive a share of inheritance" (p. 7).

Son preference was also used as an indicator of
women's gender status in this study. The results of this
study show that women with son preference had more
children ever born in both Egypt and Bangladesh. This
finding suggests that male children are more valuable than
female children which, in turn, leads to higher
fertility. Sheps (1963) argued that the strong desire for
sons influences the total number of children wanted by a
couple. This may increase the number of children ever
born alive until the desired number of sons have been

born. One implication regarding the son preference

97

finding is that improving the social security systems in
both countries may benefit people, especially in old age,
and reduce their dependence on sons for their social and
economic needs. In addition, improving the status of
females by formal education and involvement in the labor
force participation may reduce the high value placed on
male children which, in turn, may negatively influence
fertility rates.

However, an implication for development policy is
that policies that create jobs for all women--Muslim and
non-Muslim--who want to work after marriage will be
followed by a fertility decline. Policies that improve
women's earning capacities might also soften a preference
for sons. Nevertheless, the People's Republic of China
has used direct propaganda to proclaim the value of
daughters and this frontal attack on son preference has
probably spurred the impressive fertility decline there.

In addition tx: the results discussed above, the
control variables, women's age at first marriage and
women's age at interview, accounted for most of the
explained variance in the number of children ever born in
both Egypt and Bangladesh. f Women's age at first marriage
was found to be inversely related to the number of
children ever born. Also, the older the women, the
greater the number of children ever born. The other
control variable, urban-rural residence, was found to be

important in Egypt, where rural women had higher fertility

98

than urban women, but this relationship was weak in
Bangladesh, where the majority of urban women had migrated
from rural areas and probably brought with them strong
values for large families. A similar result was observed
by Ahmad (1985) in his study of. urban-rural fertility in
Bangladesh, where he found no significant differential in
marital fertility by place of residence.

Another policy implication that should be considered
in both Egypt and Bangladesh is to raise the age at first
marriage. This has been a successful policy for reducing
fertility in China (Coale et al., 1991). However, raising
the age at first marriage may be difficult to apply in
countries with a lack of birth registration or those that
have other social and religious obstacles to enforce
raising the age at first marriage. Another possibility is
to delay the age at first birth. Policymakers may
emphasize the health risks of both the mother and the
child if births occur before the mother is 17 years old.

With regard to urban-rural residence, coordination
between population policymakers and development planners
is an important issue influencing population growth.
Equity of income distribution and other social and
economic services in urban and rural areas is a crucial
issue for improving lifestyle, especially in rural areas.
Clausen (1984), for example, observed that the subsequent
decline of fertility in developing countries was a result

of a "different kind of development: education, health,

99

the alleviation of' poverty, improved opportunities for
women, and government effort to assure widespread access
to family planning services" (p. 18). Sri Lanna,
Thailand, and Turkey were cited as examples of countries
experiencing fertility decline, where income and other
social and economic services were relatively more evenly
distributed among the population.

The findings of this study confirm that both economic
status and gender status covary inversely with the number
of children ever born to women in both Egypt and
Bangladesh. Policymakers in both countries should direct
attention to the improvement of women's status: without
such improvements, women's status depends largely on the
number of children they bear (Germain, 1975).
Availability of, access to, and encouragement of female
education, especially in rural areas, are crucial steps to
improving women's economic status. The involvement of
women in economic development will also benefit the status
of women, beginning with the provision of more employment
opportunities in both urban and rural areas. Women's
situations, in relation to men's, will improve when they
are provided increased access in) material and social
resources. In addition, in) forestall women's dependence
on men and to provide alternative sources of old-age
security, social security systems can be improved, and in

some cases created especially for women.

100
Sgggestions for Further Research

This study has investigated the influence of women's
status on fertility. While it is successful in answering
some of the major research concerns, it leaves some
questions unanswered and raises additional ones that need
to be addressed. It is clear from the findings of this
study that the status of women is one of the most
important factors influencing fertility and fertility
behavior. The status of women is significantly associated
with the number of children ever born. Women who had some
indicators of higher status, such as education,
employment, or no son preference, had a lower fertility
rate. However, the measurement of the status of women in
this study depended on and derived from otherwise
available data, rather than being the object itself of
precise and accurate measurement. Along with the
important variables used in this study to measure women's
status, other variables are needed for more comprehensive
measurement of the status of women.

The results of this study lead.tx) the strong
recommendation that further research on the status of
women and fertility explore more fully the effect of the
status of women on reducing or raising the fertility
rate. These future studies should include many more
variables in order to give a more extensive measurement of
women's status, which in turn will contribute beneficial

information to the fertility literature and to

101

policymakers. Variables such as income, general decision
making (including familial decision making), and
reproductive decision making might be included. Future
research should ask about the nature of work women do and
what type of returns they receive. It would be
interesting to explore whether women paid in cash rather
than in kind have lower fertility, since cash is a more
versatile, liquid asset. It should also be interesting to
see whether women who retain control of their cash income
have lower fertility than women who deposit their earned
money into a common family pot. It has been suggested
that higher economic status empowers the wife to be more
involved in reproductive decision making. When both
spouses decide the timing, as well as the number of
children they should have, their fertility must be
different than when only the husband makes the
reproductive decisions.

In this study, religion was found to be an important
independent predictor of the number of children ever
born. ‘ It seems that levels of religiosity may have
different effects on both women's status and fertility
behavior. There are certain core religious duties that
must be observed by all Muslims, such as the five daily
prayers, paying the zakat tax, and fasting during the
month of Ramadan. However, there are many peripheral
religious norms in Islam. For example, not all Muslim

women are requested to cover their entire body when they

102

go outside the home. Also, some Muslim women can work
outside the home whereas others must stay at home and look
after the children. In addition, the use of contraception
is permitted for some Muslim women but forbidden for
others. All these norms depend on the husband's level of
religiosity. It is recommended that future research not
only include religion as an important variable but also
measure the degree of husband's religiosity within each
religion. For example, future research may ask if a woman
is required by her husband to cover her entire body, not
to work outside the home, and not practice birth control.

In future research, other factors related to women's
status which cluster around women's control of self and
property need to be considered. The age difference
between spouses can be an important indicator of women's
status. A large difference between spouses' ages at first
marriage, for example, may indicate an arranged marriage,
which may indicate a lower status for the woman. This
leads to another important factor related to the status of
women, especially in the developing countries: women's
freedom of choice regarding spouse and time of marriage.

In addition, the use of age at cohabitationlrather
than age at marital contract would offer a more sensitive
prediction of the number of children ever born. Freedom
of movement is also a good indicator of women's status
that needs to be included in future research. In some

instances, women with higher status may go outside the

103
home whenever it is possible, while lower-status women are
restricted to the home most of the time.

Future research might include the husband's opinion
about women, in general, and his wife, in particular. His
views on women's decisions in the family, the types of
work they should do, and other related issues may
contribute useful information to the issue of the status

of women and fertility.

REFERENCES

Ahmad, S.
1985 Rural-urban differentials in marital fertility in

four Moslem populations. Journal of Biosocial
Science 17(2):157-165.

Ahmed, B.
1981 Differential fertility in Bangladesh: a path
analysis. Social Biology 28:102-110.

 

Ahmed, B.

1987 Determinant of contraceptive use in rural
Bangladesh: the demand for children, supply of
children, and costs of fertility regulation.
Demography 24(3):361-373.

 

Ahmed, N.
1981 Family size and sex preference among women in

rural Bangladesh. Studies in Family Planning
12:100-109.

 

Allman, J.
1978 Family life, women's status, and fertility:
Middle East and North African perspectives. In:
Women's Status and Fertility in the Muslim World,
edited by J. Allman. New York: Praeger.

Amin, R., and R. Farugee
1980 Fertility and its regulation in Bangladesh. World
Bank Staff Working Paper No. 383. washington,
D.C.: The World Bank.

Amin, R., and A. Mariam
1987 Son preference in Bangladesh: an emerging barrier
to fertility regulation. Journal of Biosocial
Science 19:221-228.

 

Anderton, D., and L. Bean
1985 Birth spacing and fertility limitation: a
behavioral analysis of nineteenth century
frontier populations. Demography 22(2):169-183.

Anker, R.
1982 Demographic change and the role of women: a
research programme in developing countries. In:
Women's Roles and Population Trends in the Third
World, edited by R. Anker, M. Buvinic, and N.
Youssef. London: Croom Helm.

104

105

Arnold, F.
1985 Measuring the effect of sex preference on

fertility; the case of Korea. Demography
32(2):280-288.

 

Balasubramanian, K.
1977 Socio-cultural factors influencing fertility in
India. In: The Economic and Social Support of
High Fertility, edited by L. Ruzicka. Canberra:
Australian National University.
Basu, A.
1989 Is discrimination in food really necessary for
explaining sex differentials in childhood
mortality? ngulation Studies 34(2):l93-210.

Blalock, Hubert M., Jr.
1972 Social statistics. New York: McGraw-Hill.

Behrnsted, G., and D. Knoke
1982 Statistics for social data analysis. Chicago:
F.E. Peacock Publishers, Inc.

Bray, J., and S. Maxwell
1985 Multivariate analysis of variance. Newbury Park,
California: Sage Publications.

Cain, M.
1977 The economic activities of children in a village

in Bangladesh. Population and Development Review
3:201-228.

 

1981 Risk and insurance: perspectives on fertility and
agrarian change in India and Bangladesh.
Population and Development Review 7(3):435-474.

Cain, M., S. Khanam, and S. Nahar
1979 Class, patriarchy, and work in Bangladesh.
Population and Development Review 5:405-438.

Caldwell, J.
1976 Toward a restatement of demographic transition

theory. Population and Development Review
2:321-366.

1979 Education as a factor in mortality decline: an
examination of Nigerian data. Population Studies
33(3).

 

1980 Mass education as a determinant of the timing of
fertility decline. Population and Development
Review 6(2).

 

106

1981 Fertilty decline in Africa. In: Fertility decline
in the less developed countries, edited by N.
Eberstadt. New York: Praeger Publishers.

1982 Theory of Fertility Decline. New York: Academic
Press.

1986 Routes to lowrnortality in poor countries.
Population and Development Review 12(2).

Caldwell, J., P. Reddy, and P. Caldwell
1983 The social component of mortality decline: an
investigation in South India employing alternative
methodologies. Population Studies 37:185-205.

Callum, C., S. Farid, and M. Moussa
1988 Child 1055 and its impact on fertility. In: Egypt:
Demographic Responses to Modernization, edited by
A. Hallouda, S. Farid, and S. Cochrane. Cairo:
Central Agency for Public Mobilisation and
Statistics.

Cassen, R.
1978 India: population, economy, society. New York:
Holmes and Meier.

Chamie, J.
1981 Religion and fertility: Arab Christian Muslim
differentials. Cambridge: Cambridge University
Press.

Chen, L., E. Huq, and S. D'Souza
1981 Sex bias in the family allocation of food and
health care in rural Bangladesh. Population and
Develgpment Review 7(1).

Chowdhury, A., A. Khan, and L. Chen
1978 Experience in Pakistan and Bangladesh. In: The
Effects of Infant and Child Mortality on
Fertility, edited by S. Preston. New Yerk:
Academic Press.

Clausen, A.
1984 Population growth and economic and social
development. washington, D.C.: World Bank.

Coale, Ansley J., wang Eeng, Nancy E. Riley, and Lin En De.
1991 Recent trends in fertility and nuptiality in
China. Science 251 (January 25):389-93.

Cochrane, S.
1979 Fertility and education: what do we really know?
World Bank. Baltimore: The Johns H0pkins
University Press.

107

Cochrane, S., M. Khan, and I. Osheba
1990 Education, income, and desired fertility in
Egypt: a revised perspective. Economic
Development and Cultural Changg 38(2):313-339.

Das, N.
1987 Sex preference and fertility behavior: a study of
recent Indian data. Demography 24(4):517-530.

 

Dixon, R
1985 New jobs needed for rural women. Populi 12(2):
16-28.

Dyson, T., and M. Moore
1983 On kinship structure, female autonomy, and
demographic behavior in India. Population and
Development Review 9(1):35-60.

 

Egyptian Demographic and Health Survey
1989 Egypt National Population Council: Demographic
and Health Surveys. Columbia, Maryland: Institute
for‘Resource Development/Macro System, Inc.

El-Deeb, B., and J. Casterline
1988 Determinants of contraceptive use. In: Egypt:
Demographic Responses to Modernization, edited by
A. Hallouda, S. Farid, and S. Cochrane. Cairo:
Central Agency for Public Mobilisation and
Statistics.

Entwisle, B., J. Casterline, and H. Sayed
1989 Villages as contexts for contraceptive behavior
in rural Egypt. American Sociological Review
54:1019-1034.

Entwisle, B., and W. Mason
1985 Multilevel effects of socioeconomic development
and family planning programs on children ever
born. American Journal of Sociology 19(3):
616-649.

Faour, M.
1989 Fertility policy and family planning in the Arab
countries. Studies in Family Planning
20(4):254-263.

Farid, S.
1987 A review of the fertility situation in the Arab
countries of Western Asia and Northern Africa.
In: Fertility Behaviour in the Context of
Development: Evidence from the World Fertility
Survey. New York: United Nations.

108

Farques, P.
1989 The decline of Arab fertility. Pepulation:
English Selection 1:147-175.

 

Gadalla, S.
1978 Is there hope? Fertility and family planning in a
rural Egyptian community. Cairo, Egypt: The
Social Research Center, The American University
in Cairo Press.

Gadalla, S., J. McCarthy, and 0. Campbell
1985 How the number of living sons influence
contraceptive use in Mehoufia Governorate, Egypt.
Studies in Family Plannipg 16(3):164-169.

Germain, A.
1975 Status and roles of women as factors in fertility
behavior: a policy analysis. Studies in Family
Planning 6:192-200.

 

Gray, E.
1972 Influence of sex of first two children on family
size. Heredity 63:91-92.

Gray, E. and M. Morrison
1974 Influence of combinations of sex of children on
family size. Heredity 65:169-174.

Harbison, S., T. Khalegue, and W. Robinson
1989 Female autonomy and fertility among the Garo of
north central Bangladesh. American Anthropologist
91(4): 1000-1007.

Haub, C., M. Kent, and M. Yanagishita
1990 1990 World population data sheet. washington,
D.C.: Population Reference Bureau.

Heer, D., and H. Wu
1978 Effects in rural Taiwan and urban Morocco:
combining individual and aggregate data. In: The
effect of infant and child mortality on
fertility, edited by 8. Preston. New Yerk:
Academic Press.

Heitzman, J. and R. Worden
1989 Bangladesh: a country study. Federal Research
Division, Library Congress. United States
Government.

Hess, P.
1988 Population growth and socioeconomic progress in
less developed countries. New Yerk: Praeger
Publishers.

109

Hirschman, C., and P'. Guest
1990 Multilevel models of fertility determination in
four Southeast Asian countries: 1970 and 1980.
Demography 27(3):369-396.

 

Hoffman, M. (ed.)
1991 The World Almanac and Book of Facts. New Ybrk:
Sc r ipps Ho ward .

Hoque, N.
1988 Rural electrification and its impact on
fertility: evidence from Bangladesh. Unpublished
dissertation, Pennsylvania State University.

Horne, A., and M. El-Khorazaty
1987 Childbearing indices in the Arab world.
Population Bulletin of ESCWA #31, United Nations.

 

Islam, S.
1982 Case studies of indigenous abortion practitioners
in rural Bangladesh. Studies in Family Planning
13(3): 86-93.

 

Jain, A.
1981 The effect of female education on fertility: a
simple explanation. Demography 18(4):577-595.

 

Johnson, N.

1977 Farmrnonfarm differentials in fertility: the
effects of compositional and sex-role factors.
Unpublished dissertation, Pennsylvania State
University.

Johnson, N., and A. Sufian
1991 Effect of son mortality on contraceptive practice
in Bangladesh. Journal of Biosocial Science
23(4): forthcoming.

 

Jones, G.

1978 Social science research on population and
development in Southeast and East Asia. In:
Pepulation Policy: Research Priorities in the
Developing World, edited by C. Miro and J.
Potter. London: Frances Printers.

Kabir, M., M. Uddin, S. Chowdhury, and T. Ahmed
1986 Characteristics of users of traditional
contraceptive methods in Bangladesh. Biosocial
Science 18:23-33.

Kasarda, J.
1971 Economic structure and fertility: a comparative
analysis. Demography 8(3):307-317.

 

110

Kelley, A., A. Khalifa, and M. El-Khorazaty
1982 Population and development in rural Egypt.
Durham, North Carolina: Duke University Press.

Kent, M, and A. Larson
1982 Family size preferences: evidence from the World
Fertility Surveys. Reports on the World Fertility
Survey (4). Washington, D.C.: Population
Reference Bureau.

Khan, M., and I. Sirageldin
1977 Son preference and the demand for additional
children in Pakistan. Demography 14(4): 481- -495.

 

Khan, 8.

1973 Factors affecting spacing of births. The Journal
of Family Welfare 20(2):54-67.

 

Kirk, D.
1966 Factors affecting Moslem natality. In: Family
Planning and Population Programs, edited by B.
Berelson. Chicago: University of Chicago Press.

1967 Factors affecting Moslem natality. Paper
presented to the International Conference on
Family Planning, Geneva, August 23-27, 1965.
Summarized in: Moslem Attitudes Toward Family
Planning. New Yerk: Population Council.

1967 Factors affecting Moslem natality. In: Population
and Society, edited by C. Nam. Boston: Houghton
Mifflin.

Knodel, J.
1987 Starting, stopping, and spacing during the early
stages of fertility transition: the experience of
German village populations in the 18th and 19th
centuries. Demography 24(2):l43-162.

 

Koening, M., J. Phillips, R. Simmons, and M. Khan
1987 Trends in family size preferences and
contraceptive use in Matlab, Bangladesh. Studies
in Familprlanning 18(3):117-127.

lee, S., and J. Kim
1979 The value of children: a cross-national study.
Honolulu, Hawaii: East-West Population Institute.

Lightbourne, R.
1985 Individual preferences and fertility behavior.
In: Reproductive Change in Developing Countries:
Insights from the World Fertility Survey, edited
by J. Cleland and J. Hobcraft. Oxford: Oxford
University Press.

111

Mannan, M.
1989 Family, society, economy and fertility in
Bangladesh. Bangladesh Development Studies
17(3):67-99.

Mason, K.
1987 The impact of women's social position on
fertility. Sociological Forum 2(4):7l8-745.

McNamara, R.
1977 Possible interventions to reduce fertility.
Population and Development Review 3(1):163-176.

Miller, B.
1981 The Endangered Sex: Neglect of Female Children in
Rural North India. Ithaca, New Yerk: Cornell
University Press.

Nagi, M.
1983 Trends in Moslem fertility and the application of
the demographic transition model. Social Biology
30(3):245-262.

1984 Trends and differentials in Moslem fertility.
Journal of Biosocial Science 16:189-204.

Nathanson, C.
1984 Sex differences in mortality. Annual Review of
Sociology 10:191-213.

Noman, A.
1983 The Status of Women and Fertility in Bangladesh.
Dhaka: University Press Limited.

Nugent, J.
1985 The old-age security motive for fertility.
Population and Development Review 11(1):?5-97.

Oyeka, I.
1989 Influence of the number of living sons on
contraceptive use among female teachers in
Nigeria. Sgpdies in Family Planning
20(3):l70-174.

Park, C.
1978 The fourth Korean child: the effect of son
preference on subsequent fertility. Journal of
Biological Science 10:95-106.

 

1983 Preference for sons, family size, and sex ratio:
an empirical study in Korea. Demography
20(3):333-352.

112

Preston, S.H. '
1978 Introduction. In: The Effects of Infant and Child
Mortality on Fertility, edited by S.H. Preston.
New York: Academic Press.

Rahman, M., J. Akbar, and J. Phillips
1990 Sex composition of children and contraceptive use
in Matlab, Bangladesh. Paper presented at the
Annual Meeting of the Population Association of
America, Toronto.

Rao, K., and R. Beaujot
1986 Effect of infant mortality on subsequent
fertility in Pakistan and Sri Lanka. Journal of
Biosocial Sciences 18(3):297-303.

 

 

Raymont, A.
1989 Development, 'welfare,‘ and fertility: a simple
macroanalysis. Social Biology 36(1-2):110-ll3.

 

Repetto, R. .
1972 Son preference and fertility behavior in

developing countries. Studies in Family Planning

 

Rindfuss, R., L. Bumpass, J. Palmore, and D. Woo Han
1982 The transformation of Korean child-spacing
practices. Pepulation Studies 36(1):87-104.

Rizk, H.
1963 Social and psychological factors affecting
fertility in the United Arab Republic. Marriage
and Family Living 25(1):69-73.

 

 

Rizk, I., C. Stokes, and M. Nelson
1982 The influence of individual and community-level
child mortality on fertility in Egypt. Studies in
ggmparative International Development
17(2):74-86.

 

 

Rosen, B., and A. Simmons
1971 Industrialization, family and fertility: a
structural-psychological analysis of the
Brazilian case. Demography 8(1):49-69.

 

Sayed, H., M. Osman, F. El-Zanaty, and A. way.

1989 Egypt Demographic and Health Survey. Egypt
National Population Council, Cairo, Egypt. Egypt
Institute for Research Development/Macro System,
Inc., Columbia, Maryland.

Schmida, L., and D. Keenum
1983 Education in the Middle East. washington, D.C.:
AMIDEAST.

113

Sen, A., and S. SengUpta
1983 Malnutrition of rural children and sex bias.
Economic and Political weekly 18.

Shathar, 2., N. Crook, C. Callum, and S. Kazi
1988 Women's status and fertility in Pakistan.
Population and Development Review l4(3):415-432.

 

Sheps, M.
1963 Effects on family size and sex ratio of
preferences regarding the sex of children.
Population Studies 17(1):66-72.

 

Shryock, H., and J. Siegel
1976 The method and materials of demopgraphy. Orlando,
Florida: Academic Press.

Sloane, D., and C. Lee
1983 Sex of previous children and intentions for
further births in the United States, 1965-1975.
Demography 20(3):353-367.

 

Speare, A., M. Speare, and H. Lin
1973 Urbanization, non-familial work, education and
fertility in Taiwan. Population Studies
27(2):323-334.

 

Stinner, W., and P. Mader
1975 Sons, daughters, or both? An analysis of family
sex composition preference in the Philippines.
Demography 12(1):67-79.

 

Stowasser, B.
1984 The status of women in early Islam. In: Muslim
Women, edited by F. Hussain. New Yerk: St.
Martin' 5 Press.

Stycos, J., H. Sayed, R. Avery, and S. Fridman
1988 Community development and family planning: an
Egyptian experiment. Boulder, Colorado: Westview
Press.

Stycos, J., and R. Weller
1967 Female working roles and fertility. Demography 4:
210-217.

 

Sufian, A.

1984 The effect of infant and child mortality on
fertility in Bangladesh. Unpublished
dissertation, Michigan State University, East
Lansing, Michigan.

114

Sufian, A., and N. Johnson
1989 Son preference and child replacement in
Bangladesh: a new look at the child survival

hypothesis. Journal of Biosocial Science
21:207-216.

Syed, S.
1978 Female studies and fertility in Pakistan. The
Pakistan Development Review l7(4):408-430.

 

Tabah, L.
1989 From one demographic transition to another.
Population Bulletin of the United Nations, No.
28.

Teachman, J., and P. Schollaert

1989 Gender of children and birth timing. Demography
26(3):411-423.

 

United Nations
1987 Women's employment and fertility. In: Fertility
Behavior in the Context of Development: Evidence
from the World Fertility Survey. Pp. 255-280. New
York: United Nations.

1989 Correlates of fertility in selected developing
countries. Population Bulletin of the United
Nations, No. 28.

ware, H.

1978 Population and development in Africa of the
Sahara. In: Population Policy: Research
Priorities in the DevelOping World, edited by C.
Miro and J. Potter. London: Frances Printers.

Weeks, J. .
1988 The demography of Islamic nations. Pepulation
Bulletin 43(4).

 

 

Williamson, N.
1976 Sons or daughters: a cross-cultural survey of
parental preferences. Beverly Hills, California:
Sage Publications.

1978 chs or girls? Parents' preference and sex
control. Pepulation Bulletin 33(1).

 

Y0ussef, N.
1978 The status and fertility patterns of Muslim
women. In: Women in the Muslim World, edited by
L. Beck and N. Keddie. Cambridge: Harvard
University Press.

115

Zurayk, H. -
1985 Women's economic participation. In: Population

1987

Factors in Development Planning in the Middle

East, edited by F. Shorter and H. Zurayk. New
York: The Population Council.

An overview of the development of studies on

fertility. Pepulation Bulletin of ESCWA #31,
United Nations.

"11111111111111.1111