final thesis complete

Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials

ESTIMATION OF FIRST CONCEPTION WAITS, FECUNDABILITY AND AGE AT FIRST CONCEPTION

AMONG WOMEN IN BANGLADESH

A ThesisSubmitted to the Department of Statistics, University of Chittagong, in Partial

Fulfillment of the Requirements for the Degree of Master of Science in Statistics

Submitted by Arif Ahmed

Examination Roll No. 2006/11Registration No. 12223

Session: 2005-2006Examination Year: 2006

Department of Statistics, University of Chittagong, Chittagong-4331, Bangladesh

January-2010

98


99

DedicatedTo

My BelovedParents

AndEldest Brother


To Whom It May Concern

This is to certify that Mr. Arif Ahmed bearing the Exam. Roll No. 2006/11,

Registration No. 12223, Session: 2005-2006 has prepared a thesis entitled

“Estimation of first Conception Waits, Fecundability and age at first Conception

among Women in Bangladesh” under my supervision in partial fulfillment for

M.S. Degree in Statistics of the University of Chittagong, Chittagong, Bangladesh.

(Dr. Md. Abdul Karim)

Professor

Depertment of Statistics

Faculty of Science

University of Chittagong

100


ACKNOWLEDGEMENT

At first I am remembering Almighty Allah for giving me strength, patience and ability to

accomplish this research work.

I wish to express my deepest sense of gratitude and respect to my reverend supervisor

Dr. Md. Abdul Karim, Professor, Department of Statistics, University of Chittagong for his

encouragement; constant supervision, valuable suggestions and instructions and kind help along

with giving me relevant books, some literature and national and international journals during the

entire research work. I have acquired invaluable knowledge through the proper guidance from

him. The insights gained from interaction with him have not only shaped this thesis, but also will

continue to be a valuable asset throughout my professional life. I am thankful to my honorable

teacher Dr. Md. Abdul Maleque, Chairman, Department of Statistics, University of Chittagong,

providing me all the facilities available for the completion of this research work.

I am extremely grateful to all the teachers of my department for their invaluable advice and

inspiration.

I especially concede my thanks to Prof. Dr. N.S.M.Yahya; Prof. Dr. S.M.Shafiqual Islam; Prof.

Dr. Manindra Kumar Roy; Prof. Prof. Dr. Rabindra Nath Shill; Dr. Jiban Chandra Paul; Prof. Dr.

Soma Chowdhury; Mr. Md. Shakhawat Hossain, Assistant Professor; Mr. Mohammad Salim

Zahangir, Assistant Professor; Mr. Md. Rokonuzzaman Azad, Assistant Professor and MS.

Zamilun Nahar, Lecturer, Department of Statistics, University of Chittagong for giving me

guidelines on different Sophisticated topics relavent to my thesis work. I am also thankful to all

of the staffs of my department. Especially, Parvin Akter, section officer (Computer) who helped

me in many ways.

I am grateful to all of my friends and elders who encouraged me to complete the research work

successfully.

Finally I would like to wish my profound gratitude to my family members and relatives for their

immense blessings, continuous advice and inspiration and immeasurable sacrifices that lead me to

all success in my life.

January, 2010

Department of Statistics The Author

101


University of Chittagong

Chittagong, Bangladesh.

CONTENTS

ACKNOWLEDGEMENT

ABSTRACT I

LIST OF TABLES III

LIST OF FIGURES V

CHAPTER ONE: INTRODUCTION 1-37

1.1 Historical backgrounds of Bangladesh

1

1.2 Geographical characteristics of Bangladesh

2

1.3 Socio-economic and demographic characteristics of Bangladesh

4

1.4 Objective of the present study 9

1.5 Applied computer software’s 10

1.6 Overview of fecundability 11

1.7 Previous research

19

CHAPTER TWO: DATA AND METHODOLOGY 38-57

2.1 Data 38

2.2 Survey objectives and implementing organizations 38

2.3 Sample design

39

2.4 Questionnares 41

2.5 Training and fieldwork 43

2.6 Data processing 44

102


2.7 Coverage of the sample 45

2.8 Methods of finding relevant data 47

2.9 Independent variables 47

2.10 Analytical approach 57

CHAPTER THREE: ESTIMATION OF CONCEPTION WAITS AND

FECUNDABILITY: LEVELS, TRENDS AND DIFFERENTIALS 58-114

3.1 Introduction 58

3.2 Model of homogeneous fecundability (geometric distribution) 62

3.3 Model of heterogeneous fecundability 63

3.4 Estimation of the parameters 68

3.5 Variance-covariance matrix of the moment estimators 71

3.6 Goodness of fit

74

3.7 Effect of memory and truncation biases on fecundability

78

3.8 Age at marriage, conception wait and fecundability 83

3.9 Inter dependence between age at first marriage and marital duration 88

3.10 Differentials of conception waits and fecundability

89

3.11 Trends in mean conception delay 103

3.12 Trends in fecundability level 109

CHAPTER FOUR: DIFFERENTIALS OF AGE AT FIRST MARRIAGE, AGE AT FIRST CONCEPTION AND AGE AT FIRST BIRTH 115-131

4.1 Differentials of age at first marriage 115

4.2 Differentials of age at first conception and age at first birth 121

CHAPTER FIVE: BIVARIATE ANALYSIS AND COX’S MULTIVARIATE

PROPORTIONAL HAZARD REGRESSION ANALYSIS 132-144

5.1 Bivariate analysis 132

103


5.2 Proportional Hazard Regression analysis 138

CHAPTER SIX: DETERMINANT OF FIRST CONCEPTION WAIT A PATH ANALYSIS 145-159

6.1 Introduction 145

6.2 Historical background 147

6.3 Analytical method 147

6.4 Results 153

CHAPTER SEVEN: SUMMARY AND CONCLUSION 160-169

7.1 Introduction 160

7.2 Summary of the major findings 161

7.3 Conclusion 166

7.4 Policy implications 168

7.5 Suggestions for the further research 169

APPENDIX-1 170

REFERENCES 171-176

104


Abstract

Reproductive behavior of women is generally influenced by age at first marriage,

which intern influences time required to conceive, age at first conception, level of

fecundibility and age at first motherhood because births are confined within

wedlock in Bangladesh. Fecundability is regarded as one of the important

proximate parameters of fertility performance of the married women. Due to the

complex nature of fecundability, we have attempted in this study to estimate mean

fecundability from the first conception interval, which is not associated with

postpartum infecundability. The first conception intervals have been estimated

indirectly by utilizing the data extracted from the 2007 Bangladesh Demographic

and Health Survey. Since the cohort of women is not homogenous in regards to

reproductive performance, we have attempted to estimate the mean recognizable

effective fecundability by fitting the Pearson Type-I beta geometric model with

parameters a and b to the observed distribution of first conception delay in

addition to geometric distribution. In our analysis, we have estimated the

parameters by the method of moments. The purpose of the present study includes

age at first marriage, estimating the mean conception delay, mean and

corresponding variance of fecundability and age at first conception, level trends

and differentials of fecundability and also to study the age at first motherhood of

the Bangladeshi women. The mean age at first marriage of the study women is

found only 16.42 ± 3.04 years. The mean conception delay of the Bangladeshi

women has been found 18.32 months after their first marriage and the mean

fecundability is 0.055, which is estimated by geometric distribution. The

theoretical arithmetic and harmonic mean fecundabilities are found 0.070 and

0.055 respectively by fitting Beta geometric distribution. The study has observed

the marked differentials of both conception wait and fecundability among

Bangladeshi women by to their different socio-economic, demographic and

105


cultural characteristics. The mean age at first conception and mean age at first

motherhood are found 17.94 and 18.69 years respectively. This study reveals that

the women with higher education have lower mean conception delay and higher

mean fecundability. We have also seen from this study that age at first marriage

has negative relation with conception wait and positive relation with fecundability,

age at first conception and age at first motherhood. It is observed that conception

wait is decreasing and level of fecundability is increasing with the increasing age

at first marriage whatever be the marital duration. Moreover, the fecundability

decreases with the increasing marital duration whatever be the ages at first

marriage. This indicates that the more the age at first marriage the higher the

fecundability level and less the conception wait and vice-versa. Furthermore, the

more the marital duration the less the fecundability and higher the conception wait

and vice-versa. The significant regression coefficient between age at first marriage

and age at first conception is found 1.135, which indicates that with the increase of

age at first marriage by one year, age at first conception tends to increase by 1.14

years. We also get the significant regression coefficient between age at first

marriage and conception wait as -1.252, which reflects that with the increase of

age at first marriage by one year, conception wait tends to decrease by 1.25

months. The trend analysis shows that conception wait is lower consequently

fecundability is higher in the recent past than at some distant point of time. The

multivariate analysis through the Cox’s proportional Hazard Regression model

shows that the respondent age at first marriage, spousal age difference, marital

duration, use of contraception, husband’s occupation, division, current age of

respondents and body mass index are found to have statistically significant

association with the marriage to first conception wait. Finally, path coefficients

reflect that along with direct effect, current age of respondents, marital duration

and use of contraception have indirect positive effect while age at first marriage,

106


husband’s education and spousal age difference have negative direct significant

effect on conception wait.

LIST OF TABLES

TABLE NO.

TITLE OF THE TABLES PAGE NO.

1.1 Population Estimates and Projections 062.1 Results of the household and individual interviews

(Number of households, number of interviews, and response rates, Bangladesh 2007

46

2.2 Independent variables, their categories, ranges, number of events (n) and percentage in each category.

55

3.1 Observed and estimated(a) number of months required to conceive

for Bangladesh.

76

3.2 Estimates and their standard errors by the method of moments 783.3 Estimates of conception delay and fecundability for women by

marital duration, Bangladesh.

80

3.4 Relationship among age at first marriage, conception wait and fecundability

85

3.5 Mean fecundabilities for women by marital duration and respondent’s age at first marriage cohorts.

89

3.6 Mean conception wait and fecundability by the fecundability differentials (by geometric distribution and Beta geometric distribution by the method of moments

98-102

3.7 Mean conception delay for women by year preceding the survey 104

3.8 Mean Conception wait by year preceding the survey and

respondent’s age at first marriage.

104

3.9 Mean conception wait by marital duration and respondent’s age at

first marriage

105

3.10 Mean conception delay and fecundability by year preceding the survey

110

3.11 Mean fecundability by year preceding the survey and respondent’s

age at first marriage

111

107


3.12 Mean fecundabilities by marital duration and respondent’s age at

first marriage

111

4.1 Differentials of mean and median age at first marriage by background characteristics, Bangladesh, 2007

118

4.2 Mean age at first conception and first birth by background characteristics, Bangladesh, 2007

125-127

4.3 Relationship between ages at first marriage and mean age at first conception

130

5.1 Cross tabulation of conception interval by different backgroundcharacteristics of the respondents.

136-137

5.2 Cox’s Proportional Hazard Regression of conception waits by differentcharacteristics of respondents

143-144

6.1 Variables and their Measurements used in the Path Analysis 1506.2 Path coefficients for specified combinations of variables of Marriage to

first conception wait155

6.3 Analysis of the Effects of the variables used in the Path model for explaining Conception wait for women in Bangladesh

158

6.4 Zero-order Correlation Coefficients among the selected socio-economic, cultural and Demographic Variables

159

108


LIST OF FIGURES

FIGURE NO.

FIGURE CAPTIONS PAGE NO.

3.1 Observed and Estimated months required to conceive

(method of moments) for females, Bangladesh, 2007.

77

3.2 Mean conception wait by marital duration 813.3 Mean fecundability by marital duration 82

3.4 Mean conception wait by age at first marriage 863.5 Mean fecundability by age at first marriage 873.6 Conception wait by year preceding the survey 1063.7 Conception wait by years preceding the survey for different

age at first marriage107

3.8 Fecundability by age at first marriage for different marital duration

108

3.9 Fecundability by years preceding the survey 1123.10 Fecundability by years preceding the survey for different age

at first marriage113

3.11 Fecundability by age at first marriage for different marital duration

114

4.1 Mean age at first conception at different ages at first marriage.

131

6.1 A causal Model for Factors affecting timing of first conception

151

6.2 Path diagram of marriage to first conception wait and predetermined variables

154

109


110

CHAPTERONE


CHAPTER ONE

INTRODUCTION

1.1 HISTORICAL BACKGROUND OF BANGLADESH

The history of Bangladesh, in full People’s Republic of Bangladesh, republic of

southern Asia, is an eventful combination of turmoil and peace, as well as

prosperity and destitution. It has thrived under the glow of cultural splendor and

suffered under the ravages of war. The territory now constituting Bangladesh was

under the Muslim rule for over five and a half centuries from 1201 to 1757 A.D.

Then, it was ruled by British India, after the defeat of the last sovereign ruler,

Nawab Sirajuddowla, at the Battle of Palashi on the fateful day of June 23, 1757.

The British ruled over the entire Indian sub-continent including this territory for

nearly 190 years from 1757 to until Britain withdrew in 1947. During that period

Bangladesh was a part of the British Indian provinces of Bengal and Assam. With

the termination of the British rule in August, 1947 the sub-continent was

partitioned into two independent states, India and Pakistan. Bangladesh was then a

part of Pakistan and was known as East Pakistan. Pakistan consisted of the Muslim

areas of the northwest and northeast parts of the Indian subcontinent, known

respectively as West Pakistan and East Pakistan. The 1,500 miles of foreign

territory separating the two wings, the difference in climate, topography, language,

111


socio-cultural characteristics, and above all sharply an unequal economic

development and political representation resulted in the de facto existence of two

separate nations within Pakistan (Brass et al., 1981). A political debacle escalated

into a civil war in March, 1971 which culminated in an all-out war of liberation

against West Pakistan. After a tremendous sacrifice of life and property in the

nine months war Bangladesh emerged as sovereign and independent state since the

declaration of freedom on March 26, 19971.

1.2 GEOGRAPHICAL CHARACTERISTICS OF BANGLADESH

Bangladesh is a low-lying, riverine country located in South Asia with a largely

marshy jungle coastline of 710 km (441 mi) on the northern littoral of the Bay of

Bengal.Bangladesh is one of the largest delta lands in the world covering the area

of 56,977 square miles or 147,570 square kilometers of which 8,236 square

kilometers is riverine and 1,971 square kilometers is under forest. It lies in the

north eastern part of South Asia between 23034/ and 26038/ north latitudes and

88001/ and 02041/ east longitudes. The country is bounded by India to the east,

north, and west (about 1500 miles long boundary) and shares a short frontier with

Myanmar ( about 120 miles) on the south east. To the south lies the Bay of

Bengal, and to the northeast lies the broad mass of the Assam range on the

Shillong plateau. The Himalayas lie not far from the boundary on the northwest.

Except for some mountainous regions in the east and southeast part more than 85

percent of the total area of Bangladesh is flat alluvial pain crisscrossed by the

mighty rivers Padma, Meghna, and Jamuna and by their innumerable tributaries.

These rivers play a significant role in the lives of the people and in the physical

environment of Bangladesh and these rivers system traverse the country to reach

the Bay of Bengal (Johnson, 1975). Bangladesh's alluvial soil is highly fertile, but

vulnerable to flood and drought. Hills rise above the plain only in the Chittagong

Hill Tracts in the far southeast and the Sylhet division in the northeast. Straddling

the Tropic of Cancer. . Natural disasters, such as floods, tornadoes, and tidal bores

112

http://en.wikipedia.org/wiki/Tidal_bore

http://en.wikipedia.org/wiki/Tornadoes

http://en.wikipedia.org/wiki/Tropic_of_Cancer

http://en.wikipedia.org/wiki/Chittagong_Hill_Tracts

http://en.wikipedia.org/wiki/Chittagong_Hill_Tracts

http://en.wikipedia.org/wiki/Drought

http://en.wikipedia.org/wiki/Flood

http://en.wikipedia.org/wiki/Alluvial_soil

http://en.wikipedia.org/wiki/Bay_of_Bengal

http://en.wikipedia.org/wiki/Bay_of_Bengal

http://en.wikipedia.org/wiki/Littoral

http://en.wikipedia.org/wiki/South_Asia

http://en.wikipedia.org/wiki/Riverine


affect the country yearly. Bangladesh also is affected by major cyclones — on

average 16 times a decade. One cyclone struck the southeastern coast in May

1991, killing 136,000 people. The cyclone Sidr struck the southwestern coast on

November 15, 2007 affecting not only the coastal districts of the administrative

division Khulna but also about half of the tropical forest Sundarbans.

Bangladesh is a predominantly a rural country with an economy that is dominated by

agriculture and has a sub-tropical monsoon climate. While there are six seasons in a year,

three namely, Winter, Summer and Monsoon are prominent. Winter which is quite

pleasant begins in November and ends in February. In winter there is not usually

much fluctuation in temperature which ranges from minimum of 7 o C—13 o C

(45 o F—55 o F) to maximum of 24 o C—31 o C (75 o F—85 o F). The maximum

temperature recorded in summer months is 37 o C (98 o F) although in some

places this occasionally rises up to 41 o C (105 o F) or more. Monsoon starts in

July and stays up to October. This period accounts for 80% of the total rainfall.

The average annual rainfall varies from 1429 to 4338 millimetre. The maximum

rainfall is recorded in the coastal areas of Chittagong and northern part of Sylhet

district, while the minimum is observed in the western and northern parts of the

country (source: BANGLADESH BUREAU OF STATISTICS, MARCH 2009).

Mean annual temperature lies between 570F and 800F. The annual rainfall varies

from 50 inches in the west to 100 inches in the southeast and to 200 inches in the

Assam hills in the north. These factors make the climate very unpleasant with high

relative humidity (about 90 percent) accompanied quite high temperature

(Johnson, 1975). Sometimes tropical cyclones often accompanied by tidal bores

from the Bay of Bengal and floods have, on numerous occasions, brought about

havoc in the country resulting in considerable loss of millions of people and

properties (Nyrop et. al., 1975)

For administrative purposes the country is divided into seven divisions (Barisal,

Chittagong, Dhaka, Khulna, Rajshahi,Rangpur & Shylhet) and 64 districts. Each

113

http://en.wikipedia.org/wiki/Sundarbans

http://en.wikipedia.org/wiki/Sidr

http://en.wikipedia.org/wiki/Cyclones


district is divided into thanas, classified as rural, urban and semi-urban areas. The

2001 census listed 507 thanas, which intern was subdivided into 4484 unions. A

union is composing of a group of villages; the 2001 census, reported the existence

of 87319 villages in Bangladesh. In Bangladesh 77 percent population is living in

the rural areas while that urban 23 percent. According to 2001 census, the

percentage of Muslim population is 89.7 while that of Hindu, Buddhist and others

are 9.2, 0.7 and 0.4 respectively. The national language of Bangladesh is Bangla,

and about 98. 2 percent of the population has Bengali as their mother tongue

(BBS, 2001).

1.3 SOCIO-ECONOMIC AND DEMOGRAPHIC CHARACTERISTICS OF

BANGLADESH

Bangladesh is predominantly a rural country with about 77 percent population

living in the rural areas (BBS, 2003) with the exception of some tribal population

in the hilly regions. The economy of Bangladesh is overwhelmingly dominated by

Agriculture. Unemployment/ Underemployment is serious problem and pressure

on the land in rural areas has lead to movement of the people from rural to urban

areas. Currently unemployment including underemployment rate is 40 percent in

Bangladesh. Bangladesh has recorded impressive economic and social gain in

1990s. Though the adoption of and implementation of sound policies and

strategies Bangladesh has manage to sustain a large measure of economic stability

and microeconomic growth.

The per capita income has grown steadily from $273 in 1990-91

to $470 (World development indicator database, 2005) in

2005.The country’s annual GDP growth reached a robust 5.9% in

Fiscal Year 2009 (FY09), a slight drop from the 6.2% growth achieved in

FY08 (2009 Bangladesh Economic Update) while 4.5 percent in

2000. Bangladesh has made remarkable achievement in reducing

poverty between 1991 and 2000. Poverty rate fell by 9 percent

114


during the decade. However, some 63 million (Roughly hall of the

population) people still live in severe deprivation, two-thirds of

them in extreme poverty. Bangladesh is still struggling to emerge

from the realm of poverty. The HPI-1 value of 36.1% for

Bangladesh, ranks 112th among 135 countries for which the index

has been calculated( Human Development Report 2009,

Bangladesh) Over the last 10 years Bangladesh has made

impressive gain in key human development indicators The HDI for

Bangladesh is 0.543, which gives the country a rank of 146 th out

of 182 countries (Human Development Report 2009,

Bangladesh). The UNDP Gender Development Index (GDI) for

2004 ranked Bangladesh 110 among 144 countries, an increasing

of 13 positions since 1999. This improvement reflects a closing of

the gap between men and women in key indication such as live

expectancy. Bangladesh ranks 108th out of 109 countries in the

GEM, with a value of 0.264 (Human Development Report 2009,

Bangladesh).

The literacy rate in Bangladesh is very low. It is remarkably lower

among females than their male counterparts. Only 40.3 percent

of males and 33.4 percent of females were able to read and write

in any language (BBS, 2003). The literacy rate of the population

aged five years and over is 42.5 percent; of them 46.4 percent of

males and 38.3 percent for females. The rural-urban and religious

differential in literacy rate is substantial. Among urban population

the literacy rate for both sexes was 52.3 percent as against 32.7

for their rural counterparts. While, the literacy rate among

Muslims was 44.62 percent as against 52.4 for Hindus (BBS,

115


2003). Although literacy continues to remain low quite low in

Bangladesh it has shows some improvement over the years. It

increased gross primary enrolment from 72 percent in 19990 to

98 percent in 2001 and has already attain Millennium

Development Goals (MDG) of eliminating gender disparity in

primary and secondary enrolment (Bangladesh country overview,

2006). The school attendance rate of the population aged 5 to 24

years for males and females are 55.8 and 46.3 percent

respectively, which indicate the considerable narrowing of gaps

by gender.

Bangladesh is the 7th most populous country in the world (PRB, 2006). According

to the United States Census Bureau (BUCEN-IDB-2009), the population was

156.05 million as against 130 million for 2001 census. The increase in adjusted

population stands 26.05 million over the Eight years. Density of population is one

of the highest in the world with an average of 977 persons per square kilometer

(Economic survey, 2009). Bangladesh has made significant strides in lowering its

population growth rate. The intercensal population growth rate peaked in the early

1970s at about 2.5 percent per annum, followed by a decline to 2.2 percent during

1081-1991 percent and in 2001 the rate became 1.53 (BBS,2003), according to

BUCEN-IDB-2009 the growth rate became 1.3 percent. But according to

Economic Survey the population of Bangladesh in 2009 is 142.4 million with an

estimated growth rate of 1.26 percent per annum; its population may be double in

about 30 years. A statistical report on Population Estimates and Projections

(Source: BUCEN-IDB) is mentioned below:

TABLE 1.1 Population Estimates and Projections

Year Population

1960 54,592,652

116


1970 67,331,349

1980 87,937,333

1990 111,436,651

2000 136,681,493

2010 158,065,841

2020 180,753,264

2030 204,141,664

2040 221,691,969

2050 233,587,279

Source: BUCEN-IDB

The population of Bangladesh is characterized by a young age

structure. The relatively young age structure of the population

indicates continued rapid population growth in the future.

According to 2001 census, 43 percent population is under 15

years of age , 52 percent between 15 and 64 years and 3 percent

are age 65 and above ,in the year 2003 the percentage become

32.9, 63.6 and 3.5 respectively (2006 CIA , World Fact Book). But

in the year 2009 the total population of Bangladesh is

156,050,883 (July 2009 est. Source: BUCEN-IDB) and It’s age

structure are 0-14 years: 34.6% (male 27,065,625/female

26,913,961), 15-64 years: 61.4% (male 45,222,182/female

117


50,537,052) & 65 years and over: 4% (male 3,057,255/female

3,254,808) (2009 est.). The young age structure constitutes a

build-in “population momentum” which will continue to generate

population increase well in to the future, even in the face of rapid

fertility decline.

The contraceptive prevalence rate for married women in Bangladesh has

increased from 8 percent in 1975 to 56 percent in 2007, a sevenfold increase over

more than three decades (Table 5.6 and Figure5.1). In Bangladesh the CPR rose

from 7.7 percent in 1975 to 31 percent in 1989 and farther to 39.9 percent (REF

10) in 1991. Contraceptive prevalence rate again increased from 43 percent in

1993-94 to 58 percent in 2004 (BDHS, 2004). Overall, current contraceptive use

has declined by two percentage points in the past three years, from 58 percent in

2004 to 56 percent in 2007, but use of modern methods has remained unchanged.

Use of oral pills has continued to rise, but a two-decade trend of increasing

injectable use was interrupted in 2007 (Table 5.6 and Figure 5.2). The observed

decline in injectable use, from 10 percent in 2004 to 7 percent in 2007, could be

the result of a significant shortage in injectable supplies during some periods of

2006-2007 that affected public sector and NGO family planning service delivery.

Between 2004 and 2007, use of traditional methods also declined from 11 percent

to 8 percent.

Among the remarkable demographic transition the Crude Death Rate (CDR) has

fallen dramatically, from about 19 per thousand population in 1975 to about 8

deaths/1,000 population ("CIA World Factbook 2007"2009 est). Although infant

and under five mortality rates are declining, they are still high. The infant mortality

was 145 deaths per thousand live births in 1975 and fell to 47 in 2007 (UNICEF

Global Database/Child Health Malaria-2009) and under-five mortality was 239 in

1970 and reduces to 61 per 1000 in 2007(UNICEF Global Database/Child Health

Malaria-2009 ).

118

https://www.cia.gov/library/publications/the-world-factbook/geos/bg.html#People


Maternal mortality has declined from 620 deaths per 100,000 in 1982 to 380 in

2005, which is still high as compared to other developed countries (Women of

the World, 2005). Poor nutrition, poverty and lack of access to health services

contribute to some 20,000 maternal deaths each year. This small but important

decline of mortality is mainly attributed to increased availability of family planning

and immunization services, improved antenatal and delivery care and a reduction

in the number of births to high risk mothers(BBS, 1997A:144). Because of the

mortality decline there is evidence of modest improvement in life expectancy

during the past decade. Life expectancy at birth was 46 years for males and 47

years for females in 1974(UN, 1981:60) but in 2006 the life expectancy at births

was estimated as 62.47 and 62.45 for males and females respectively. Life

expectancy at birth in Bangladesh is 65.7(Bangladesh’s human development

index 2007. The human development index trends tell an important story in that

respect. Between 1980 and 2007 Bangladesh's HDI rose by 1.86% annually from

0.328 to 0.543 today. , which gives the country a rank of 146th out of 182

countries.

The couples of Bangladesh have accepted small family norm. About 70 percent of

ever married women prefer a two-child family to be ideal and 16 percent consider

a three-child family ideal, Overall, the mean ideal family size among currently

married women is 2.3 children, a decline from 2.9 in 1989(BDHS). Fertility in

Bangladesh is still high even by the standards of developing countries. The crude

birth rate (CBR), the number of births per 1000 population which gives a rough

estimate of the fertility level is a measure of period fertility. CBR in Bangladesh in

2009 was 24.7 per 1000 (BUCEN-IDB-2009) which indicate the modest decline in

fertility as the same was 43.8, 40.6 and 29.8 in 1981, 1987 and 2006 respectively.

The total fertility rate has declined from nearly seven births per women in 1975 to

about five births in 1989 (BFS). Further fertility has declined from five births per

women in 1989 to 4.3 in 1991 (CPS). TFR dropped almost imperceptibility from

4.3 to 3.4 in 1993-94 (BDHS). Further fertility has declined from3.4 to 3.3 in

119


1996-97(BDHS) and then remain stagnant till 1999-2000. But after almost a

decade of stagnation the TFR Has declined slightly to 3.0 children per woman.

Further fertility has declined from 3.0 to 2.7 in 2007(BDHS). Differentials in

fertility by background characteristics are substantial. Women in rural areas have

more children then their urban counterparts (2.8 and 2.4 children per woman

respectively) The TFR is highest in Sylhet division 3.7 and lowest in Khulna 2.0,

in Rajshahi 2.4, in Barisal and Dhaka 2.8 and in Chittagong 3.2. Fertility

differentials by women’s educational status are notable; women who have no

formal education have an average of 3.0 children while women with at least some

secondary education have 2.3 children.( Source: BDHS 2007)

Crude activity rate in Bangladesh is very low. The percentage of the total

population economically active has declined from 28.7 in 1974 to 27.5 in 1980

(BBS,2003). About 60 percent of the economically active population 10 years and

over were engaged in agriculture, and only 9.3 percent of the females aged 10

years and over were economically active in 2001 and it was only 2.5 in 1974

(BBS, 2003). The reductions in crude activity rate over the decades indicate

partially the higher attendance of the boys and girls in school and colleges. The

increase female activity rate shows a positive indicator for participation of the

women in economic activity as well as the development level of the country.

Bangladesh has a high dependency ratio of about 83 persons which is mainly due

to an exceptionally high proportion of young persons in its age structure

(BBS,2003).

1.4 OBJECTIVE OF THE PRESENT STUDY

The main objective of this study is to analyze the important components of

respondents of resproductive behavior at the beginning of reproductive span. This

includes the age at first marriage, conception waits, fecundability, and age at first

conception and age at first motherhood among women in Bangladesh. We have

seen that a very few attempts have been made to estimate the mean fecundability

120


for Bangladeshi women as it is not a directly observable event. For this, we know

very little about such an important fertility parameter of Bangladeshi women.

However, the specific objectives of the study are:

I) To analyze the pattern of waiting time from marriage to first conception among

women in Bangladesh.

II) To observe the trend and differentials of conception wait and also to observe

the pattern of fecundability.

III) To estimate level of fecundability by fitting geometric and beta geometric

distribution.

IV) To observe the mean age at first marriage among Bangladeshi women and

its differentials by available background characteristics.

V) To estimate the mean age at first conception from age at first marriage and

conception waits and it’s differentials.

VI) To find the relationship between age at first marriage and conception wait

and also to age at first marriage and age at first conception and finally.

VII) To find the mean age at first motherhood and it’s differentials with socio-

economic, demographic and cultural characteristics.

1.5 APPLIED COMPUTER SOFTWARE’S

The computer software’s which are used in this research work are:

I) Statistical Package for Social Science (SPSS for Windows version 11.5) has . been used to analyze the data. II) Application Package Microsoft Excel has been used to draw various graphs.

121


III) Programming language FORTRAN has been used to obtain the expected frequencies of conception waits in method of moments.

1.6 OVERVIEW OF FECUNDABILITY

One of the important but difficult areas of investigation in fertility is that of

fecundability, or the probability of conception. Along with many other variables

fecundability determines the observed reproductive behavior of women. In

developed countries with a high use of contraceptives, fecundability may be of less

122


relevance to the study of fertility. However, in non-contraceptive societies or the

societies where contraceptive use is quite low, fecundability could be one of the

prime determinants of fertility. Fecundability effects fertility through its

relationship with the average time required for a conception to occur, and can also

be thought of as the transition probability for the passage from susceptible state to

pregnancy (Perrin and Sheps, 1964).

Defining fecundability as the “probability of conceiving during a month in the

absence of conscious efforts to limit procreation” leads to the interesting question

of why the sizes of families often vary widely in non-contraceptive societies.

Hutterites have a mean completed family size of eleven children, while six

children even fewer, is the mean in some Asian communities, especially in India

and Bangladesh. Nuptiality patterns by themselves can not explain such large

differences. The variation in natural fertility could be due to differentials in

fecundability.

The Italian demographer Gini (1924) first introduced the concept of fecundability

into the vocabulary of demography. The concept of fecundability was popularized

only thirty years later by Henry and others (1987). According to Gini,

fecundability is the probability for a married woman to conceive during a month,

in the absence of any Multhusian or neo- Multhusian practice intended to limit

procreation. By “Multhusion Practice” Gini probably meant complete voluntary

abstinence from sexual intercourse while “neo- Multhusion Pracitce” refers to

contraception. Gini gives the unit of time as “a month” without clarifying further;

he probably meant a calendar month, not a monthly menstrual cycle. The

distinction between the two is often not worth making, since the precision that

could be gained would be small compared with the additional complication in the

calculations. Such a gain would even be illusory in most cases; as we have noted

123


the most frequent length for a cycle is 28 days, but the mean length is generally

about 30-31 days, the length of a calendar month. The general concept of

fecundability, or the monthly probability of conception, grew out of Gini’s (1924)

early work on birth intervals and time-to-conception, while Henry (1972)

expanded upon the concept and its measurement. Most demographers involved in

studies of conception time during the past two decades have focused on

mathematical models of birth interval dynamics, or on techniques for estimating

population fecundability (Sheps and Menken 1973; Bongaarts 1975). Some

research has been done on differential fecundability and its determinants within

populations, but few studies have been multivariate. The main focus has usually

been on the relationship of fecundability to age or age at marriage (see Leridon

1977). But in an early multivariate study based on Taiwan data, Jain (1969) found

that fecundability varies directly with wife’s socio-economic status net of age at

marriage and marriage duration. Bogue and Bogue (1980) used multiple regression

analysis to examine some predictors of conception time for the first live birth for

women of proven fertility (two or more births) using CELADE data from seven

Latin American cities. Pregnancy loss was found to have a relatively large and

significant inverse effect on fecundability, while age at marriage generally showed

a positive and significant effect on fecundability. Education of mother also was

positively related, but less consistently and with a weaker effect.

The United Nations (1958) defined fecundability as the probability of a conception

in a menstrual cycle, a conception being the fertilization of an ovum by a sperm.

Fecundability is measured in women, who ovulate regularly, i.e., Pregnant, Sterile

or Postpartum anovulatory women are excluded. We now need to be a little more

precise in the definition of fecundability because not all conceptions are

recognized. To be noticed by the women, a pregnancy must at least delay the first

menses after conception as modern tests of pregnancy are not efficient before this

124


limit either. Even after these two weeks, many women will not be aware of a

pregnancy if menses come back within a few weeks. Since there is a little

difference between the average duration of a cycle and a calendar month

(Matsmoto et al., 1962 ; Voltman, 1956) ; it is practically convenient to use a

calendar month without losing much precision rather then using a menstrual cycle

while measuring the time unit for the measurement fecundability and in this

investigation are also refer to the same. In practice, fecundability is measured in

women who are ovulating regularly that is, Pregnant, Sterile or Postpartum

anovulating women are excluded.

Fecundability, as defined by the Italian demographer Gini, is the monthly

probability of conception in the absence of contraception, outside the gestation

period and the temporary sterile period following the termination of pregnancy.

An obvious definition of fecundability is required for the convenience of our

forthcoming analysis of the data. Since true fecundability is virtually impossible

to measure accurately, simply because many conceptions are unrecognizable.

Hertig (1967) estimated that nearly half of all the conceptions are never identified,

either because the fertilized ovum aborts shortly after implantation. Women often

are seen not aware of their conception as menstruation regulation stopped when it

come back after two or three weeks, they do not think that their conceptions were

aborted spontaneously. For a group of women the un-weighted average of

fecundabilities for individual woman is denoted by mean fecundability. It is

obvious that the forgoing definitions may have different implications and various

interpretations depending upon the outcome of the conceptions. To be more

precise, in the definition of fecundability, we make the following distinctions.

1. Total (or physiological) Fecundability (TF): Total fecundability is the

probability that any conception regardless of outcome occurs during a cycle

125


including non-implemented ova and conception aborted spontaneously before the

end the cycle ( Bongaarts, 1975, Leridon, 1977)

2. Recognizable Fecundability (RF): It is the probability of a conception, which is

recognizable at the end of the conception cycle by the non-occurrence of the

menstruation; excluding pregnancies ending within two weeks after conception

(first missed menses). A fraction 1 of all conception fails to implement or aborts

before the beginning of the next cycle. Thus

RF = TF (Bongaarts, 1975)

3. Effective Fecundability (EF) : It is the fecundability of a conception which will

end in a live birth. A fraction 2 of the recognizable conceptions aborts

spontaneously after the first cycle of gestation; therefore

EF = (1-α2) RF

= TF.

The parameter 2 will be referred to as the incidence of late spontaneous abortion

and includes the risk of still birth.

4. Apparent Fecundability (AF): It is the fecundability when including all

pregnancies recognized and declared by the women at the time of interview. In

addition to above definitions, further distinction can be made between gross and

net fecundability measures:

a. Gross Fecundability (GF): The Gross fecundability is used for the probability

of conception in females actually engaging in regular intercourse.

126


b. Net Fecundability (NF): Net fecundability measures the conception probability

in a population living in a normal social and biological environment in which

temporary separations and illness occur.

The following diagram illustrates that total (or physiological) fecundability is

related with all conceptions; recognizable fecundability is related with pregnancy

ending in period B, C, and D; apparent fecundability is related with pregnancies

ending in period C and D; effective fecundability is related with live births only

(period D)

SOME CONCEPTS RELATED TO FECUNDABILITY

Fecundability, treated as a biological characteristic of the females, is clearly affected by

several factors, both physiological, generally the natural side of fecundability and

behavioral factors, that are within the jurisdiction of the control fertility. Some of

127

Fertilization (conception)

Last First 2d month 3d month 6th month Full term

Menstrual Missing (approx.)

Period menses

A B C D


these factors which are of frequent occurrence in any fecundability study are

discussed below:

(i) The reproductive period:

It is difficult to clearly define the bounds of the woman’s reproductive period.

There are of course, two objective limits –menarche ( or puberty) and menopause,

the cessation of menstrual period.

Menarche: Menarche (the on set of menstruation) is thought to signal the time

when a female becomes capable of reproduction. Menarche usually occur

somewhere between the age of 11 and 15 with considerable variation in the mean

age among different population.(Sheps et ,al, 1973 : Leridon 1977,) The menses

are often rather irregular at first, and anovular cycles (without the release of ova

from ovary)may be frequent. The release of behavioral and ova does nor

necessarily indicates that these ova can be fertilizes or implemented in the uterus

and brought to term. Hence full sexual maturity probably occurs some time after

menarche. However, it is difficult to study the socio-cultural factors influencing

age at menarche, since it will be chiefly determined by the biological and

nutritional factors ( Shella, J., 2001)

Menopause: Menopause is the end of a woman’s reproductive life. Generally it

occurs between the age of about 42 and 50 ( Sheps and Menken 1973, Wyon ND

Gordon, 1971). A woman is postmenopausal if she has experienced at least 12

months since the last menses in the absence of a known pregnancy (Wood, 1995).

After the menopause a woman is sterile.

The interval between puberty and menopause is termed “Reproductive period of a

female. The total length of reproductive life could be longer than 35 years if we

consider the complete interval between puberty and menopause : actually,

however, the fecund period is on the average only 27-38 years ( Leridon, H. 1977)

128


Sterility:Sterility is a couples incapacity to produce a live-born child, A sterile

couple can not procreate a child. The sterility may be due to either or both

partners.. If conception is impossible physiologically to a female she is said to be

sterile or infecund. If the fecundability is zero throughout the life of a female, she

is said to be primarily sterile: if it becomes zero at any time during her

reproductive period ( which arises after one or more children have been born ) ,

she is known as secondary sterile (Heter , 1970: Weir and Weir, 1961).

Fetal wastage: A conception may not always result in a live birth, the outcome of

the corresponding pregnancy may end ina spontaneous fetal death, in an induced

abortion, in a still birth or in a live birth. For convenience, if a conception results

in a live birth, it is termed as complete and if the conception result in a fetal

wastage, it is termed as incomplete. It seems reasonable to consider that in a given

context each of these outcomes have a numerical probability which may vary with

maternal age and health, rank order of the pregnancy and the time elapsed since

the previous pregnancy(Abramson, 1973: Potter at. Al., 1965).

Non-susceptible Period: The fecundability of a female is temporarily suspended

following each conception when menstruation discontinues for some times. This

period of non-susceptibility is the sum of two parts: first, gestation period and

second, the interval after its terminative and before the resumption of ovulation

which is known as post partum amenorrhea (PPA) period. The outcomes of a

pregnancy depends upon the duration of pregnancy and PPA, while the later

depends also upon the beast feeding practices and their physiological impact.

Duration of fertile period: The exact duration of fertile period has proven

difficult to determine, and, as a consequence, the existing estimates vary widely

from less than a day (Tietze, 1960) to a week or more (Hartman, 1962). The fertile

period is a brief interval around the time of ovulation during which an

129


insemination can result in a fertilization. Fertilization is possible if insemination

occurs before ovulation because sperm retains its fertility for a short time after

insemination. Fertilization can also take place following ovulation because the

ovum remains viable for a brief interval. An approximate estimate of fertile period

therefore would be the sum of the fertile life times of sperm and ovum, but this

estimate is slightly too high because it does not take into account the time required

for the sperm capacitation. Sperm needs 6 hours after insemination before

becoming fertile (James, 1979) and this time should be subtracted to yield an

accurate of the fertile period.

Best available estimates of the fertile life times range from 24 to 48 hours for

sperm (Robert G and Potter Jr. 1961 ;Vander Vliet and Hafez 1974) and from 12

to 24 hours for the ovum (Barret and Marshall 1969). Summing these life times

and subtracting 6 hour for capacitation gives a range of 30 to 66 hours for the

fertile period with an average of about 48 hour or two days.

Fertility, Fecundity and Fecundability:

130

Fertile period

Fertile life time sperm Fertile life time of ovum

Period of capacitation

Earliest possible fertile insemination

Exact time of ovulation

Latest possible fertile insemination


It is useful to briefly discuss the basic terms, fertility, fecundity and fecundability.

Fertility refers to actual reproduction, whereas fecundity denotes the ability to

reproduce. A woman who is bearing children is fertile; a women is considered

fecund if she is capable of bearing live off spring. The opposite terms are

infertility, also called childlessness and infecundity, which is synonymous with

sterility. Sterility (infecundity) implies the existence of infertility but the reverse is

not necessarily the case. A fecund woman may choose to remain in fertile by not

marrying or by practicing highly effective contraceptive. Infertility then is either

due to a voluntary decision not to have children or it is caused by (biological)

infecundity.

The term fecundability refers to the ability to conceive. Fecund and fertile women

are necessarily fecundable, although the may experience temporary periods of

infecundabilit. However, some fecundable women are infecund, and consequently

infertile, because they are physiologically unable to successfully complete a

pregnancy. The term fecundability has taken on a specific meaning as the

probability of conceiving per month (among cohabiting women who are not

pregnant, sterile or temporary infecundable)

These definitions are given in the Multilingual Demographic Dictionary (United

Nations, 1959) and are constantly used in the English demographic literature. It

should be noted that in France and in other Roman Language the terms fertility

and fecundability are reversed; that is, fecundity is the equivalent of fertility and

fertiles equal to fecundity. To ad further to the confusion, the words fertility and

fecundity are used virtually synonymously in the biological and medical literature.

131


More specific meaning can be given in the terms fertility and fecundity by adding

objectives. For example, natural fertility is found in populations where no

contraception or induced abortions is practiced ( no deliberate birth control is

practiced; controlled or regulated fertility is observed in societies where fertility

control practices are widespread. Similarly, natural fecundability is the monthly

probability of conception in the absence of contraception, and residual or

controlled fecundability refers to the conception risk in the presence of

contraception.

In practice recognizable and effective fecundability are two most widely used

measure of fecundability. In this study, by the term fecundability we will mainly

refer to net recognizable and effective fecundability. For convenience, the term

‘fecundability’ will be reserved to refer to it. The time unit for the measurement of

fecundability is the duration of one menstrual cycle. However, many investigators

use the more convenient calendar month, a practice that is accepted in the present

model. In any case, the difference between the average duration of a cycle and a

month is small (Matsumoto, 1962; Vollmen 1956).

1.7 PREVIOUS RESEARCH

A considerable amount of effort has been expanded in attempts to estimate

fecundability since Gini (1964) first define it. In 1975 Bongaarts derived a new

method for the estimation of the mean and variance of fecundability from the

distribution of interval from marriage to first birth or from the resumption of the

conception risk after contraception to the subsequent birth. The estimates of the

mean and variance of fecundability are obtained by fitting a model for the

distribution of intervals from marriage to first birth.

Although the theoretical importance of fecundabiliy is beyond question as a major

proximate determinants of natural fertility and as a major standard by which it

132


assess the impact of fertility regulation but the enormous variation of its estimates

undoubtedly attributable to difference in methodology and definitions. Bongaarts

(1975) pointed out that intervals of fecundability vary greatly from study to study

because of deferent methods and different definitions.

Yet even when the same method and the same definition were applied in a study of

several historical populations, fecundability ranged from 0.18 to 0.31 (Wilson,

1987). Larsan and Voupel, (1993) provided the estimates that are based on the

models that incorporate the effects of persistent heterogeneity and that are use the

full information provided by multiple spell duration data. Empirical studies based

on birth histories are few. Estimates in the studies vary from 0.144 to and 0.189 in

United States (Potter and Parker, 1964; Westoff et al., 1961) and from 0.163 in

Taiwan (Jain, 1969) to 0.318 in eighteen centaury European population (Henry,

1964). Differences in the characteristics of the sample population make the

validity of comparison tenuous.

For a group of women the un-weighted average of fecundabilities for individual

women is defined by mean fecundability. In the past, at least five different

methods for the estimation of mean value of fecundability in a population have

been explored.

(a) Calculating fecundability from coital frequency and the duration of the

viability of sperm and ovum. (Glass and Grenbanik, 1954; Lachenbruch 1967;

Westoff at al., 1961; Tietzre 1960)

(b) Observing the proportion of women conceiving during a one-month period

(cycle) of exposure to the risk of conception (Barrett, 1969, 1971 : Gini, 1924;

Henry, 1953; Porter 1961; Sheps 1965; Titze et al., 1950; Whelpton and Kiser,

1950)

(c) Fitting models to the distribution of waiting times to conception (Henry.

1964(a); Jain, 1969; Mojumder and Sheps, 1970; Potter and Parker, 1964; Sheps

133


1964; Sheps and menken, 1972; 1973; Mode, 1985). This method can also be yield

estimates of the variance of the fecundability (Mojumder and Sheps, 1970; Henry.

1964(a): Jain, 1969; Sheps and menken, 1973; Leridon, 1973).

(d) Fitting Models to birth interval distributions ( D’Sousa, 1973; Srinivasan,

1966, 1967).

(e) Fitting Models to the distribution of parities attained within a certain period of

time by a group of women (Brass, 1958; James, 1963; Singh, 1963, 1969 Islam et

al., 1997, Balakrishnan, 1972).

The resulting estimates of mean fecundability have been less than consistent,

ranging from below 0.05 to near 1. Holemberg (1970) reviewed most of the

aforesaid studies and some of the methodological problems have been described

by Leridon (1973) and Sheps and Menken (1973). The wide range of fecundability

values may in part be due to real differences between conception rates in various

populations, but it is probably often the consequence of a lack of comparatively of

the different methods and definitions that have been used.

Each of the aforementioned methods has its particular merits and weakness.

Method 4 and 5 depend on a substantial number of assumptions and on data

inputs, which may not be available. For example James obtained estimation of

fecundability by making use of the total number of children born to couples age

over 45 in a population reported completely to avoid contraception. He also made

use of Brass model, describing the distribution of mothers of completed fertility by

number of births, to estimate the fecundability among Hutterite women aged 45

and over. In the case of these models too, assumptions has to be made without data

for guidance –namely, assumptions about the frequency of fetal wastage, length of

temporary infecundable periods, and absence of deliberate birth control throughout

the reproductive period. Further more James results yield estimates lower than

those of other works and his results have met with some skepticism.

134


The lack of reliable estimates for physiological parameter, such as the duration of

the fertile life times of sperm and ovum, constitutes the principal weakness of

method 1 (and 2). Glass and Grenbenik, Titze and Potter used two methods of the

first type to estimate theoretical values of fecundability. These models were based

on the length of the fertile period (F) and the frequency of coitus per menstrual

cycle (n). The probability that at least one coitus will occur during the fertile

period (F). Based on the assumptions that at most one effective coitus occurs

during any 24 hours is

,

and based on the assumption that coitus can occur at any time during the inter-

menstrual period of 25 days, is

.

These probabilities provide estimates of mean fecundability on the assumption that

the coitus, which occurs during the fertile period, yields a conception. Models of

this type are subject to the limitation that results from them are highly sensitive to

assumptions about the length of the fertile period and the fraction of the fertile that

overlaps with the resulting identifiable gestation, but no precise estimates exist for

either of these two factors.

Models of the 3rd type produced fairly consistent estimates of fecundability and

make relatively few assumptions but they often rely on a woman’s memory for

determining the number and timing of conceptions and also have the limitation

that they apply only to the beginning of married life.

Initial stage by Titze and potter (1960) involved arbitrary three-point distribution.

Pearson’s Type I distribution has been recommended and used by Henry (1964).

Following his work Potter and Parker constructed the Type 1 geometric model

135


with parameters a and b for predicting the time required to conceive and for

estimating the mean fecundability for women in the United States. A critical

analysis of this method has shown that moment estimates of the parameters a and

b are moderately reliable only within a specified range of a. Outside this range,

either of the estimates are extremely inefficient or their variances are not defined

at all. Furthermore, no moment estimate is defined when a is less than 2.

Estimating by the method of moments for this model has several disadvantages.

First as pointed out by Potter and Parker since the second moment of the

conception month is not defined unless a exceeds 2, the method is restricted to the

situation where the estimate is greater than 2. Secondly, although the variance of

the moment estimates has not been discussed in the record cited, it is clear that

since both estimates are the functions of the first two moments of the conception

month, the variances of the estimates will depend on the fourth moment of the

conception month, which indicate that the variance of the moment estimate

undefined when a is not greater than 4.

To overcome the limitations of moment estimates, an alternative method of

estimation, the maximum likelihood estimation, which does not suffer from these

disadvantages, was provided by Mujumder and Sheps (1970). Such estimates have

certain optimal properties including consistency and asymptotically minimum

variance and can be defined for all permissible values of a and b which are always

desirable. Jain (1969) showed that with the aid of improved measures of

contraceptive delays, one can obtained a good fit with the Type 1 geometric

model. He also presented new data on fecundability comparatively free from

memory and truncation biases at the beginning of the married life of Taiwanese

women. He also studied the relationship between age and fecundability at the

beginning of married life. He fitted a theoretical distribution to first contraceptive

delays. The great advantage of this model is that it makes minimum assumption

and has yielded relatively consistent results.

136


Brass (1958) has considered a continuous time model when it is assumed that the

fecundability of a given women is constant, but that the constant varies from

woman to woman depicting a gamma density. He has also assumed that a

pregnancy can terminate only as a live birth and that of the total infecundable

period following a conception is constant. He has derived the expression for the

probability of m conceptions in time t corresponding to the model. In fitting the

special case of zero infecundable period to data Brass has excluded the women

with no conceptions and therefore dealt with the problem of fitting a truncated

negative binomial distribution. The same probability for the special case of zero

infecundable period following a conception had been obtained earlier by

Greenwood and Yule (1920) in connection with disease and accidents, the fitting

of this negative binomial distribution to actual data had been shown by Dandaker

(1955). Sing(1963) has extended Dandekar’s discrete-time model assuming that

each women has constant fecundability and that the fecundabilities of all women

follow a beta distribution. He has obtained the expression for the probability of m

conceptions in time t in this model. Restricting the model to that portion of the

reproductive period where a woman is susceptible and waits for a conception, and

making the assumption of constant fecundability for each women and a gamma

distribution for the fecundabilities for all women Singh (1964b) has given the

expression for the distribution of the waiting time in the continuous -time case.

In the discrete -time approach, restricting again to the waiting time for a

conception, Potter and Parker (1964) have assumed constant fecundability for each

woman and suggested the Type 1 Geometric distribution as a useful model.. They

have obtained the distribution of waiting time, estimating its two parameters α and

β by the methods of moments. Singh (1964) proposed a continuous probability

distribution based on another set of assumptions for this situation and outlined a

method to obtain best asymptotically normal estimates of the parameters. These

137


estimates are obtained after several iterations starting from any set of consistent

estimates. Majumder and Sheps (1970) pointed out the limitations of these

moment estimates and have estimated the same parameters by the method of

maximum likelihood and proved that the maximum likelihood estimates fit the

data better than do the moment estimator. Singh and Bhaduri(1972) have showed

that it is relatively easier to compute maximum likelihood estimates of the

parameters of the continuous model than for the discrete Type 1 Geometric

distribution. Biswas(1973) later studied the Brass model and provide a simple

procedure to estimate the parameter of the model. The problem of truncation in

this model for waiting time has been discussed in Das Gupta (1973a), Das Gupta

and Hickman (19734) and Suchindran and Lachenbuch (1974). They have derived

truncation version of the distributions of the waiting time of first conception.

Incorporating the fetal wastage to Dandekar’s discrete model, Potter (1960, 1961)

has outlined the probability distribution of the number of births in a given time

period assuming the use of relatively effective birth control method. Singh (1963,

1964a) also modified Dandekar’s discrete-time model making the assumptions of

a fixed fecundability for each women and a beta mixing distribution to adjust for

the fecundabilities of all women and derived an expression for the probability of

m conceptions in time t. Taking the mean duration of gestation and postpartum

amenorrhea at 12 months and estimating the proportion fecund and fecundability

using Dandekar’s data, he has found a better fit than Dandekar has done. Later

restricting the to that portion of the reproductive period where a woman is

susceptible and waits for a conception , and making the assumption of constant

fecundability for each women and a gamma distribution for the fecundabilities for

all women , Singh (1964b) has give the expression for the distribution of waiting

time in the continuous cases confining the model to the part of the reproductive

period where a woman is susceptible and waits for conceptions , assuming that

138


each women has constant fecundability and that the fecundabilities of all women

follow a gamma distribution.

Sheps and Perein (1963) have used a special case of Henry’s discrete time model

namely, constant infecundable period associated with live births and still births, to

study the relationship between birth rate and effectiveness of contraceptives and to

obtain the probability generating function for the distribution of intervals between

two live births. In a later paper the same probability generating function has been

found by Srinivasan (1967). Sheps and Perrin (1966) have again used the same

special case of Henry’s discrete time model to obtain the distribution of time

required for the occurrence of a fixed number of conceptions terminating in live

birth. In their model, They have taken into account more than one type of

pregnancy, i,e., any pregnancy that terminates either in fetal wastage, stillbirth or

live birth with a given probability. The model also allows for variable periods of

gestation and of nonsuscptibility to conception.

Although Henry’s models recognize the possibility of live births and stillbirths,

they consider the infecundable period following a conception as a random variable

and do not allow for the variation in the gestation period and the amenorrhea

period separately. This limitation has been removed by Perrin and Sheps (1964) in

the model in which human reproduction is viewed as a Markov renewal process

with a finite number of states. In this model, they have assumed that at any time

after marriage (before the occurrence of secondary sterility or menopouse) a

woman is in any one of the following five states: (1) nonpregnant fecundable, (2)

pregnant, (3) postpartum amenorrhea period associated with abortion or fetal

wastage, (4) postpartum amenorrhea state associated with still birth, and (5)

postpartum amenorrhea period associated with live birth. They have also assumed

that the length of stay in each state is a random variable and that each pregnancy

terminates in one of the following outcomes: fetal death, stillbirth or live birth

with fixed probabilities. In their model, they have further assumed constant

139


fecundability for women in the nonpregnant fecundable stat, and the waiting time

for a conception follows a geometric distribution. Expression have been derived

by them for the mean and variance of different time intervals (e.g., the interval

between successive live births), and for the monthly probability of a live birth.

They have also obtained expressions for the mean and variance of live births

stillbirths, and miscarriages in a given period after marriage and the probabilities

of the different states at a given point in time.

Assuming time as a discrete random variable in the nonpregnant fecundable state

and the length of stay in any other state as the continuous random variable

Sheps(1964) has derived the distribution of time needed for conception and the

distribution of conception delays for a heterogeneous population of couples with

unequal monthly probability of conception. In a later paper Sheps (1967) has

formulated a class of stochastic models for a homogeneous cohort of women in

terms of markov renewal process on the assumption that the process is

independent of age and parity and the length of stay in the non pregnant state has

been to be continuous with the arbitrary probability density function. She has

explored the effects of various levels of contraceptive effectiveness and abortion

rates on birth rates, and the use of these models in evaluating population policies.

She has also derived the moments of the distribution of conception delays in a

group where the monthly probability of conception remains constant for any

women but varies between women in an unspecified manner. Sheps and Menken

(1971) have presented a unified approach to the derivation of the existing models

of human reproduction. While introducing the various models, they have included

a comprehensive review of the techniques and methods used in their construction.

In addition to developing a stochastic model of conjugal history in terms of

Markov renewal process, Krisnan (1971) has introduced a special type of Markov

renewal process to set up a model of human fertility taking into consideration only

140


social institutional factors.. He has also suggested alternative models of conjugal

history and fertility incorporating mortality factor explicitly into the model.

Chiang (1971) has developed a stochastic model of human

reproduction considering two transient states namely nonpregnant fecundable and

pregnant infecundable (combining both gestation and amenorrhea) and an

absorbing state (death state). The model is non homogeneous with respect to time

as he has treated the transition from fecundable state to pregnant or in

infecundable state and the corresponding probability as a function of woman’s age

and the transition from infecundable to fecundable state a function of the length of

time she has been pregnant and her age. Expressions have been derived for the

multiple transition probabilities between the three states. He has also estimated the

length of time needed to for a female to have certain number of pregnancies. As

he has considered both gestation and postpartum amenorrhea into a single state,

namely infecundable state there is some limitations in his model. The probability

that a women aged x who has been in the pregnant or infecundable state for a

length of time t will have a transition to the fecundable state in the age interval (x,

x+dx) will not only be influenced by x and t, but also on the derivation of t into its

two parts, (segments) namely gestation period and postpartum amenorrha period.

The analysis of human reproduction process using different types of models to

various kinds of data available has enable to derive several theoretical behaviour

of fertility. Mode (1972) has studied the intrinsic rate of geometric growth of a

population in terms of a modification of the Markov renewal model of human

reproduction formulated by Perrin and Sheps (1964). I n his stochastic model,

mode has taken into account the biological factors that the reproductive period of

every woman terminates eventually and that every woman runs the risk of death

throughout her life span. He has derived an expression for evaluating what

influences a population policy consisting, for example, of contraceptive practices

and laws of abortion may have on population growth. In a later paper considering,

141


considering discrete and continuous time simultaneously, Mode (1975) has

suggested a modification of classical renewal theory to construct a model of

terminating reproductive process in which waiting times among live births are age

and parity dependent. He has presented the distribution of waiting times among

live births by restricting the Perrin-Sheps model as a non-homogeneous semi-

Markov process. He has also given numerical examples illustrating how models of

human reproduction may be linked to generalized age dependent branching

process to give insight into the problem arising in the evaluation of family

planning programs. Mode and Litman (1975), and Mode (1985) also developed

the more comprehensive population growth model required for simulating the

events to the women in the simulation population.

Extending the results of Perrin and Sheps (1964), Das Gupta (1973a, 1973b) has

derived a more general probability model of human reproduction. This stochastic

model of human reproduction removes some of the limitations of the Perrin and

Sheps model. It considers the possibility of dependence of amenorrhea period on

the preceding gestation period, and also for live births, on the breastfeeding

practices of of the mother. He has formulated two models, virtually identical in

concepts are considered depending on whether time is treated as a continuous or

discrete variable. He has derived exact probability distribution of various

characteristics of fertility by solving integral equations by Laplace transformations

in the continuous model and by solving difference equations with the help of

probability generating function in the discrete model. The results he derived

include the distribution of time intervals that do not involve more than one

conception and their means and variances; the distribution of the time intervals

that may involve more than one conception and their means and variances; and the

exact probabilities of different states at time t and their asymptotic forms. He also

illustrates the applications of the models for special case and generates many

known results. In a later paper, Das Gupta and Hickman (1974) have derived

142


truncated version of the distribution of waiting time of first conception and live

births.

Singh et al. (1974) derived a probability model for a number of births to a female

during a given time interval since marriage assuming fecundability to be parity

dependent, no fetal wastage, and a constant period of non-susceptibility associated

with each conception. The distribution takes into account the changes of primary

sterility and secondary sterility following each pregnancy termination. It is

demonstrated that the model can be used to predict the number of children born to

females of a marriage cohort during a given period under different hypothetical

situations involve in family planning programmers. In a later paper Singh et al.

(1975) have derived a probability model of the number of births to a female during

a specified period of t years assuming that females have the same conception rate.

They have discussed the application of the model and as on illustration. It has

been fitted to the observed data collected in the demographic survey of Varanahi

(India), 1967-70. They have suggested that the model may be used to assess the

effectiveness of a family planning program in the case where couples want to limit

their family size after a certain number of births.

Ruzica and Bhatia (1982) have reported that in rural Bangladesh (Matlab), around

25 per cent of the husbands frequently stay away from home and almost half of

them for periods exceeding three months. A large number of married men also go

to the oil-rich Middle East countries leaving their families behind. All the above

factors affect women's exposure to the risk of conception and are responsible for

lowering fecundability.

Edmonston (1983) has developed a micro analytic stochastic model of human

reproduction with special features for use in examining reproduction in high

mortality populations of Bangladesh. The conditions of micro analytic are

143


biological, and do not rely any explicit-formulation of social and cultural aspects.

The model assumes that the biological factors (fecundability, live births, stillbirths,

fetal deaths and sterility) vary with woman’s age. The motivation of this paper is

not to estimate the exact condition of human reproduction but also to incorporate

mortality and breastfeeding aspects in a fertility simulation model and to analyze

the results of the simulation interns of major fertility factors. In both the cases, the

incorporation in the model appears to be successful. This simulation model

incorporates a logit transformation on a Weibul survival function for a

model schedule for postpartum amenorrhea and for early childhood mortality

developed by Lesthase and Page (1980) which depends on the duration of

breasfeeding. Employing the monte carlo simulation model Edmonston (1983)

has found that only use at marriage is allowed for very and the mean

family size reported in Bangladesh Fertility survey 1975-76 agrees in

general with non-contraception simulation. Fetal deaths and still births are

assumed to depend linearly on a woman’s age and fetal deaths are

geometrically distributed. Edmonston also derived age specific marital fertility

rates and display the mean birth intervals by birth order for the three ages of

marriage of 15, 17.5 and 20 years for simulation rate. He observed that the Coale-

trussel (1974) natural model fertility schedule show slightly higher fertility

through about 35 years of age.

Islam (1986) has formulated a stochastic model of human reproduction along with

lines suggested by Perrins and Sheps (1964). In this model death state s6 have

been considered as an absorbing state. Hence it is more realistic than that of

Perrin and Ships particularly in cases of developing countries where mortality rate

and in particular, the maternal mortality rate is significantly higher. (Of high

mortality populations). He has attempted to develop a stochastic model to study

the fertility pattern of two marriage cohorts of women in terms of conception

144


intervals, live birth intervals and annual probability of occurrence of live birth

under certain assumptions regarding socio-biological factors that influence

reproductive behavior. He has found a woman belonging to the 1965-69 cohorts

in nonpregnant state for 6.7 months on an average while the corresponding

average time spent by a woman in the 1970-74 cohorts is 7.2 months before direct

transition to pregnant state. Mean conception intervals are at 32.0 and 32.7

months for women in the 1965-1974 and 1970-1974 first marriage cohort

respectively. The estimate of mean live birth interval is found to be 33.6 and 36.0

month for the two cohorts respectively. His estimates reflects that average time

spent in non-pregnant state before a direct transition before a direct

transition to pregnant state, mean conception and mean live births intervals

for women in both the cohorts are found to be higher among urban than

rural women, among women with at least some schooling than those with

no schooling and among Muslims than their Non-Muslims counterparts.

Annual fertility rate per woman are estimated as 0.375 and 0.334 for the 1965-69

and 1970-94 first marriage cohorts respectively. The model estimates of fertility

for both the cohorts have exhibited negative relationships with urbanization and

education and fertility for the non-Muslims is found to be lower than that for

Muslims.

Meridith et al. (1987) attempted for the first time to disentangle the relationships

between nutrition, lactation and a woman’s monthly probability of conception

from the data taken from the Determinants of Natural Fertility Study, Conducted

under the auspices of the International Center for Diarrhoeal Disease Research,

Bangladesh. Applying Multivariate Hazard models to the study of fecundability

they observed that there is strong seasonal patterns of conception in Bangladesh,

which can more plausibly attributed to seasonal variations in coital frequency than

to seasonality in nutritional status. They showed that fecundability varies both

145


among women of a given age and for a particular women by age. The variation is

related to four variables: separation, which effects coital frequency, age, which

represent biological and perhaps decline in coital frequency; lactation practices;

and the duration of amenorrhoea. The study revealed that the longer a woman’s

post partum amenorrhoea the higher the fecundability when she resumes menses

and the more rapidly she conceives. Undoubtedly, the most sticking finding of this

study is the effect of the patterns of breastfeeding in the monthly probability of

conception as woman begins to wean her child, her probability of conceiving

increase, presumably the result of decreased levels of serum prolactin which

inhibit ovulation.

Balaksisnan (1988) analysis the probability of conceiving and conception delay by

duration from the date of entry into union from the data collected in National

Fertility Surveys in rural and semiurban areas of Mexico, CostaRica, Colombia

and Perue for all noncontrcepting women in sexual unions by applying life table

methods using both cloded and open birth intervals. However fecundability

estimates were constructed using only closed intervals in other words based on

women who had at least one pregnancy.He observed overall low mean

fecundability compared to other populations, but this difference is largely due to

the lower mean age at entry into union neither than to lower age-specific

fecundability.

Goldman and Montgomery (1990) considered a fundamental demographic issue

which surprisingly little is known: on the effects of husband’s age on

fecundability. They evaluated the effects of husband’s age on the probability of

conception from world fertility surveys data in five developing countries. The

Ivory cost, Ghana, Kenya, The Sudan and Syria. Proportional hazard model, which

include wive’s age husband’s age, marriage duration, union type and post-partum

exposure as covariates were used by them to describe the monthly conception rate

for second and higher order birth intervals in which no contraception was used.

146


With the exception of Syria, the resulting models indicate that the effect of male

age is generally small in relation to the influence of marital duration and age of the

woman.

Tewari et al. (1994) have derived some stochastic models to describe the

variation in the length of open birth interval of women having given birth to a

child during the last “T” years of their current reproductive age. Assuming the

reproduction process as steady state derives the first model, the second is obtained

by varying the fecundability parameter involved in the first model after the last

birth. The present exercise intends to propose simple cohort model of open birth

interval conforming to the steady state conditions and to estimate biological

parameters of the model along with standard error and best asymptotic normal

(BAN) estimates of parameter.

Pathak (1996) has outlined a modified stochastic model of family formation by

incorporated age patterns of marriage characterized by a displaced lognormal

distribution under some simplified assumptions. He has found that the mean age of

mothers at the time of births of the first and last offspring’s are sensitive to age

patterns of child bearing which is characterized by age specific marital fertility

rate, He has also shown that the average number of children per mother is

dominantly mounded by mortality and level of fertility

Islam et al (1997) have estimate fecundability by the model fitting technique to

data on the distribution of the number of births to women with a fixed marital

duration. The study was based on data from two national level fertility surveys, the

Bangladesh Fertility survey (BFS) 1975 and 1989. The patterns and trends of

fecundability in Bangladesh were investigated by fitting by model developed by

Singh(1969, singh et al. 1971) with a slight modification. They obtained low mean

fecundability in Bangladesh compared to western countries. The urged that the

low fecundability of Bangladeshi women as compared to that of western countries

147


may be due to a number of social, biological and cultural factors, and also the

many social taboos. One of such factors is low age at marriage of girls especially

in rural areas. For which the fecundability during the first few years of marriage

could be lower

Thomas et al. (1997) has constructed a discrete survival model that allows time -

dependent covariates to assess the influence of the covariates on time to pregnancy

(TTP). Time to pregnancy (TTP), the number of menstrual cycles it takes a

couple to conceive, is potentially an informative measure of human reproduction.

A random effect was included to account for unobserved heterogeneity. The

collected waiting times are obtained through retrospective ascertainment and are

analyzed as truncated data. Fisher scoring through iterative reweighed least

squares implemented maximum likelihood estimation. The analyze of TTP data

revealed that, fecundability increased with the mother’s age contrary to what could

be expected for prospectively obtained waiting times, whereas it decreased with

the fathers age. The analysis further revealed a weak, but significant, effect of

mother’s body mass index (BMI) around 22 had slightly higher fecudability than

women with higher or lower BMI. Finally the negative effect of recent use of

contraception appeared to wash out soon after the end of use.

Islam, Mazharul M; Yadava, R.C sept (1997) has observed that Bangladesh is a

poor country, and malnutrition and ill-health are largely prevalent particularly

among women, and may also contribute to low fecundability though the fertility

rate is rather high due to the low contraceptive use rate and/or lack of effective use

of contraception, as also due to the higher desired family size.

The BDHS 1999-2000 demonstrates lower interval between marriage to first birth

but higher interval between subsequent births than observed in the BDHS 1996-

97(Azad, 2001).

148


To control for variation in fecundity among women, we include information on the

length of time elapsing between marriage and the first birth for women (the vast

majority) who were not pregnant when they married. Only 0.3 percent of married

women in the EDHS used contraception before the birth of their first child, so the

length of the first birth interval is likely to be largely determined by a couple’s

fecundity. We do not have data on frequency of intercourse, nor on the duration of

viability of ova and sperm. These will, we hope, be captured by the inclusion in

the model of an unobserved heterogeneity term. Publications, Delhi (February,

2001), 31-46.

To investigate the pattern and differentials of interval by selected demographic and

Socioeconomic variables, they (M. Nurul Islam, Department of Statistics,

University of Dhaka and Salehin Khan Chowdhury, Daffodil University, Dhaka)

used Cox’s proportional Hazard regression model. The data used for the

completion of this work is extracted from the Bangladesh Demographic and

Health Survey (BDHS) conducted in 1999-2000. The mean of first birth interval is

slightly shorter than the mean of subsequent birth intervals and it is around 36

months. The median first birth interval is 28 months, which is four months shorter

than that of subsequent birth intervals. They observed that, about 63.4% of the

women take their first child within three years of marriage. They concluded that in

Bangladesh marriage to first birth interval decreases and subsequent birth interval

increases over time. The life table analysis of birth intervals reveals that about 76

percent of Bangladeshi women take their first child within five years of marriage

and their average first birth interval is 25 months. To identify the determinants of

birth interval Cox’s proportional Hazard Regression has been applied. This result

indicates that while controlling the other variables women living in the urban areas

tend to have a longer birth interval than the women living in the rural areas. The

life table analysis of birth intervals reveals that about 76 percent of Bangladeshi

women take their first child within five years of marriage and their average first

149


birth interval is 25 months. This percentage is almost the same for overall

subsequent births but the average subsequent birth interval is 30.5 months. The

differential analysis of quantum and tempo of fertility has been done for place of

residence, working status, level of education and region of residence. To identify

the determinants of birth interval Cox’s proportional Hazard Regression has been

applied. This result indicates that while controlling the other variables women

living in the urban areas tend to have a longer birth interval than the women living

in the rural areas. Subsequent birth interval of urban women is also longer than

rural women. In the multivariate analysis longer duration of first and subsequent

birth intervals are observed among the working women. Educated women are

likely to have shorter first but longer subsequent birth interval than women having

no education. Upper social class women used to have shorter first but longer

subsequent birth interval than middle or lower class women. The results indicate

that marriage to first birth interval increases significantly with age of the

respondent but decreases with age at marriage. An unexpected finding was that

ever users of contraception have distinctly shorter birth interval than never users.

Gender preferences also have significant effect on birth interval in Bangladesh.

Their analysis also shows that religion and region of residence have no significant

effect on marriage to first birth interval but appears to have significant effect on

subsequent birth interval.

Greater education participation can cause a delay in births because being in

education can be incompatible with having children, for financial and life-style

reasons. Moreover, after leaving education, highly educated women are likely to

spend more time in investment in job search and in finding the right job. There is

also a positive correlation between investment in schooling and investment in on

the job training, both of which combine to provide higher wages for educated

women (Gustafsson, 2001). This results in those with higher education levels

tending to have higher growth in earnings in the years immediately following

150


education. Therefore more highly educated women may wish to delay having a

child until a stable and desired career pattern has been established

Gustafsson, (2002) cited two principle reasons given by Hotz et al., (1997) for the

general postponement of first birth in Western countries; the consumption

smoothing and women’s career planning motives. The consumption-smoothing

motive relates to having enough resources to afford having a child, while the

career-planning motive refers to the need to have the time for child caring and

rearing.

Bangladesh’s adolescent fertility rate is one of the highest in the world, and

consistent high adolescent fertility is one of the main reasons for the slow fertility

decline in recent years (NIPORT, Mitra and Associates, and ORC Macro, 2005).

(Quamrun Nahar and Hosik Min 2008) have observed using four sets of

Bangladesh Demographic and Health Survey data collected during 1993/94,

1996/97, 1999/00, and 2004, the paper examines the trends and determinants of

adolescent childbearing in Bangladesh, and identifies area-level variation in

explaining differentials in adolescent first birth. Discrete-time multilevel hazard

modeling is used to estimate the hazard of first birth before age 20 after

controlling the effects of other individual and household factors. The results

suggest that the overall probability of first birth before age 20 among Bangladeshi

women remained static or even increased slightly over time. There was a

significant area level variation in teenage first birth in 1993/94 and 1996/97.

However, over time the effect of area is decreasing. At the individual level,

women’s education, especially higher education, has the strongest effect in

delaying first birth during adolescence. Age at marriage has a strong association

with age at first birth: a one-year increase in age at marriage decreases the chance

of teenage first birth by 10% or more. Frequent media exposure has a significant

delaying effect, and the effect is more distinct in the most recent year.

151


152

CHAPTERTWO


CHAPTER TWO

DATA AND METHODOLOGY

2.1 DATA

In the estimation of conception waits, fecundability and timing of first birth among

women in Bangladesh, we have used the data extracted from the response of

women questionnaire of 2007 Bangladesh Demographic and Health Survey

(BDHS). The survey was conducted under the authority of the

National Institute for Population Research and Training (NIPORT)

of the Ministry of Health and Family Welfare. The survey was

implemented by Mitra and Associates, a Bangladeshi research

firm located in Dhaka. Macro International Inc., a private research

firm located in Calverton, Maryland, USA, provided technical

assistance to the survey as part of its international Demographic

and Health Surveys program. The U.S. Agency for International

Development (USAID)/Bangladesh provided financial assistance.

2.2 SURVEY OBJECTIVES AND IMPLEMENTING ORGANIZATIONS

The BDHS is a part of the worldwide Demographic and Health

Survey Program.The 2007 Bangladesh Demographic and Health

Survey (BDHS) is the fifth BDHS undertaken in Bangladesh. This

periodic survey is conducted every three to four years to serve as

a source of population and health data for policymakers, program

managers, and the research community. In general, the aims of

the BDHS are to:

153


Provide information to meet the monitoring and evaluation needs

of health and family planning programs, and Provide program

managers and policy makers involved in these programs with the

information they need to plan and implement future

interventions. More specifically, the objectives of the survey are

to provide up-to-date information on fertility and childhood

mortality levels; nuptiality; fertility preferences; awareness,

approval, and use of family planning methods; breastfeeding

practices; nutrition levels; maternal and child health; awareness

of HIV/AIDS and other sexually transmitted diseases; knowledge

of tuberculosis; and domestic violence. Although improvements

and additions have been made to each successive survey, the

basic structure and design of the BDHS has been maintained over

time in order to measure trends in health and family planning

indicators.

2.3 SAMPLE DESIGN

The 2007 BDHS employs a nationally representative sample that

covers the entire population residing in private dwelling units in

Bangladesh. The survey used the sampling frame provided by the

list of census enumeration areas (EAs) with population and

household information from the 2001 Population Census.

Bangladesh is divided into six administrative divisions: Barisal,

Chittagong, Dhaka, Khulna, Rajshahi, and Sylthet. In turn, each

division is divided into zilas, and each zila into upazilas. Rural

areas in an upazila are divided into union parishads (UPs), and

UPs are further divided into mouzas. Urban areas in an upazila

are divided into wards, and wards are subdivided into mahallas.

These divisions allow the country as a whole to be easily divided

154


into rural and urban areas. EAs from the census were used as the

Primary Sampling Units (PSUs) for the survey, because they could

be easily located with correct geographical boundaries and

sketch maps were available for each one. An EA, which consists

of about 100 households, on average, is equivalent to a mauza in

rural areas and to a mohallah in urban areas.

The survey is based on a two-stage stratified sample of

households. At the first stage of sampling, 361 PSUs were

selected. Figure 1.1 shows the geographical distribution of the

361 clusters visited in the 2007 BDHS. The selection of PSUs was

done independently for each stratum and with probability

proportional to PSU size, in terms of number of households. The

distribution of the sample over different parts of the country was

not proportional, because that would have allocated the two

smallest divisions, Barisal and Sylhet, too small a sample for

statistical precision. Because only a small proportion of

Bangladesh’s population lives in urban areas, urban areas also

had to be over-sampled to achieve statistical precision

comparable to that of rural areas. Therefore, it was necessary to

divide the country into strata, with different probabilities of

selection calculated for the various strata. Stratification of the

sample was achieved by separating the sample into divisions

and, within divisions, into urban and rural areas. The urban areas

of each division were further subdivided into three strata:

statistical metropolitan areas (SMAs), municipality areas, and

other urban areas. In all, the sample consisted of 22 strata,

because Barisal and Sylhet do not have SMAs.

155


The 361 PSUs selected in the first stage of sampling included 227

rural PSUs and 134 urban PSUs. A household listing operation was

carried out in all selected PSUs from January to March 2007. The

resulting lists of households were used as the sampling frame for

the selection of households in the second stage of sampling. On

average, 30 households were selected from each PSU, using an

equal probability systematic sampling technique. In this way,

10,819 households were selected for the sample. However, some

of the PSUs were large and contained more than 300 households.

Large PSUs were segmented, and only one segment was selected

for the survey, with probability proportional to segment size.

Households in the selected segments were then listed prior to

their selection. Thus, a 2007 BDHS sample cluster is either an EA

or a segment of an EA.

The survey was designed to obtain 11,485 completed interviews

with ever-married women age 10-49. According to the sample

design, 4,360 interviews were allocated to urban areas and 7,125

to rural areas. All ever-married women age 10-49 in selected

households were eligible respondents for the women’s

questionnaire. In addition, ever-married men age 15-54 in every

second household were eligible to be interviewed.

2.4 QUESTIONNARES

The 2007 BDHS used five questionnaires: a Household

Questionnaire, a Women’s Questionnaire, a Men’s Questionnaire,

a Community Questionnaire, and a Facility Questionnaire. Their

contents were based on the MEASURE DHS Model Questionnaires.

156


These model questionnaires were adapted for use in Bangladesh

during a series of meetings with a Technical Task Force (TTF) that

included representatives from NIPORT, Mitra and Associates,

ICDDRB: Knowledge for Global Lifesaving Solutions, the

Bangladesh Rural Advancement Committee (BRAC),

USAID/Dhaka, and Macro International (see Appendix E for a list

of members). Draft questionnaires were then circulated to other

interested groups and reviewed by the BDHS Technical Review

Committee (see Appendix E). The questionnaires were developed

in English and then translated and printed in Bangla.

The Household Questionnaire was used to list all the usual

members of and visitors to selected households. Some basic

information was collected on the characteristics of each person

listed, including age, sex, education, and relationship to the head

of the household. The main purpose of the Household

Questionnaire was to identify women and men who were eligible

for individual interviews. In addition, the questionnaire collected

information about the dwelling unit, such as the source of water,

type of toilet facilities, flooring and roofing materials, and

ownership of various consumer goods. The Household

Questionnaire was also used to record height and weight

measurements of all women age 10-49 and all children below six

years of age.

The Women’s Questionnaire was used to collect

information from ever-married women age 10- 49. Women were

asked questions on the following topics:

157


Background characteristics, including age, residential

history,

education, religion, and media exposure,

Reproductive history,

Knowledge and use of family planning methods,

Antenatal, delivery, postnatal, and newborn care,

Breastfeeding and infant feeding practices,

Vaccinations and childhood illnesses,

Marriage,

Fertility preferences,

Husband’s background and respondent’s work,

Awareness of AIDS and other sexually transmitted

diseases,

Knowledge of tuberculosis, and

Domestic violence.

The Men’s Questionnaire was used to collect information from

ever-married men age 15-54. Men were asked questions on the

following topics:

Background characteristics, including respondent’s

work,

Marriage,

Fertility preferences,

Participation in reproductive health care,

Awareness of AIDS and other sexually transmitted

diseases,

Knowledge of tuberculosis, injuries, and tobacco

consumption, and

Domestic violence.

158


Questions on domestic violence (which were included in both the

Women’s and Men’s Questionnaires) were administered to only

one eligible respondent per household, whether female or male.

In households with two or more eligible respondents, special

procedures were followed to ensure that the selection of the

woman or man was random and that these questions were

administered in private.

The Community and Facility Questionnaires were administered in

each selected cluster during listing. These questionnaires

collected information about the existence of development

organizations in the community and the availability and

accessibility of health services and other facilities.1 This

information was also used to verify information gathered in the

Women’s and Men’s Questionnaires on the type of facilities

respondents accessed and the health service personnel they saw.

2.5 TRAINING AND FIELDWORK

Forty-two field staff were trained and organized into six teams to

carry out the listing of households and delineation of EAs and to

administer the Community and Facility Questionnaires. In

addition, six supervisors were deployed to check and verify the

work of the listing teams. Listers were also trained in the use of

Global Positioning System GPS) units so that they could obtain

locational coordinates for each selected EA and for facilities

located within each EA.

159


The Household, Women’s, and Men’s Questionnaires were

pretested in February 2007. Fourteen interviewers were trained

for the pretest. The questionnaires were pretested on 100 women

and 100 men in two rural areas in Barisal district and two urban

areas in Dhaka. Based on observations in the field and

suggestions made by the pretest teams, revisions were made in

the wording and translation of the questionnaires.

Training for the main survey was conducted for four weeks from

February 25 to March 23, 2007. A total of 128 field staff were

recruited based on their educational level, prior experience with

surveys, maturity, and willingness to spend up to five months on

the project. Training included lectures on how to complete the

questionnaires, mock interviews between participants, and field

practice. Fieldwork for the BDHS was carried out by 12 interview

teams, each consisting of one male supervisor, one female field

editor, five female interviewers, two male interviewers, and one

logistics staff member. Four quality control teams ensured data

quality; each team included one male and one female data

quality control worker. In addition, NIPORT monitored fieldwork

with another set of quality control teams. Data quality was also

monitored through field check tables generated concurrently with

data processing. This permitted the quality control teams to

advise field teams about problems detected during data entry.

Tables were specifically generated to check various data quality

parameters. Fieldwork was also monitored through visits by

representatives from USAID, Macro International, and NIPORT.

160


Fieldwork was implemented in five phases and carried out from

March 24 to August 11, 2007.

2.6 DATA PROCESSING

All questionnaires for the BDHS were periodically returned to

Dhaka for data processing at Mitra and Associates. The

processing of data collected in the field began shortly after

fieldwork commenced. Data processing consisted of office

editing, coding of open-ended questions, data entry, and editing

inconsistencies found by the computer program. The data were

processed by 10 data entry operators and two data entry

supervisors working in double shifts using six microcomputers.

Data processing commenced on April 16 and ended on August

31, 2007. Data processing was carried out using CSPro, a joint

software product of the U.S. Census Bureau, Macro International,

and Serpro S.A.

2.7 COVERAGE OF THE SAMPLE

In the next page; Table 2.1 shows the results of the household

and individual women’s and men’s interviews. Of the 10,819

households selected for the survey, 10,461 were found to be

occupied. Interviews were successfully completed in 10,400

households, or 99.4 percent of households. A total of 11,178

eligible women age 15-49 were identified in these households

and 10,996 were interviewed, for a response rate of

98.4 percent.2 Eligible men in every second household were

selected to yield 4,074 potential male respondents, of whom 92.6

percent or 3,771 were successfully interviewed.

161


Table 2.1 Results of the household and individual interviews (Number of households, number of interviews, and response rates, Bangladesh 2007)

Result

Total Number

Percent

Household SampleHousehold occupied 10461 96.7Households absent for extended period of time 199 1.8 Dwelling vacant or Destroyed 132 1.2Other 27

0.2

Total household Selected 10819 100.0

Household interviews

Household interviewed 10400 96.1

Household Not interviewed 61 0.6

Total household Occupied 1046 96.7Household Response rate 99.4

Individual Interviews: womenEligible women Interviewed 10996 98.4

Eligible women Not interviewed 182 1.6Total Eligible women 11178 100.0

98.4Eligible women Response rate Individual Interviews: men

Eligible Men Interviewed 3771

162


92.6

Eligible men Not Interviewed 303 7.4Total Eligible Men 4074 100.0Eligible men Response rate 92.6 Source: Bangladesh Demographic and Health Survey- 2007The principal reason for non-response among eligible women and men was their absence from home despite repeated visits to the household. The household and eligible women’s response rates were similar to the response rates in the 2004 BDHS. However, the male response rate was lower than in the last survey.

2.8 METHODS OF FINDING RELAVANT DATA

To estimate “the Conception wait, fecundability and Timing of first birth” for

Bangladeshi women we have extracted 1651 women out of 10996 women from

the 2007 Bangladesh Demographic and Health Survey(BDHS) who have at least

one live birth. Since our study is based on the first birth history data, so we have

excluded those of the women having first birth occuring more than 5 years

preceding the survey to avoid the memory lapse of the respondents. Next, we have

excluded the women who were pre-marital pregnant (101). After that we have

excluded those conceptions of the women whose pregnancy have terminated

before first live birth (240).To compute the mean and variance of ‘time required to

conceive’, the first pregnancy intervals (periods between marriage and first

conception), which are coded as zero months have been assigned as a value of

0.25 months. This is done because the coded value of zero months represents a

period of 15 days or less, the mid-point of which is 0.25 months.

Calculation of the dependant variable:

Since we have considered at least one live birth, for creating the dependent variable we are to follow the following:

163


Conception wait = (Date of first birth - 9 - Date of first marriage)

Age at first conception = Age at first marriage + Conception wait

2.9 INDEPENDENT VARIABLES

This study includes the several explanatory variables. However some of the

variables with their original coding can be used for the analysis of the data. On the

other hand, we recode some of the variables ignoring their original coding for the

convenience of the analysis. We also assign new coding of some independent

variables for the purpose of the easiest analysis of the data in which the coding of

the variables are not available. We also create some new variables for our analysis

by using the relative information obtained from the other variables.

The newly recoded variables are Respondent’s education level, Husband’s

education level, Respondent’s occupation, Husband’s occupation, Religion,

Current age of respondent, Use of contraception, Husband’s age, Marital duration,

Body mass index, Wealth index, Respondents age at first marriage. The

independent variable Spousal age difference is created by using the variables

Husband’s age & Women’s age in years from household report with the

computation (Husband’s age - women’s age in years from household report).Also,

the variable Mass media contact is created with the help of the variables Reads

News paper or magazine, listens to radio and watches TV. The variables Type of

place of residence and Division are recoded same as the original recoding. Above

seventeen explanatory variables are included and examined in this study. They are

classified as-

Ι. Demographic variables

II. Socio-economic variables and

III. Cultural variables

164


I. Demographic variables:

The explanatory variables that summarize the demographic behaviors of the

respondents are considered as demographic variables. Demographic variables are

includes

1) Use of contraception

2) Respondents age at first marriage

3) Current age of respondent

4) Husband’s age

5) Marital duration

6) Spousal age difference

Use of contraception:

Contraception (birth control) prevents pregnancy by interfering with the normal

process of ovulation, fertilization, and implantation. Ever use of contraception

method is an important factor in any demographic analysis. This variable is

included in this study. Ever use of contraceptive woman means any woman who

wanted to prevent pregnancy using a reliable form of birth control. In this data set

the variable is given by four categories and I categorized it into two as Not use of

contraception and Use of contraception. In the present study 21.1 percent of

women never used any method of contraception before their first birth. On the

other hand, 78.9 percent of them used any one of the contraceptive methods. This

indicates almost universality of use of contraception.

Respondents age at first marriage:

Age at first marriage is an important factor in any demographic analysis. Age at

first marriage has a direct effect on fecundability as well as conception waits &

timing of first birth. This variable is included in this study. The information on age

165


at first marriage was collected by asking the respondent ‘How old are you when

you got married?’ and reported incomplete years. Then we get age of first

marriage of respondents. Plainly speaking, age of women when they first get

married is easily known as age of first marriage. The respondents are categorized

into three categories as 11 to 15 years, 16 to 19 years and 20 years & above. In this

study 44.3 percent women get married in the age group 11 years to 15 years, 43.0

percent women get married in the age group 16 years to 19 years and only 12.7

percent women get married at age 20 years & above.

Current age of respondent:

Current age of respondent has a remarkable contribution on fecundability. Though

the variable current age of respondent has no original category, I have recoded

them into three categories as 15 to 19 years, 20 to 22 years and 23 years & above.

This study includes 36.8 percent women in the current age group 11 to 15 years,

36.6 percent women in the current age group 16 to 19 years and only 26.6 percent

women at the age 20 years & above.

Husband’s age:

Husband’s age (partner’s age) are also sub-divided into three categories such as 18

to 26 years, 27 to 32 years and 33 years & above. Here, I replaced (missing 99 =5

& System =49) by the mean value 29.88. It is evident from the research work that

32.5 percent husband’s age lies in the age group 18-26 years, 40.7 percent

husband’s age lies in the age group 27-32 years and 26.8 percent husband’s age

lies at age 33 years & above.

Marital duration:

The variable marital duration has six groups originally but I have re-grouped into

two as (0-4) & (5-29) years. In this study 60.1 percent respondent’s marital

166


duration is between 0 to 4 years and 39.9 percent respondent’s marital duration is

5 years and more.

Spousal age difference:

The variable spousal age difference is calculated by subtracting women’s age (in

years from household report) from Husband’s age. Spousal age differences are

grouped into four groups such as Less equal 4 years, 5 to 8 years, 9 to 12 years and

13 years & above. In the present study 17.1 percent spousal age difference is less

equal 4 years, 38.0 percent spousal age difference is between 5 to 8 years, 25.4

percent spousal age difference is between 9 to 12 years and 19.4 percent spousal

age difference is at 13 years & more.

II. Socio-economic variables:

The variables, which relate with social and economical status of any community,

are known as socio economic variables. These variables are important in any

demographic studies. The weighted frequency distributions of the socio-economic

variables are discussed in this chapter. In the present study, following predominant

socio economic variables have been used in various analyses and the variables are-

Respondent’s education level

1) Husband’s education level

2) Respondent’s occupation

3) Husband’s occupation

4) Mass media contact

5) Wealth index

6) Body mass index

Respondent’s education level:

In the most of the analysis performed on the context of the Bangladesh, it is

observed that education of respondents has significant effect on the fecundability

167


as well as on the timing of first birth. So, the variable respondent education level is

included into the study and it is recoded into three categories Illiterate, Primary

and Secondary & above. This study includes 12.4 percent respondents who are

illiterate, 24.0 percent respondents who are Primary educated and 63.5 percent

respondents whose education level is at secondary & above.

Husband’s education level:

Husband education level is also taken in the analysis and this education level is

also recoded same as the Respondent education level. This study also includes

23.9 percent husbands who are illiterate, 25.9 percent husbands who are Primary

educated and 50.2 percent respondents whose education level is at secondary &

above

Respondent’s occupation:

Respondents’ occupation has a large number of levels for its broad descriptive

nature. The recoding program of variables is so difficult. In this analysis their

occupations are recoded into two categories, Not working and working. This study

includes 78.3 percent respondents who are in not working group and 21.7 percent

respondents are in working group.

Husband’s occupation:

Husband occupation has also a large number of levels for its broad scope. The

recoding is done, recoding them into four categories, Agriculture, Business,

Professional worker & Others. This study also includes 20.9 percent husband

whose occupation is Agriculture, 22.4 percent husband whose occupation is

Business, 35.1 percent husband who are professional worker and 21.7 percent

husband whose occupation is others.

Mass media contact:

Whether the respondents have put to access any one of the Reads News paper or

magazine, listens to radio and watches TV or not, a new variable named mass

168


media contract is formed by computing the frequencies and categorized as No and

Yes. In this study 27.6 percent respondents have no contact with any one of the

medias as Reads News paper or magazine, listens to radio and watches TV or not.

On the contrary 72.4 percent respondents have a contact with at least one of the

medias as Reads News paper or magazine, listens to radio and watches TV or not

Wealth index:

One of the background characteristics used throughout this report is an index of

household economic status. The wealth idex used in this study was developed and

tested in a large number of countries to measure inequalities in household income,

use of health services and health outcomes (Rutstein et al., 2000). It is an indicator

of the level of wealth that is consistent with expenditure and income measures

(Rutstein, 1999). The wealth index is constructed from data on household assets,

including ownership of durable goods (such as televisions and bicycles) and

dwelling characteristics (such as source of drinking water, sanitation facilities, and

construction materials). To create the wealth index, each asset was assigned a

weight(factor score) generated through principal component analysis, and the

resulting asset scores were standardized in relation to a normal distribution with a

mean zero and standard deviation of one (Gwatkin et al., 2000). Each household

was then assigned a score for each asset, and the scores were summed for each

household; individuals were ranked according to the total score of the household

in which they resided. The sample was then divided into quintiles from one

(lowest) to five (highest). Poor class is created by merging first two quintiles,

Middle is made by taking itself from the original data and Rich class is created by

taking last two quintiles altogether. In this study 50.3 percent respondent’s wealth

index is high, 18.9 percent respondent’s wealth index is middle and 30.8 percent

respondent’s wealth index is low.

Body mass index:

169


Body mass index is one measure of obesity (or Non- obesity). The 2007 BDHS

also collected data on the height and weight of ever-married women age 10-49.

The data are used to derive two measures of nutritional status: height and body

mass index (BMI). A woman’s height can be used to predict the risk of having

difficulty in pregnancy, given the relationship between height and pelvic (basin-

shaped cavity in most vertebrates, formed from the hip-bone with the sacrum and

other vertebrae) size. The risk of giving birth to low-weight babies is also higher

among women of small stature. The cutoff point at which mothers can be

considered at risk because of short stature is normally taken to be between 140 and

150 centimeters. The BMI index is used to measure thinness or obesity. It is

defined as weight in kilograms divided by height in meters squared (kg/m2). The

main advantage of the BMI is that it does not require a reference table from a well-

nourished population. A cutoff point of 18.5 is used to define thinness or acute

undernutrition. A BMI of 25 or above usually indicates overweight or obesity.

Thus, for analysis a women is categorized into three categories according to BMI

as Under weight if BMI is less than 18.5, Normal weight if BMI is 18.6 to 24.9,

Over weight if BMI is more than 24.9(25 and more Body mass index is also

known as Quetelet index. In the present study 30.6 percent respondents have under

weight, 61.3 percent respondents have normal weight and 8.1 percent respondents

have over weight. Body mass index is a guideline used to judge whether we are at

risk for health problems associated with our weight. Body mass index is calculated

by using the formula

BMI = where Weight is in kilograms and

hight is in meters

III. Cultural variables:

170


The cultural settings and tradition has an important effect on fecundability as well

as timing of first birth. That is why we included the following cultural variables in

our study;

1) Type of place of residence

2) Division

3) Religion

Type of place of residence:

The type of place of residence of the respondents is given into two categories,

Urban as well as Rural areas.

Division:

The respondents are also divided into six administrative divisions Barisal,

Chittagong, Dhaka, Khulna, Rajshahi and Sylhet. In the present study 10.5

percent respondents are from Barisal division, 21.1 percent respondents are from

Chittagong division, 21.5 percent respondents are from Dhaka division, 15.3

percent respondents are from Khulna division, 18.5 percent respondents are from

Rajshahi division and 13.1 percent respondents are from Sylhet, Division.

Religion:

Religion of women is divided into the categories Muslims and Non Muslims. The

non Muslim category contains Hindus, Buddhists and Christians mothers,

whereas, Muslim subgroup holds only the women who come through Islam. This

study includes 90.2 percent Muslim respondents and 9.8 percent Non-Muslim

respondents.

171


The frequency and percentage distribution of all the characteristics are as

follows:

Table 2.2: Independent variables, their categories, ranges, number of events (n) and

percentage in each category.

Name of the variables

Categories Ranges Number of respondents (n)

Percentage(%)

.Respondent’s education level

0

1

2

Illiterate

Primary

Secondary & above

205

397

1049

12.4

24.0

63.5

.Husband’s education level

0

1

2

Illiterate

Primary

Secondary & above

395

427

829

23.9

25.9

50.2

172


Respondent’s occupation level

0

1

Not working

Working

358

1293

78.3

21.7

Husband’s occupation

1

2

3

4

Agriculture

Business

Professional workerOthers

345

369

579

358

20.9

22.4

35.1

21.7

Type of place of residence

1

2

Urban

Rural

617

1034

37.4

62.6

Division

1

2

3

4

5

6

Barisal

Chittagong

Dhaka

Khulna

Rajshahi

Sylhet

174

348

355

252

305

217

10.5

21.1

21.5

15.3

18.5

13.1

Religion1

2

Islam

Others

1490

161

90.2

9.8

Current age of respondent

0

1

2

(15-19)

(20-22)

(23-42)

608

604

439

36.8

36.6

26.6

Use of contraception

0

1

Not use

Use

349

1302

21.1

78.9

Social status1

2

High

Middle

831

312

50.3

18.9

173

Table continued....


3 Low 508 30.8

Husband’s age

1

2

3

(18-26)

(27-32)

(33-70)

536

672

443

32.5

40.7

26.8

Marital duration 1

2

(0-4)

(5-29)

993

658

60.1

39.9

Spousal age difference

01

2

≤4(5-8)

9 +

283628

740

17.138.0

44.8

Body mass index

1

2

3

Below 18.5

18.5-24.9

24.91-37.79

499

998

132

30.6

61.3

8.1

Mass media contact

0

1

No

Yes

455

1196

27.6

72.4

Age at first marriage

1

2

3

11-15

16-18

19 +

731

609

311

44.3

36.87

18.80

2.10 ANALYTICAL APPROACH

In the present study, the main aim is to study mean age at first marriage, mean conception

wait, mean fecundability, mean age at first conception and mean age at first motherhood

among Bangladeshi women. For the purpose the methodology applied are, i) cross

tabulation analysis, in this section, we have tried to trace out those independent variables

174

Table continued....


which have a significant effect on first conception wait. ii) Cox’s multivariate

proportional Hazard regression analysis, in this study, we have tried to find out those

independent variables which have a simultaneous significant effect on marriage to first

conception wait. iii) Geometric and beta geometric distribution to find levels and

differentials of both conception wait and fecundability. iv) Finally, a path analysis is done

to observe the direct and indirect effects of the significant variables (in Hazard analysis).

175

CHAPTERTHREE


CHAPTER THREE

ESTIMATION OF CONCEPTION WAITS AND FECUNDABILITY: LEVELS, TRENDS AND DIFFERENTIALS

3.1 INTRODUCTIONBangladesh is a poor country, and malnutrition and ill-health are largely prevalent

particularly among women, and may also contribute to low fecundability though

the fertility rate is rather high due to the low contraceptive use rate and/or lack of

effective use of contraception, as also due to the higher desired family size (Islam,

Mazharul M; Yadava, R.C sept 1997. 43(3). p.13-20. Location: SNDT

Churchgate.) Fecundability effects fertility through its relationship with the

average time required for a conception to occur, and can also be thought of as the

transition probability for the passage from susceptible state to pregnancy (Perrin

and Sheps, 1964). The human reproduction starts from the on set of marriage and the

timing of first conception following it depends on the biological characteristics of

women. As such the analysis of the waiting time of first conception signifies couples

fertility at early stage of married life. The time, a woman takes to conceive for the first

time after her marriage is called the first conception wait or conception delay. A

176


conception delay is defined as the exposure months preceding, but not including, the

month of conception, whereas the conception wait or the time required to conceive

includes that month as well (Potter and Parker, 1964). To avoid ambiguity between a

“conception delay” and “waiting time of conception” the present analysis has considered

the conception wait. Only pregnancies recognizable by the delay of the first menses after

fertilization is included in this analysis, and fertilized ova that fail to implant or abort

spontaneously before the women knows she is pregnant are not counted as conceptions

The time of first conception, especially when the age at effective marriage is under teens,

is said to be influenced by a number of biological, social and cultural factors. The

biological factors include the age at menarche, length and level of adolescent sterility

while the social factors include the visit of female partner to parents for the observance of

festivals, social functions, and other taboos causing physical separation between the

spouses. It is often very difficult to ascertain the exact age at attaining maturity and thus

the actual waiting period for the first conception for such a woman after her marriage.

Because of this biological phenomenon, fertility behavior of women who marry in

adolescence is quite different from that of women who marry at late ages. Thus, analysis

of the first conception wait following marriage has required special attention for the

population where women marry under the age of 20 years, adolescent sterility elongates

the time until their first conception and they can not be treated fecund right from the time

of marriage (Prathak and Prasad,1977). Thus a woman takes several months to conceive

after entering the susceptible state. A woman is assumed have become biologically

mature when her menstrual cycle becomes ovulatory and she continues to ovulate

regularly (Pathak, 1978). She may enter the susceptible state by marriage or resumption

of menses after a live birth while living with her partners. It has been observed that for a

homogeneous group of women, the reciprocal of mean waiting time for first conception

gives the arithmetic mean of fecundability, whereas for a heterogeneous group of women,

the reciprocal of the mean waiting time for the first conception gives the harmonic mean

of fecundability (Henry, 1972; James, 1963; Sheps, 1964). Since fecundability is

inversely related to conception wait, estimating fecundability, conception wait may be

177


calculated either from marriage to first birth or to first conception or from subsequent

birth intervals.

After a birth a woman takes some time to have resumed her menses. In this time

she passes postpartum infecundable period due to breastfeeding. Waiting times

tend to be longer for younger women in the years immediately following

menarche (Balakrishnan, 1979; Jain, 1969), presumably because the incidence of

anovulatory is then higher. Conception waits are also longer among women who

experience prolonged period of separation from their spouses due to seasonal

migration for work (Bomggarts, 1983). But in general women beyond adolescent

the interval between marriage and first conception or first birth is, on average,

substantially shorter then the interval between first and second birth and also

higher order births (Gautier and Henry, 1958; Islam and Islam, 1995; Henripin,

1954). Plausible explanation for the difference in duration is the existence of an

infecundable period called postpartum amenorrhea period because both intervals

contain gestation and ovulatory segments of approximately the same duration and

conception are subject to similar risk of intrauterine mortality (Henry, 1964).

Besides, the interval following a birth is held by the prolonged duration of

breastfeeding, malnutrition, which increase the duration of postpartum

infecundable period

Therefore, the waiting time for first conception is generally measured from the

date of marriage. However, the assumption that the female is exposed to the risk of

conception immediately after marriage has some biological implications because

waiting time of first conception has some unique features to investigate it. Firstly,

the effect of amenorrhea period which is random and is largely determined by the

social and cultural factors of breastfeeding practices among women is completely

absent. Secondly, a woman, generally, does not like to use contraceptives to

postpone first birth. Thirdly, there is little chance of recall lapse in reporting the

178


time of first birth from the date of marriage because birth of the fist child is the

most important event in the conjugal life of a couple. As a result, the time of first

conception or birth is determined largely by the biological characteristics of the

women and can, therefore, be utilized for estimating various biological

determinants of human reproduction, the knowledge of which may be helpful in

the assessment of the impact of contraception and induced abortion along with the

impact of social and economic factors on fertility.

Determinants of biological parameters of human reproduction from the data on first

conception waits has been rendered possible through the development of some

probability models under the assumption that the occurrence of conception to a married

women depends on chance. The development of such models has considered the time

discrete as well as continuous. Although treating the time elapsed from the marriage or

from the beginning of reproductive process to the first conception as continuous makes

mathematical treatments more convenient and easy, the initial development treated time

as discrete. Discretisation of the time seems more logical at the micro level as the

reproductive life of a woman can be treated as a sequence of menstrual cycles.

In certain populations, such as India, Pakistan, Nepal and Bangladesh, where most of the

females consummate their marriage at adolescence stage, they may not be exposed to the

risk of conception immediately after marriage because of the adolescent sterility or due to

some socio-cultural taboos. The later may include the frequent visits of female partners to

their parent’s house especially in the beginning of their married life (Yadava, 1971). To

cope up with the with such problems it has been suggested the model of waiting time of

first conception by assuming that certain proportions of female are adolescent sterile at

the time of marriage(Prasad et al.,1977 ; Nair,1983). Further, in some situations it has

been reported that some of the women who are pregnant at the time of marriage report to

have conceived within a short interval of time (generally first months) after marriage

(James, 1973). In such cases, the observations will be inflated in the first few months. To

overcome the problems of premarital conceptions, inflated form of probability

179


distributions of the waiting time for first conception have been suggested (Singh, 1964;

Pathak, 1967; Singh et al., 1976). There has been another problem with regard to the data

collection that the duration of the observation period for collecting data on first

conceptive delays is not sufficiently large in the sense that one can not wait till the last

women gives birth. Das Gupta and Hickman (1974) and Suchindran and Lachenbruch

(1974) have derived truncated version of the distributions of the waiting time of first

conception. In addition, Dwivedi(1985) accounted for both, the period of temporary

separation between husband and wife and the adolescent sterility among females at early

days of married life while analyzing the first conception wait. Battaacharya et al.,(1988)

have proposed a time dependent probability model for the study of first conception

interval under the consideration that female may be exposed to the risk of conception at

different points of time after marriage. They considered the risk of conception as a

second degree polynomial of time in order to derive the true level of fecundability in the

population.

In traditional populations like Bangladesh where marriage is almost universal and early

marriage is vogue, all the female are not exposed to the risk of conception immediately

after marriage either mainly due to biosocial immaturity or due temporary separation

from their spouses. Since premarital conception is forbidden in Bangladesh, inflated form

of probability distributions of the waiting time for first conception has not been

considered in this analysis. Since data has been used in this study mainly based on

retrospective enquiry, right censoring occurs because the information about the duration

of risk period is incomplete due to limited observation period. As with the timing of a

first conception, we usually can not follow all sampled women until they either have a

fist conception or reach menopause and are no long able to have a first conception.

Therefore, instead of deriving the truncated version of the distributions of waiting time of

first conception proportional hazard model has been employed in this analysis. In

addition, temporary separation of husband and wife before first conception was not taken

in to cognizance in the present study. To estimate the Levels, Trends and

Differentials of fecundability; we are going to apply” Geometric distribution for

180


Homogenous women and type 1 geometric distribution for Heterogenous women”.

These two cases are discussed in brief in bellow:

3.2 MODEL OF HOMOGENEOUS FECUNDABILITY (GEOMETRIC

DISTRIBUTION)

The process of human reproduction starts from the on set of effective marriage and

the timing of first conception following it depends on the biological characteristics

of women. According Gini (1924), conception is a random event even though all

the possible biological and sociological factor influencing conception are

controlled. This randomness of conception gave a clue for the application of

Probability Theory. Treating fecundability to be constant for a long span till she

conceives and time as a discrete random variable Gini (1924) derived the

geometric distribution for the time of first conception. Further applications are

found in the Works of Henry (1953), Henripon (1954) and Vincent (1961). Gini’s

results obtained the mean fecundability of the population and the co-efficient of

variation from the data on the proportion of women conceiving during the first and

second months of exposure to risk. The simplest case to consider is that in which

fecundability is not only constant among women but also from month to month.

Let for a woman P, 0<P<1, be the Probability of conception in any month and

assume that the month represent independent trials. If the month in which

conception occur is denoted by the random variable T, then Prob(T=1)=P which is

the probability that the conception occurs in the first month. If T=t such that t>1,

then the preceding (t-1) months conception has not occurred with probability

and first conception occurs in the t-th month with probability P.

181


Therefore, we can write

is the desired formula for the

probability density function of T. This probability function is, of course, that of the

well-known geometric distribution.

The probability function is a monotonic decreasing function of t with a mode, the

most probable value, at t=1. The survival function at time t, s(t) defined by

s(t)=prob[T>t] is given by . The distribution function of the

random variable T is given by

The mean time of conception is , represent the mean number of

ovulations before conception. The mean time to conception is between m and m-1,

depending on the interval that separated marriage from the first ovulation thus

follows.

The variance of distribution is

In the homogeneous case, the mean waiting time for the first conception is

Since , we underestimate fecundability if we use homogeneous

model.

The moment estimator of P is . For the geometric distribution, the maximum

likelihood estimator coincides with the moment estimator.

182


3.3 MODEL OF HETEROGENEOUS FECUNDABILITY

In the last centuries, demographers have employed a variety of techniques to study

the mean value of fecundability and it’s distribution discussed in chapter one.

Among these techniques, the most commonly and widely applicable technique, is

the fitting of a theoretical distribution to the observed distribution of waiting time

to conception. In such case the waiting time to conception or the conception

interval is measured by subtracting the date of first marriage from the date of first

conception. The theoretical fitting of the type I geometric distribution has the

limitation that it can be used only after the beginning of the marital life of the

women as the conception interval for the pre-marital pregnant woman are not

possible to calculate. However, the model has the great advantages that it makes

minimum assumption and is equivalent with the Gini’s definition of fecundability

because the range of the Pearsonian type I geometric distribution lies between 0

and 1. Pearson’s type I distribution has been recommended and used by the Henry

(1964) for the first time to study the mean value of fecundability. Following his

work, Potter and Parker (1964), Majumdar and Sheps (1970) and Anrudh Kumar

Jain (1969) fitted the type I beta geometric distribution for predicting the time

required to conceive and for estimating the mean fecundabilities for women in the

United States, Princeton fertility survey data and Taiwanese women respectively.

The Model

For analyzing the data on the conception interval the type 1 geometric distribution

is considered as a useful model. The model relies on the following assumptions:

(i) The fecundability of each couple remains constant from month to month

until pregnancy.

183


(ii) Among couples, fecundability is distributed as a Pearson Type 1 curve,

i.e., Beta distribution with parameters and b.

(iii) Conception is a random event conditional on fecundability.

(iv) The number of couples is large.

In a population, not all women may be having same fecundability to bear children.

There is simple evidence that couples vary in their fecundability. A significant

proportion of sexually active couples get pregnancy in their first non-contraception

cycle, a smaller proportion of the remaining couples conceive in the conception

rate continues to decline as the risk dampens. Therefore, the first assumption may

be violated if spouse are temporarily separated, if the couple intentionally changes

the timing and frequency of intercourse, and / or if a miscarriage of six or eight

weeks is not reported. Even among healthy, regularly menstruating women, the

proportion of anovulatory cycles is put at 5 percent or thereabouts ( Potter, 1961)

During the period of separation occurs for a short period and does not coincide

with the fertile period during the month, then it will no effect on the monthly

probability of conception. The pronounced decrease in the probability of

conception over time is not purely due to time effect but as a sorting effect in a

heterogeneous population (Leridon 1977; Weinberg and Gladen, 1986). As such,

the fecundability, P of a particular woman, which is assumed to be constant earlier

from month to month may be though as a realization of the random variable P.

hence the distribution of T is the conditional distribution of T for given P, that

the unconditional probability that a conception occurs at t for a randomly selected

couple is given by

184


t=1,2,3,…….. is the probability density function of the random variable T. it is

easy to see that is a proper probability density function for each P, then so

is h(t).The given probability density function are frequently referred to as mixture

distributions.

It can also be shown that the variance of the waiting time of first conception in

case of heterogeneity is always greater than the variance of the same in the

homogeneous estimation. For the application of the above Mathematical model,

we need to have a specific from of f(P). A Judicious Choice of f(P), when is

the probability density function of the geometric distribution is the well-known

two-parameter Beta distribution whose probability density function is

Where, 0<P<1, a>0 and b>0

The distribution is also known as a Type 1 Geometric, so named because of a

classification system introduced by the British Statistician, Karl Pearson. Pearson

Type 1 distribution has been recommended first by Henry (1961) to study the

fecundability. Following his work Potter and Parker (1964) constructed the Type I

geometric model for predicting the time required to conceive and for estimating

the mean fecundability for women in United States. Pearson type 1 distribution is

convenient and gives a good approximation to unimodal distributions that are

encountered in reality, for a variable that, like fecundability, ranges between 0 and

1. The normalizing constant B(a,b) is the famous beta type 1 function defined by

185


; which may also be written

with and

and ,

The fundamental equation connect the gamma and beta function is

The mean , mode and variance of fecundability are

, =

and

For a>1 and b>1, the mode of the Beta geometric distribution M (P) is given by

.

It is, moreover, simple to calculate the value of the coefficient of variation of

fecundability in case of a beta distribution. Starting from

and ; where c is the coefficientis of variation

and v is the variance of fecundabilitry.

The proportion of conceiving during the first month of exposure is given by

186


More generally, the number of conceiving during the jth month of exposure is

equal to

for j=2,3,…………..

and

The number of women conceiving during the jth month is

N(j) = N.P(j)

Where, N is the total number of women in the sample.

The rate of conceiving in month j is given by

Under the assumption of Type 1 geometric model the theoretical average and

variance of time required to conceive are given by the following expression:

187


and this is true when a>1. Simply, the variance of time of first

conception can be obtained as

if a>2

3.4 ESTIMATION OF THE PARAMETERS

There are various statistical methods for estimating the parameters. Among them,

the method of “moments” is used in this context for it’s simplicity.

Method of Moments:

Let m1 and m2 are the two observed first and second sample raw moments of the month

required for women to conceive for the first time after their marriage. The corresponding

two population moments of T about origin, conditional on p as given by the simple

geometric distribution are

To obtain the unconditional moments of X we also have

and

188


Hence the variance of T is given by

Now equating m1 with 1 & m2 with 2/ we get

or

and

or

or

or and

Note that s2m(m-1)0; otherwise a will be less than 2 for which the theoretical

variance S2 is not defined. Whenever this condition is not satisfied the Type-1

geometric model can not be fitted. Again to find the variance-covariance matrix of

the moment estimate of & b we need the third and fourth moments of T since the

estimate of & b are the functions of the first two moments of T and in this case

the method is restricted to the situations when a is greater than 4. The analysis of

the method of moments has shown that the moment estimators of & b are

moderately reliable only within a specified range of values of a. Outside this

range, either the estimators are extremely inefficient or their variances are not

defined at all. Fortunately our moment estimates of a and b for 1651 women

189


having at least one recognizable effective conception in last five years preceding

the survey after their marriage are 4.195 and 55.32.

The values of and v2 can be estimated by substituting the estimated values of

and b in the formulae of mean, mode and variance of fecundability.

The expression for estimating is given as

Estimated

It can be seen that the estimated is a function of mean conception delay m and

variance s2. If there were some unreported pregnancy losses and some women had

reported their first live birth, instead of miscarriage or abortion as their first

pregnancy, than the estimated mean m would be higher than true M and the

estimated would be deflated. The under reporting of pregnancy losses would

also increase the estimated variance s2, but this will have smaller effect on the

estimated , because s2 appears both in denominator and numerator. Similar

expressions for p/ and v2 can be obtained easily.

The theoretical values of Pj and Nj for j = 1,2,3,..........etc. are obtained by using

FORTRAN language (Appendix –1).

The survival function, in this case, is given by

, which is valid for t>1.

Recursively, it can be seen that

190


, t = 1,2,.......; where s(0) = 1.

3.5 VARIANCE-COVARIANCE MATRIX OF THE MOMENT ESTIMATORS

For large sample, this asymptotic variance of , and covariance of and

is given by Rao(1952).

Var( ) = A12 var(m1) + A2

2 var(m2) + 2A1A2 cov(m1m2) = 4.76

Var( ) = B12 var(m1) + B2

2 var(m2) + 2B1B2 cov(m1m2) = 1812.18

and cov( ) = A1B1 var(m1) + A2B2 var(m2) + (A1B1+A2B2) cov(m1m2) = 92.83

Where,

= 0.769

= -0.0076

= 16.95

191


= -0.149

If mr and r/ are respectively the rth sample and population moments about zero

then by Kandall and Stuart (1958)

= 0.367

= 92869.77

and

= 70.30

Here ;

= 18.32 = 941.47

= 133311.66

= 154258935.1

192


Hence the variance-covariance matrix (V1) of the estimates of the parameters a and b obtained by the method of moments is given by

The correlation co-efficient between the estimators of the two parameters given by

= 0.99966

The value of co-efficient of correlation of the estimators obtained by the method of

moments indicates that they are highly correlated. The estimates of the arithmetic

mean fecundability and harmonic mean fecundability are obtained as

=0.070

= 0.055 and the estimate of the variance of fecundability is

= 0.0011

These mean fecundabilities and variance are obtained by using the marriage to

first conceptions interval and parameters are estimated for all month’s conceptions

of 1651 women having at least one conception in the last five years preceding the

survey.

And the corresponding variances of the aforementioned estimates are

193


=0.00027

= 0.0000213

= 0.000001101

The variance of exists only when the relevant covariance matrix exists.

where =0.000207

and = -0.0000353

3.6 GOODNESS OF FIT

The model is fitted with data extracted from the 2007 BDHS which provides

conception waits of 1651 women who had not aborted before first conception.

Before fitting the model we have to make little adjustment to data for computing

mean & variance of conception waits

(1) For computing the mean and variance of time required to conceive, those first

pregnancy intervals period between marriage and first conception-which are coded

as zero months have been assigned value of 0.25. This is done because the coded

value of zero months represents a period of 15 days or less, the mid-period of

which is 0.25 months, coded months j (j>0) represent the number of women who

conceive between (j-1/2) and (j+1/2) months, and in the cases j represents the

mid point of the interval (j1/2).

(b) For comparing the estimated number of women conceiving during jth month

with observed one, the later values needed some adjustment. In order to make

194


observed values comparable to the estimated values, the coded values have been

estimated as follows:

(i) Women conceiving during first month = (Women for whom the length

of first pregnancy interval is coded as zero) + ½ (women for home it is

coded as one)

(ii) In general, women conceiving during the jth month (j >1) =1/2[women

for home the length of the first pregnancy interval is coded as (j-1)

months +women for whom it is coded as (j) months].

The observed and expected data are presented in table (3.1). It is seen from the

Table that the average interval between marriage and first recognizable effective

conception wait is 18.32 months. The observed proportion of women conceiving

during each successive months after marriage are compared with the

corresponding theoretical proportions obtained in the basis of the Type 1 beta

geometric model fitted by methods of moments. The results are shown in Table

3.1 and Figure 3.1. Contingency Chi-square test statistic has been employed to examine

whether each of the selected socioeconomic and demographic factors has significant

association with selected independent variables. The calculated value of chi-square is

found 27.37. Whereas, the tabulated value of chi-square at 18 degrees of freedom with

5% level of significance is 28.87. Therefore we can say that the data of this study poorly

fitted the beta geometric distribution.

195


Table 3.1.Observed and estimated(a) number of months

required to conceive for Bangladesh. Months

required

to conceive

(xi)

Observed

Number of

women

(Oi)

Method of MomentsExpected

Number of women

(Ei)

196


≤ 34-78-1112-1516-1920-2324-2728-3132-3536-3940-4344-4748-5152-5758-6465-7273-7980-9091-109110-138157-303

40023119519112110280565641272523231611111210119

405.12279.4207.31177.52118.2890.1869.6954.5238.1336.4429.8023.6220.5525.0017.7612.9510.1311.399.179.045

0.068.380.7311.0240.061.551.530.048.370.570.260.0810.2920.160.170.2940.070.030.0750.4243.2

Total 1651 1651 27.37

Mean 18.32 - -

Variance 606.07 - -

4.195 -

55.78 -

(0.05;18) 28.87 (Tabulated) 27.37

(Calculated)

(a)Estimated on the assumption that fecundability among couples is distributed

according to a Pearson Type-I- curve.

(b)Women who are definitely pre-martially pregnant and or who had never conceived by the time

of interview are excluded.

(c) Periods required to conceive are presented in “ordinal” months.

197


Number of months required for first conception: Observed and Expected (Method of moments)

012345678

1 10 19 28 37 46 55 64 73 82 91 100

109

Number of Month

Per

cent

age

of w

omen

Observred

Expected

Figure-3.1 Observed and Estimated months required to conceive (method of

moments) for females, Bangladesh, 2007.

Table-3.2. Estimates and their standard errors by the method of moments

Estimates Method of Moments

198


4.195 2.18

55.32 42.57

r 0.9997

0.070 0.0164

0.055 0.0046

Var( ) 0.0011 0.00105

c 0.234

Note: r is correlation coefficient of a and b and c is coefficient of variation of

fecundability

The estimated mean fecundability –the monthly probability of conception for this

study is 0.070 0.0164 and the estimated modal fecundability is 0.056, which is

less than the mean. According to the pearson’s measure of skewnees this indicates

that the distribution of fecundability is positively skewed.

3.7 EFEECT OF MEMORY AND TRUNCATION BIASES ON

FECUNDABILITY

In estimating fecundability there is a possibility to occur two types of biases: (a)

‘memory’ bias and (b) truncation bias. As a result of memory bias the overall

fecundability estimate is deflated and it is inflated by the truncation bias. The

memory bias arises because in the BDHS survey the information about pregnancy

histories was collected retrospectively at the time of survey. Usually two types of

memory bias which may effect the mean fecundability are: (a) underreporting of

pregnancy losses and (b) misreporting of the time of marriage, date of birth and

hence the time of first conception. The effect of unreported conceptions prior to

the first reported conception is to lengthen the first pregnancy interval and hence

to reduce mean fecundability because it can be observed from the expression **

that the estimated is a function of mean conception wait m and variance s2. If

199


there were some under reporting of pregnancy losses and some women had

reported their first live birth, instead of a miscarriage or abortion as their first

pregnancy, then the estimated m would be higher than true mean M and

consequently would be deflated. The underreporting of pregnancy losses would

also increase the estimated variance s2, but this will have smaller effect on the

estimated , because s2 appears both in denominator and numerator. On the other

hand the effect of memory bias depends upon the nature of the error: if the first

pregnancy interval is lengthened then the mean fecundability will be reduced and

if the first pregnancy interval is shortened then the mean fecundability will be

increased. It may be assumed that recall lapse is function of the length of the

period between the occurrence of an event and its reporting date. Therefore,

duration of marriage being a good measure of this elapsed period, it may be

assumed that both types of memory bias will increase with marriage duration.

The truncation bias arises because the interview date curtailed the observed child

bearing experience of women as the survey was conducted retrospectively. In

estimating overall fecundability we included women with at least one pregnancy;

hence given short marriage duration, a woman could not be included in the

analysis at all unless her first recognizable effective conception is considerably

prompt. This indicates that short marriage duration with a conception select for

quick conceptions, thereby increasing mean fecundability. On the other hand, if

long marriage duration is included comparatively sub-fecund women will also

included in the analysis and the overestimation will be less. Thus it can be said

that the longer the marriage duration the less the effect of truncation on mean

fecundability.

To study the effect of both memory and truncation biases on mean fecundability it

is necessary to study the variation in fecundability, it is necessary to study the

variation in fecundability by marriage duration. The findings are shown in Table

3.3. Duration of marriage of the respondents, who have at least one recognizable

200


effective pregnancy, can be divided into different segments with distinct mean

fecundabilities. The result indicates that mean fecundability is very high for

women married 48 months or less and it declines substantially as we pass

successively from one segment to the next. The sharp decline in mean

fecundability from short marrital duration to long marrital duration is probably due

to the positive effect of truncation bias and the negative effect of memory bias. It

seems quite reasonable to assume that mean fecundability for women married for

5-9 years would provide a closer estimate of fecundability for Bangladeshi women

then that based on all women. Women married for 5-9 years are not as seriously

affected by the two biases, as are all women. Moreover, the effects of these two

biases on mean fecundability for the women can be assumed to be compensatory.

If this were so, the mean fecundability for women in this sample will be closer to

0.0403.

Table-3.3 Estimates of conception delay and fecundability for women by marital

duration, Bangladesh.

Marital

duration

Mean

conception

Variance Fecundabilities(a) Number

of

201


in year delay(months) women

0-4

5-9

10-14

15-19

20-24

25-29

9.41

24.82

92.67

170.00

231.67

281.50

74.504

326.287

503.154

86.000

1408.333

924.500

0.1063

0.0403

0.0108

0.0059

0.0043

0.0036

993

607

42

4

3

2

All

women

18.3150 606.068 0.055 1651

(a)Fecundabilities are estimated on the basis of Geometric distribution as the

individual marriage cohorts are homogeneous.

202


Figure-3.2. Mean conception wait by marital duration

The trend line regarding marital duration and mean conception wait indicates that

up to 9 years mean conception wait is increasing narrowly, after that it is sharply

increasing.

203


0

0.02

0.04

0.06

0.08

0.1

0.12

0-4 5-9 10-14 15-19 20-24 25-29

Marital Duration(in Year)

Fec

un

dab

ility

Mean Fecundability

Figure-3.3. Mean fecundability by marital duration

The trend line regarding marital duration and mean fecundability indicates that up to 9

years fecundability is decreasing sharply, after that it is narrowly decreasing.

204


3.8 AGE AT MARRIAGE, CONCEPTION WAIT AND FECUNDABILITY

Marriage marks the beginning of childbearing period, as it is the starting point of

legitimate sexual union between husband and wife. Fecundability being dependent

upon physiology, the age at marriage will play significant role in determining its

level. The prevailing generalizations in the literature about the variations in mean

fecundability with age as fecundability is zero until some age a, then increases to

some age b, passes through constant maximum from age b to age y and then

decreases until age z when it becomes zero again ( Henry 1965). Ignoring foetal

deaths Vincent’s estimates of mean effective fecundability for newly wed French

women, aged 16 to 25 years, and supported a part of Henry’s hypothesis. We have

an opportunity here to observe the variation in fecundability for Bangladesh by

applying theGeometric model to age at marriage classes.

Estimates of mean conception delays shown in Table 3.4 reflect that women who

marry before 14 years of age take the longest time to conceive and the delay

reduces with increasing age at marriage. The conception delay is shortest when

reaches at the age 21 years and above. Like time required conceiving, the mean

fecundabilities based on Geometric model for each age at marriage group are

shown in the same Table 3.4. The result indicates that fecundability varies with the

age at marriage group. These results also show that fecundability increases with

the age at marriage. The findings suggest that the mean fecundability is lowest

when women are in their early teens; it gradually increases as women move to

their late teens and age 21 and above. The trend of fecundability is constant with

Henry’s (1965) hypothesized relationship between age and fecundability, and with

Vincent’s (1961) estimates of effective fecundability by age. The same trend was

also obtained by Jain (1969) for the Taiwanese women. Our findings did not tell

us much about fecundability after age 23 years. In this sample there are only 53

women who married at age 24 years and above, and the mean fecundabilities of

these women are higher than that of women of other age at marriage class.

205


One of the plausible explanations for the longer conception delays at early age at

marriage is that in Bangladesh most of the females consummate their marriages

under the age of 20 years and they can not be treated right from the time of

marriage ( Pathak and Prasad, 1977). A woman is assumed to have become

biologically mature when her menstrual cycle becomes ovulatory and she

continues to ovulate regularly (Pathak, 1978). The time interval between menarche

and attainment of full biologically maturity is called the adolescent sterility. Thus

the probability of conceiving in a month in Bangladesh is lower at early age at

marriage is mainly due to adolescent sub-fecundity of the respondents or due to

some socio-cultural taboos.

206


Table 3.4 shows that as the age at first marriage increasing mean conception wait

is decreasing resulting fecundability is increasing. From this table it also reveals

that the significant correlation coefficient between age at first marriage and

conception wait is -0.154, which discloses that age at first marriage is negatively

correlated with marriage to conception wait. We also get the significant

Regression coefficient between age at first marriage and conception wait is -1.252

which indicates that if age at first marriage will increase by one year, conception

wait will decrease by 1.252 months.

Table 3.4: Relationship among age at first marriage, conception wait and

fecundability


Mean conception wait

Standarddeviation

Correlation coefficient

Regressioncoefficient

P-value

Fecundability Numberof respondents

≤13 26.03 33.514

-0.154 -1.252 0.000

0.038 23014 22.70 30.851 0.044 23515 22.23 28.869 0.045 26616 14.53 16.101 0.069 24017 14.22 16.017 0.070 20018 13.39 13.532 0.075 16919 13.99 16.709 0.071 10120 14.61 17.599 0.068 7421 16.05 25.841 0.062 3922-23 17.11 27.479 0.058 4424-25 12.03 12.682 0.083 2626+ 8.33 10.781 0.120 27Total 0.055

207


0

5

10

15

20

25

30

≤13 14 15 16 17 18 19 20 21 22-23

24-25

26+


Me

an

co

nc

ep

tio

n w

ait


Figure 3.4 Mean conception wait by age at first marriage

The trend line regarding age at first marriage and mean conception wait indicates

that up to 18 years of age at first marriage, mean conception wait is decreasing

narrowly after that it is narrowly increasing up to 23 years of age at first marriage

thereafter it is sharply decreasing.

208


209


0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

≤13 14 15 16 17 18 19 20 21 22-23 24-25 26+


Fec

un

dab

ility

Mean Fecundability

Figure-3.5 Mean fecundability by age at first marriage

The trend line regarding age at first marriage and mean fecundability indicates that

up to 18 years of age at first marriage, fecundability is increasing narrowly after that it

is narrowly decreasing up to 23 years of age at first marriage thereafter it is sharply

increasing.

210


3.9 INTER DEPENDANCE BETWEEN AGE AT FIRST MARRIAGE AND

MARITAL DURATION

The findings of the present study show that mean fecundability vary with the both

age at marriage and marital duration. We want to verify now whether both the

relationships are real or whether one is real and the other one artificial of the

association between age at marriage and the marriage duration. Age at marriage is

strongly associated with the marriage duration as shown by a value of Pearson’s

value contingency coefficient calculated on the assumption that the two variables

are independent. For studying the association, marriage duration is divided into

four groups: 0-4years, 5-9 years and 10 years or more. Age at marriage is also

divided into four groups: less than or equal 15 years, 16-18 years and 19 years or

more.

Mean fecundability after controlling for marriage duration within each age at

marriage group, are shown in Table3.5. The findings in the last column show that

the mean fecundability decreases monotonically from 0.106 to 0.009 with

ascending marriage duration. Mean fecundability declines with ascending

marriage duration within each age at marriage groups. The differentials

presumably, are mainly due to memory laps and truncation biases.

The findings in the last row of the Table-3.5 reflect that mean fecundability

increases monotonically from 0.042 to 0.0705 with advancing age at marriage

except age at marriage group 16-18 years of the women. This result suggests that

fecundability differentials are not only due to memory biases and probably due to

some psychological changes, which differentiates the promptness of first

conception.

211


Table-3.5. Mean fecundabilities for women by marital duration and respondent’s

age at first marriage cohorts.

Duration of

Marriage

(years)

Fecundability

TotalWife’s age at marriage (in completed years)

15 16-18 19 or more

0-4

5-9

10 & above

0.091(382) 0.114(396) 0.130(215)

0.037(310) 0.045(206) 0.042(91)

0.008(39) 0.013(7) 0.009(5)

0.106(993)

0.040(607)

0.009(31)

Total 0.042(731) 0.0709(609) 0.0705(311)

3.10 DIFFERENTIALS OF CONCEPTION WAITS AND FECUNDABILITY

In non-contraception population after entering susceptible state a woman who

engaged in regular intercourse taken on an average several months to conceive.

The women who are more fecund conceive more quickly then the women who are

relatively less fecund. The time, a woman takes to conceive for the first time after

her marriage is called the first conception wait or conception delay or first

conception interval. In the sense Bongaart et al. (1983) pointed out that the waiting

time to conception is synonymous with the length of the menstrual interval. Many

212


researchers including Demographers, Nutrition lists, and epidemiologists, have

shown that the women’s nutritional status, breastfeeding behaviour, socio-

economic, environment and personal characteristics have a great impact in

determining the waiting time to conception and fecundability (Meridith et.al.,).

Bongaarts (1975) mentioned that the biological capacity of women to conceive,

the coital frequency and the timing of sexual unions, the sperm count and the

mobility of the sperms, contraceptive practices and the health status of the couple

are some of the important factors affecting the contraception delay and

fecundability. The effect of various factors on fecundability can be analyzed by a

study of the patterns of distributions of these menstruating intervals and their

correlates. However, true fecundability is virtually complex to measure accurately,

simply because many conceptions are unrecognized as it is not directly observable

event. Hertig et al., (1967) estimated that nearly half of all conceptions never

identified, either because the fertilized ovum fails to implent in the uterous or

because the ovum aborts shortly after implementation. Thus the study based on

waiting time to an identified pregnancy estimate i.e, recognizable fecundability

which is lower than the true fecundability; the difference is a function of the

spontaneous abortion rate in the first few weeks of pregnancy. In this study we

have dealt with the recognizable pregnancies and for this the most reliable and

consistent estimates of the conception interval are measured from the date of first

marriage to the date of first conception.

Fecundability is inversely related to the length of conception wait; the higher the

fecundability, the shorter the conception waits and vice versa. In fact it can be

shown that there is an exact inverse relationship between the conception wait and

fecundability as in a homogeneous population of women with identical

levels of fecundability (Henry,1953; Menken, 1973). However in case of

heterogeneous women the average waiting time to conception is longer that in the

213


homogeneous case as it is seen that the fecundability is not the same for all women

because they have different frequencies of intercourse and different biological

characteristics. Potter and Parker (1983) indicated that heterogeneous women with

highest fecundability conceive rapidly, leaving slower conceivers with decreasing

levels of fecundability in successive months. Bongaarts and Potter argued that the

observed time to conceive conception is on an average about 50 percent longer

than the inverse of fecundability, so they suggested an approximation estimate of

fecundability as instead of .

Considering the methodological differentials the mean fecundability levels are

estimated by three different techniques as-

(a) Estimation of mean fecundability level from the observed duration of

conception delay by geometric distribution.

(b) Estimation of mean fecundability level by fitting Pearson Type I beta

geometric distribution where the parameters are estimated by the method

of moments;

In this chapter an attempt has been made to estimate the fecundability level from

the distribution of conception delays measured from the marriage to first

conception interval. The chapter also analyses the differentials of conception wait

and fecundability and identified the significant predictors affecting fecundability.

We have found from the conditional geometric distribution of the conception delay

(in complete months) that there exists an exact inverse relationship between mean

conception delay and fecundability under the assumption that the women are

homogeneous having constant fecundability. The relationship, however, is not

exact if the fecundability varies from women to women due to some biological and

214


behavioral reasons. In that situation fecundability can be estimated by applying the

method of beta geometric distribution with parameters and b, estimated by the

method of moments. In this study we estimate the average fecundability assuming

that the women under study are homogeneous as well as heterogeneous in nature

with respect to fecundability.

Since marriage to first conception is the most reliable and consistent estimate of

waiting time to conception. We have estimated the average fecundability on the

basis of the waiting time to conception for both the homogeneous and

heterogeneous cases. Table3.1 presents the estimate of fecundability and its

socioeconomic and demographic differentials. The results indicate that the mean

fecundability for considering the women as homogeneous obtained from the

conditional geometric distribution (1/m) is 0.055. The corresponding arithmetic

and harmonic mean fecundability for the heterogeneous women have been found

as 0.070 and 0.055 for method of moments. The results indicate that under the

assumption of homogeneity, the mean fecundability in Bangladesh is much lower

than the estimate obtained by Meredith et al. (1987). Meredith, Menken and

Alauddin obtained an estimate of fecundability for rural Matlab as 0.060 taking

data from Determinants of Natural Fertility Study, conducted under the auspices of

the International Centre for Diarrhoeal Disease Research, Bangladesh. But their

estimate is consistent with the estimated mean fecundability obtained in this study

under the assumption of heterogeneity of women with respect to fecundability.

There are several explanations for the lower estimate of fecundability. One of the

important reasons for the lower estimate of fecundability is that 87.3% of the 1651

women were found to be married who have age limit from 10 to 19 year. i.e. Most

of the respondents are adolescents at the time of first marriage, they are relatively

premature to conceive as compared to their adult counterparts. Another important

reason is that the BDHS study was based on retrospective study, which leads to a

great deal of memory bias of the respondents. And also may be due to the

215


methodological differences. Now we shall discuss the differentials of fecundability

associated with selected socioeconomic, demographic and cultural factors.

3.10.1 Age at first marriage, conception wait and fecundability

Age at first marriage is found to have an inverse relationship with the marriage to first

conception interval. The results in table 4.1 show that the conception interval decreases

as the age at first marriage increases and vice-versa. The women whose age at first

marriage is equal to or less than fifteen years have higher mean conception interval

(23.57 months) and lower fecundabilities (0.042) than the other women. This may happen

because the women who got married in her early age takes time to conceive due to her

menstruation delay and due to her reluctance to be the mother at the very beginning of

her conjugal life and this increases their conception wait but decreases their

fecundabilities. The women whose age at first marriage lies between 16 to 18 years have

lower mean conception interval (14.11 months) and higher fecundabilities (0.071) than

the other women. The women whose age at first marriage lies in the age group 19 years

and above, have very slightly higher mean conception interval (14.18 months) and

fecundabilities (0.071) than the women whose age at first marriage is lies between 16 to

18 years.

3.10.2 Respondent’s (wife’s) education level, conception wait and fecundability

Respondent’s (wife’s) education levels have found to have an inverse relationship with

the marriage to first conception interval but it has a positive association with

fecundability for both homogenous and heterogeneous cases. It has been shown in the

table 4.1 that when the education level of the respondents are increasing, then their mean

fecundability are increasing (except the case of arithmetic mean fecundability for primary

educated women by method of moments) and vice-versa.This is because, when the

women are seen gradually adapting themselves with higher education, then their mean

age at first marriage are also increased and for that they have a tendency to get a child

216


quickly which causes the mean conception delay shorter and fecundability higher.It is

also seen from the table that the women with no education has a greater mean conception

delay(25.17 ) and lower fecundability (0.040) whereas we see the reverse result in the

case of women having primary (21.69 and 0.046) and secondary and above ( 15.70 and

0.064) level of education. We have also seen that the mean fecundability for homogenous

women is equivalent with harmonic mean fecundability for heterogeneous women.

3.10.3 Husband’s education level, conception wait and fecundability

Husband’s education level has been found to have an inverse relationship with the

marriage to first conception interval but it has a positive association with fecundability

for both homogenous and heterogenous case. It has been shown in the table 4.1 that when

the education level of the husband’s are increasing, then their mean fecundability are

increasing and vice-versa. It has been found that the husband’s with no education has a

greater mean conception delay (21.83) and lower fecundability (0.046) than the

husband’s with primary education (17.65 and 0.057) or secondary and higher education

(16.98 and 0.059).This may happen because the husband’s with no education marry the

adolescents who have a lower fecundity (biological capacity to conception) due to

menstruation delay and longer conception delay and hence a lower fecundability.

3.10.4 Respondent’s occupation, conception wait and fecundability

Table-4.1. shows that the women, who do work, have higher conception delay (19.56)

and lower fecundability (0.051) than the women, who do not work (17.97 and 0.056) i.e.

home based. This is due to the fact that the women who do jobs have to spend more time

in her work place than to house comparing to the women who are home based. This will

increase their mean age at first marriage, mean age at first conception as well as decrease

their tendency to be the mother of a child rapidly after marriage. It is also seen that the

women who are doing jobs are found to use more contraceptive methods to linger their

first conception than the women who are home based. These factors are enough to

217


increase the conception interval and hence decrease the fecundability level for the women

who are employed than the women who are home based.

3.10.5 Husband’s occupation, conception wait and fecundability

It has been seen from the Table 4.1.that the husband’s whose occupation is business have

the lower conception interval (16.34) and higher fecundability (0.061) than the other

occupational husband’s. This may happen due to the fact that the husband’s possessing

business related occupation marry the matured women who have a higher fecundity

(biological capacity to conception) and shorter conception delay and hence a higher

fecundability. On the other hand, the husband’s possessing agriculture related occupation

has the higher conception delay (19.59) and lower fecundability (0.051) than the other

occupational husband’s. This is because the husband’s possessing agriculture related

occupation marry the adolescents who have a lower fecundity (biological capacity to

conception) due to menstruation delay and longer conception delay and hence a lower

fecundability.

3.10.6 Type of place of residence, conception wait and fecundability

It has been shown in table 4.1 that the urban women have higher mean conception delay

(18.67) and the lower fecundability (0.054) than rural women. This may happen because

the women living in urban areas get greater scope to do job, for that the contraceptive

practice in urban areas is high (84.9%) than in rural areas (75.2%). In the contrary, the

women living in rural areas do not get more scope to do job due to some social and

religious bindings and they have to cotribute to their family work as well as field work in

harvesting season. That’s why they need children more rapidly to take help in their

family work as well as field work. We also see the parallel results of mean fecundability

218


for homogenous women as well as for heterogeneous women obtained from the harmonic

mean.

3.10.7 Division, conception wait and fecundability

The BDHS 2007 data has been showing that the Dhaka division has a greater mean

conception delay (21.23) and lower fecundability (0.047) than the other division. This is

because the women surveyed in Dhaka division were the adolescent and less educated

and it is known that adolescent has a greater mean conception delay and lower

fecundability than the adults. The Sylhet division has a lower mean conception delay

(15.44) and higher fecundability (0.065) than the other division. This may happen

because the women surveyed in Sylhet division were found with higher mean age at

marriage, religious and holding higher educational status and hence they have low

conception wait and higher fecundability. Moreover, the contraceptive practice in Sylhet

division is the lowest (48.4%) among the other division that causes the higher fertility

rate in Sylhet and consequently the higher fecundability in Sylhet division. We also see

the parallel results of mean fecundability for homogenous women as well as for

heterogeneous women obtained from the harmonic mean.

3.10.8 Religion, conception wait and fecundability

We see from the table 4.1. that the women from the Non-Muslim community have a

shorter conception interval (17.81) and higher fecundability (0.056) than the women from

the Muslim community(18.37 and 0.054).This may happen because Non-Muslim women

got married at higher age than their counter parts and it is known that adults have a lower

mean conception delay and higher fecundability than the adolescent.

3.10.9 Current age of respondent, conception wait and fecundability

The BDHS 2007 data has been showing that current age of respondent has been found to

have a positive association with the marriage to first conception interval but it has an

inverse relationship with fecundability for both homogenous and heterogenous case.i.e.

219


The conception wait is increasing as the current age of respondent is increasing .As a

result the fecundability is decreasing. In this variable for the respondents in the age group

20 to 22 years, we get variance of fecundability is negative this is because a=-16.70

which is less than 2 but According to Anrudh Kumar Jain

Variance = S = (ab(a+b-1))/((a-1) .(a-2)) is not

defined unless a>2.

3.10.10 Use of contraception, conception wait and fecundability

We see from the table 4.1. that the women who do not use contraception have greater

conception delay (23.31) than their counter parts(16.98). This may happen because the

women who do not use contraception are greatly suffering due to adolescent sterility.

3.10.11 Wealth index, conception wait and fecundability

The BDHS 2007 data provides us a consistent result that wealth index has an inverse

relationship with the marriage to first conception interval but it has a positive association

with fecundability for both homogenous and heterogenous cases. The women whose

wealth index is high have shorter conception interval (17.39) and higher fecundability

(0.058) than the women with low (19.52 and 0.051) or middle (18.83 and 0.053) wealth

index.

3.10.12 Husband’s age, conception wait and fecundability Husband’s age has been found to have a positive relationship with the marriage to

first conception interval but it has a negative association with fecundability for

both homogenous and heterogenous case. .i.e. The conception wait is increasing as

the Husband’s age of the respondent is increasing .As a result the fecundability is

decreasingand vice-versa. In this variable for the Husband’s in the age group

18 to 26 years, we get variance of fecundability is negative this is because a=-

35.51

220


which is less than 2 but According to Anrudh Kumar Jain


defined unless a>2.

3.10.13 Marital duration, conception wait and fecundability

Marital duration has a strong positive relationship with the marriage to first conception

interval but it has a negative association with fecundability for both homogenous and

heterogenous case. For instance, the mean length of the first conception interval is found

as 9.41 months for women who have spent 4 years or less as their conjugal life since their

marriage and 31.76 months for those women who have spent more than 4 years as their

conjugal life since their marriage.This indicates that conception interval increases as the

marital duration increases and vice-versa. This also indicate that the women in the lower

conception interval are highly educated, newly got married in her matured age, highly

fertile group of women and less users of family planning. One important social factor is

the practice of permanent sexual abstinence arising from the attainment of the

grandmother status. Since the age at first marriage is very low in Bangladesh, a large

number of women become grandmothers by the time they reach their 35th year, that is,

after 15 to 20 years of marriage. Mothers attaining their grandmother status usually feel

embarrassed to 'compete' with their daughters (or daughters-in-law) and usually practice

permanent sexual abstinence although they are still physiologically capable of

reproduction. The grandmother status thus marks a cultural as opposed to a biological end

of the reproductive period of the woman's life.

In this variable for the Marital duration group 4 years or less, we get variance of

fecundability is negative this is because a= -32.29, which is less than 2 but According to

Anrudh Kumar Jain


defined unless a>2.

3.10.14 Spousal age difference, conception wait and fecundability

221


We see from the table 4.1. that the Spousals whose age difference 9 years and above

have the lowest conception interval (17.15) and higher fecundability (0.059) as compared

to the other. This also indicates that the women in the lower conception interval are

highly educated, newly got married in her matured age, highly fertile group of women

and less users of family planning. On the other hand the Spousals whose age difference 4

years or less has the highest conception interval (22.24) as compared to the other. This

also indicates that the women in the highest conception interval are less educated, got

married in her adolescent age, lower fertile group of women and higher users of family

planning.

3.10.15 Body mass index, conception wait and fecundability

The BDHS 2007 data provides us an important result that the respondents whose Body

mass index is in 18.5-24.9 i.e. weight status is normal have the lower conception interval

(17.39) and higher fecundability (0.057) as compared to the other. This indicates that the

respondents, whose weight is normal, are physically sound enough and capable to

conceive.

3.10.16 Mass media contact, conception wait and fecundability

The BDHS 2007 data has been showing that the women who have a contact with at least

one media, have the lower conception interval (17.99) and higher fecundability (0.056)

compared to their counter parts. This also indicate that the women in the lower

conception interval are highly educated, newly got married in her matured age, highly

fertile group of women and interested to be a mother as early as compared to the other.

222


223


98


Table 3.6 Mean conception wait and fecundability by the fecundability differentials (by geometric distribution and Beta

geometric distribution by the method of moments)

Background

characteristics

Mean Conception wait

(in month),

m

Variance,

No.

of

respondents

Mean FecundabilityFor

homogenous

Women

For heterogenous women

Method of Moments

Mean= Mode= Variance = H.M=

Bangladesh 18.315 606.07 1651

0.055 4.195 55.32 0.070 0.056 0.0011 0.055

Age at first marriage 11 to 15 years

16 to 18 years

19 years and above

23.57

14.11

14.18

962.77

236.68

377.40

731

609

311

0.042

0.071

0.071

4.47

9.16

3.96

78.38

107.04

39.08

0.054

0.079

0.092

0.043

0.071

0.072

0.0006

0.0006

0.0020

0.043

0.071

0.071

Respondent education

98


level Illiterate

Primary Secondary & above

25.17

21.69

15.70

1893.70

631.14

326.76

205

397

1049

0.040

0.046

0.064

2.95

6.91

6.85

47.06

122.33

85.95

0.059

0.053

0.074

0.041

0.046

0.064

0.0011

0.0004

0.0007

0.040

0.046

0.064

Husband education level

Illiterate

Primary

Secondary & above

21.83

17.65

16.98

817.23

649.54

476.77

395

427

829

0.046

0.057

0.059

4.51

3.65

4.64

73.12

44.13

58.24

0.058

0.076

0.074

0.046

0.058

0.060

0.0007

0.0015

0.0011

0.046

0.057

0.059

Respondent occupation Not Working Working

17.97

19.56

565.89

751.17

1293

358

0.056

0.051

4.34

3.87

56.63

53.32

0.071

0.068

0.057

0.052

0.0011

0.0011

0.056

0.051

Husband’s occupation Agriculture BusinessProfessional worker

19.59

16.34

18.50

518.17

749.72

584.20

345

369

579

0.051

0.061

0.054

6.73

3.01

4.49

106.45

30.76

61.07

0.059

0.089

0.068

0.052

0.063

0.055

0.0005

0.0023

0.0010

0.051

0.061

0.054

Table Continued.......

99


Others 18.82 577.29 358 0.053 4.7767.15

0.066 0.054 0.0009 0.053

Type of place of residence Urban

Rural

18.67

18.10

634.76

589.43

617

1034

0.054

0.055

4.16

4.21

55.90

54.96

0.069

0.071

0.054

0.056

0.0011

0.0011

0.054

0.055

Division Barisal Chittagong Dhaka Khulna

Rajshahi

Sylhet

19.63

15.52

21.23

19.89

18.11

15.44

796.04

479.07

655.93

637.32

605.96

514.89

174

348

355

252

305

217

0.051

0.064

0.047

0.050

0.055

0.065

3.70

3.78

5.80

4.87

4.09

3.53

50.25

40.35

97.06

73.07

52.87

36.49

0.069

0.086

0.056

0.062

0.072

0.088

0.069

0.086

0.056

0.062

0.072

0.088

0.0012

0.0017

0.0005

0.0007

0.0012

0.0020

0.051

0.64

0.047

0.050

0.055

0.065

Religion Islam

Others

18.37

17.81

620.47

475.57

1490

161

0.054

0.056

4.12

5.40

54.14

74.03

0.071

0.068

0.055

0.057

0.0011

0.0008

0.054

0.056

Current age of respondent 15 to 19

11.02

15.74

112.93

207.06

608

604

0.091

0.064

90.92

-

901.13

0.092

0.060

0.091

0.063

0.0001

-0.0002

0.091

0.054

Table Continued....

100


years

20 to 22 years

23 to 42 years

31.97 1571.77

439 0.03116.70

5.40

-260.8

136.25

0.038 0.032 0.0003 0.031

Use of contraceptionNot use

Use

23.31

16.98

1343.62

400.76

349

1302

0.043

0.059

3.26

6.19

50.49

82.87

0.061

0.069

0.044

0.060

0.0010

0.0007

0.043

0.059

Wealth index High

Middle Low

17.39

18.83

19.52

504.58

617.05

764.83

831

312

508

0.058

0.053

0.051

4.60

4.38

3.79

58.87

60.30

51.76

0.072

0.068

0.068

0.058

0.054

0.052

0.0010

0.0010

0.0011

0.058

0.053

0.051

Husband’s age18-26 years

27- 32 years

33-70 years

13.81

18.32

23.76

167.37

497.57

1250.11

536

672

443

0.072

0.055

0.042

-35.51

5.52

3.53

-467.5

78.28

57.48

0.071

0.066

0.058

0.072

0.055

0.043

-0.0001

0.0007

0.0009

0.072

0.055

0.042

Marital duration

Table Continued......

101


0-4 years

5-29 years

9.41

31.76

74.50

1108.79

993

658

0.106

0.031

-32.29

16.78

-263.1

485.35

0.109

0.033

0.112

0.032

-0.0003

0.0001

0.112

0.031


≤ 4 years

(5-8) years

9 + years

22.24

18.07

17.15

892.10

434.75

655.02

283

628

370

0.045

0.055

0.059

4.25

6.88

3.56

69.11

100.27

41.01

0.058

0.064

0.080

0.046

0.056

0.060

0.0007

0.0006

0.0033

0.045

0.055

0.059

Body mass index< 18.5

18.5-24.9

24.91-37.79

18.56

17.39

24.82

642.41

544.25

957.60

499

998

132

0.054

0.057

0.040

4.06

4.20

5.14

53.71

52.43

98.61

0.070

0.074

0.050

0.055

0.059

0.041

0.0011

0.0012

0.0005

0.054

0.058

0.040

Mass media contactNo

Yes

19.17

17.99

786.49

537.64

455

1196

0.052

0.056

3.60

4.63

47.10

61.74

0.071

0.070

0.053

0.056

0.0013

0.0010

0.052

0.056

Table Continued....

102


103

Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth

3.11 Trends in mean conception delay

It is observed that Table 4.5 shows the mean conception delay that were taken place at

different years preceding the survey date. The results show that the women who conceive

in recent years have lower conception delay than those who conceived 3 or more years

preceding the survey date. The mean conception delay is found to be 16.66 for women

who conceived within one year preceding the survey while this value is 18.21 for women

who conceived within two years preceding the survey, the value is 18.88 for women who

conceived within three years preceding the survey, the value is 18.39 for women who

conceived within four years preceding the survey and the value is 18.3 for women who

conceived within five years preceding the survey. The mean conception delay in recent

two years indicate that the women who conceived in recent year may be highly educated,

young, newly married with relative higher age at marriage, highly fertile group of women

and less contraceptive users.

.An interesting result can also be seen from table 4.6 and from table 4.7. The table 4.6

shows that the women who got married at age equal to or greater than 19 years have

lower conception delay than the women with other two age groups. This means that the

women who got married in their matured age took less time to conceive than the women

who are adolescents.

It is seen from Table 4.7 that the mean conception delay decreases as the respondent’s

(wife’s) age at first marriage increases for any of the three cases of marital duration

except for the age at marriage group (16-18) years and marital duration (5-9) years. On

the other hand mean conception delay increases as the marital duration increases for any

of the three cases of respondent’s age at first marriage.

115


Table-3.7: Mean conception delay for women by year preceding the survey

Year preceding

the survey

Mean conception Delay( )

Number of respondents

1

2

3

4

5

16.66

18.21

18.88

18.39

18.36

329

304

356

355

307

116


Table-3.8: Mean Conception wait by year preceding the survey and

respondent’s age at first marriage.

Year

precedin

g

the

survey

Respondent’s age at first marriage.

(in completed years) Total

Conception wait

15 years(16-18) years 19 years or

more

1

2

3

4

5

21.51(103)

24.07(149)

23.58(168)

25.28(145)

22.92(166)

14.63(111)

13.43(110)

14.72(131)

14.57(131)

13.15(126)

14.92(52)

12.82(49)

14.61(57)

14.92(75)

13.53(78)

16.66(307)

18.21(355)

18.88(356)

19.39(304)

18.36(329)

Total 23.57(731)14.11(609) 14.18(311)

18.32(1651

)

117


Table-3.9: Mean conception wait by marital duration and respondent’s age at first

marriage

Marital

duration

(years)

Respondent’s age at first marriage

Total


15 16-18 19 or

more

0-4

5-9

10 &

above

11.04(382)

26.78(310)

120.87(39)

8.76(396)

22.21(206)

78.14(7)

7.70(215)

24.03(91)

113.80(5)

9.14(993)

24.82(607)

114.31(31)

Total 23.57(731) 14.11(609) 14.18(311) 18.32(1651)

118


Conception wait by year preceding the survey

15.5

16

16.5

17

17.5

18

18.5

19

19.5

1 2 3 4 5

Year preceding the survey

Co

nc

ep

tio

n w

ait

Conception wait

Figure 3.6 Conception wait by year preceding the survey

The trend line regarding years preceding the survey and mean conception wait

indicates that up to three years preceding the survey conception wait is increasing

after that it is narrowly decreasing.

119


Mean conception wait by years preceding the survey for different ages at first marriage

0

5

10

15

20

25

30

1 2 3 4 5Years preceding the survey

Mea

n c

on

cep

tio

n w

ait

Age at first marriage >= 19 Age at first marriage (16-18) Age at first marriage <= 15

Figure 3.7: Conception wait by years preceding the survey for different age at first marriage

The trend line regarding years preceding the survey, age at first marriage and mean

conception wait indicates that for age at first marriage less or equal 15 years mean

conception wait is slightly increasing up to four years preceding the survey,

thereafter it is decreasing. For the other two groups of age at first marriage, up to 2

years preceding the survey, mean conception wait is sharply decreasing and after that

up to 4 years preceding the survey mean conception wait is narrowly increasing and

thereafter it is sharply decreasing.

120


Mean conception wait by age at first marriage for different marital duration

0

20

40

60

80

100

120

140

0-4 5-9 10 +Marital duration

Me

an

co

nc

ep

tio

n w

ait

Age at first marriage >= 19 Age at first marriage(16-18) Age at first marriage <= 15

Figure 3.8: Fecundability by age at first marriage for different marital duration

The trend line regarding marital duration, age at first marriage and mean

conception wait indicates that upto marital duration 9 years the conception wait is

narrowly increasing thereafter it is sharply increasing . For age at first marriage

(16-18) years the degree of peakness of the conception wait is low than the other

two groups of age at first marriage.

3.12 Trends in Fecundability Level

Table 3.5 shows the mean fecundability that were taken place at different years

preceding the survey date. The results show that the women who conceive in

recent years have higher fecundability than those who conceived 3 or more years

preceding the survey date. The mean fecundability is found to be 0.0600 for

women who conceived within one year preceding the survey while this value is

0.0549 for women who conceived within two years preceding the survey, the value

is 0.0530 for women who conceived within three years preceding the survey, the

121


value is 0.0544 for women who conceived within four years preceding the survey

and the value is 0.0545 for women who conceived within five years preceding the

survey. The fecundability in recent two years indicate that the women who

conceived in recent year may be highly educated, young, newly married with

relative higher age at marriage, highly fertile group of women and less

contraceptive users.

An interesting result can also be seen from table 3.6 and from table 3.7. The table 3.6

shows that the women who got married at age equal to or greater than 19 years have

higher fecundability than the women with other two age groups. This means that the

women who got married in their matured age took less time to conceive than the women

who are adolescents.

It is observed from Table 3.7 that the mean fecundability increases as the respondent’s

(wife’s) age at first marriage increases for any of the three cases of marital duration

except for the age at marriage group (16-18) years and marital duration (5-9) years. On

the other hand mean fecundability decreases as the marital duration increases for any of

the three cases of respondent’s age at first marriage.

Table-3.10 Mean conception delay and fecundability by year preceding the

survey

Year preceding

the survey

Mean conceptionDelay( )


Fecundability

1

2

3

4

5

16.66

18.21

18.88

18.39

18.36

329

304

356

355

307

0.0600

0.0549

0.0530

0.0544

0.0545

122


Table-3.11: Mean fecundability by year preceding the survey and

respondent’s age at first marriage

Year

preceding

the survey

Respondent’s age at first marriage (in

completed years) Total

Fecundability

1516-18 19 or more

1

2

3

4

5

0.046(103)

0.042(149)

0.042(168)

0.040(145)

0.044(166)

0.068(111)

0.074(110)

0.068(131)

0.069(131)

0.076(126)

0.067(52)

0.078(49)

0.068(57)

0.067(75)

0.074(78)

0.060(307)

0.055(355)

0.0530(356)

0.052(304)

0.054(329)

Total 0.042(731) 0.0709(609) 0.0705(311) 0.055(1651)

123


Table-3.12: Mean fecundabilities by

marital duration and respondent’s

age at first marriage Duration

of

Marriage

(years)

Respondent’s age at first marriage (in

completed years) Total

Fecundability

15 16-18 19 or more

0-4

5-9

10 & above

0.091(382)

0.037(310)

0.008(39)

0.114(396)

0.045(206)

0.013(7)

0.130(215)

0.042(91)

0.009(5)

0.106(993)

0.040(607)

0.009(31)

Total 0.042(731) 0.0709(609) 0.0705(311) 0.055(1651)

124


Fecundability by years preceding the survey

0.048

0.05

0.052

0.054

0.056

0.058

0.06

0.062

1 2 3 4 5

Year preceding the survey

Fe

cu

nd

ab

ility

Fecundability

Figure 3.9: Fecundability by years preceding the survey

The trend line regarding years preceding the survey and fecundability indicates that

up to three years preceding the survey fecundability is decreasing then it is narrowly

increasing.

125


Fecundability by years preceding the survey for different age at marriage

0.02

0.03

0.04

0.05

0.06

0.07

0.08

1 2 3 4 5Year preceding the survey

Fe

cu

nd

ab

ilit

y

Age at marriage >= 19 Age at marriage (16-18) Age at marriage <= 15

Figure 3.10: Fecundability by years preceding the survey for different age at first marriage

The trend line regarding years preceding the survey, age at first marriage and

fecundability indicates that for age at first marriage less or equal 15 years

fecundability is slightly decreasing up to four years preceding the

survey .Thereafter it is slightly increasing. For the other two groups of age at first

marriage up to 2 years preceding the survey, fecundability is sharply increasing and

after that up to 4 years preceding the survey fecundability is narrowly decreasing and

thereafter it is sharply increasing.

126


Fecundability by age at first marriage for different marital duration

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0-4 5-9 10 +Marital duration

Fec

un

dab

ilit

y

Age at first marriage >= 19 Age at first marriage (16-18) Age at first marriage <= 15

Figure 311: Fecundability by age at first marriage for different marital duration

The trend line regarding marital duration, age at first marriage and fecundability

indicates that up to marital duration 9 years the fecundability is sharply decreasing

thereafter it is narrowly decreasing for all the three groups of age at first marriage.

127


128

CHAPTERFOUR


CHAPTER FOUR

DIFFERENTIALS OF AGE AT FIRST MARRIAGE, AGE AT FIRST CONCEPTION AND AGE AT FIRST BIRTH

4.1 DIFFERENTIALS OF AGE AT FIRST MARRIAGE The age at marriage is a cultural variable, which may be influenced by a host of

socio-economic and demographic factors. Their relative influence may be expected

to vary from region to region even within the same region. These factors are

differently to produce different marital pattern at different point of time. The social

determinants of age at marriage are very important in Bangladesh. According to

McDonald (1981), “marriage customs, including norms about age at marriage,

develop in each culture in relation to the function that fulfill in the society”.

4.1.1 Respondent education level and age at first marriage

Several studies have shown that women’s education is one of the prime

determinants of age at first marriage in many developing countries (Vagliani,1980;

McCarthy, 1982; Kanitkar and Sinha, 1985 and Bhargava and Saxena, 1985). The

result in the table 5.1 exhibits the lower mean age at first marriage (15.46) for the

women who have no education. The mean age at first marriage for the women

who have primary education and secondary & above is 15.67 years and 16.92

years respectively. The median age at first marriage for the women who have No

education and primary education is same and which is equal to 15 years. The table

4.1 also exhibits that secondary & higher educated women are delayed by about

1.46 years than their illiterate counterparts.

4.1.2. Husband education level and age at first marriage

129


From the table 4.1, it is seen that the mean age at first marriage for the women

whose husbands have No education and primary education are 15.28 and 15.56

years respectively and also their median age at first marriage of the above

mentioned groups are same (15 years). The higher mean and median age at first

marriage for the women whose husbands have Secondary & above level of

education is 17.40 years and 17 years respectively than the others.

4.1.3 Respondent occupation and age at first marriage

Table 4.1 reveals that the mean age at first marriage for the women who do not

work is 16.40 and who work is 16.47. This means that non-working women

married early than the working women. The median age at first marriage for the

both group is same.

4.1.4. Husband’s occupation and age at first marriage

Standard of living and attitudes are likely to differ according to the occupational

status of male. The data shows that the women are more likely to have married

early when their husbands occupation is in others group and their mean age at first

marriage is 15.51.The second earliest married group of women are those, whose

husbands worked in agricultural sector and their mean age at first marriage is

15.53.The data shows that the women are more likely to have married lately when

their husbands are professional worker and their mean and median age at first

marriage 17.25 and 17 years respectively.

4.1.5. Type of place of residence and age at first marriage

The result in the table 4.1 exhibits the lower mean and median age at first

marriage 15.91 and 16 years respectively for the women who are living in rural

areas than their urban counterparts. This may happen that the women lived in

urban areas get greater scope to educate themselves with higher education, which

would increase their mean and median age at first marriage.

130


4.1.6. Division and age at first marriage The data shows that the highest mean age at first marriage among women occurred

in Sylhet division (17 years), followed by Barisal division (16.78

years),Chittagong division(16.77 years), Dhaka (16.28 years), Khulna (16.03

years) and Rajshahi (15.88 years). From this discussion we say that the lowest

mean age at first marriage is experienced in Rajshahi division. The same pattern is

experienced for median age at first marriage.

4.1.7. Religion and age at first marriage Table 4.1 reveals that the mean and median age at first marriage for non-Muslim

are 17.59 and 17 years respectively which are higher than their Muslim (16.29 and

16 years) counterparts.The difference is between two groups is 1.3 years, this

indicates that Non-Muslim women are delayed by about 1.3 years than their

Muslim counterparts.

4.1.8. Current age of respondent and age at first marriage The result in the table 4.1 exhibits that the higher mean and median age at first

marriage are 19.23 and 19 years respectively for the women whose current age lies

in the age range 23 to 42 years than the others.

131


Table-4.1: Differentials of Mean and Median age at first marriage by background characteristics, Bangladesh, 2007

Background characteristics

Mean age at first marriage

Standarddeviation

Median age at first marriage


Bangladesh 16.42 3.04 16 1651

Respondent education level Illiterate Primary Secondary & above

15.46

15.67

16.92

2.39

2.25

3.26

15

15

16

205

397

1049

Husbandeducation level Illiterate

PrimarySecondary

& above

15.28

15.56

17.40

2.22

2.37

2.35

15

15

17

395

427

829

Respondentoccupation Not Workingworking

16.4016.47

2.963.29

1616

1293358

Husband’s occupation Agriculture Business

Professionalworker Others

15.53

16.82

17.25

15.51

2.36

3.03

3.49

2.29

15

16

17

15

345

369

579

358

Type of place of residence Urban

17.26 3.47 17 617

132


. Rural 15.91 2.62 16 1034Division Barisal Chittagong Dhaka Khulna Rajshahi Sylhet

16.78

16.77

16.28

16.03

15.8817.00

3.77

2.60

2.93

2.80

3.093.21

16

17

16

15

1516

174

348

355

252

305217

Religion Islam

Others

16.29

17.59

2.98

3.35

16

17

1490

161Current age of respondent 15 to 19 years

. 20 to 22 years

23 to 42 years

14.42

16.38

19.23

1.43

1.89

3.67

14

16

19

608

604

439

Use of contraceptionNot use Use

16.1816.48

2.723.12

1616

49302

Wealth index High

Middle Low

17.40

15.7115.24

3.44

2.233.09

17

1515

831

312508

Husband’s age 18-26 years

. 27- 32 years 33-70 years

15.15

16.40

17.97

2.00

2.50

3.96

15

16

17

536

672

443

. Marital duration

0-4 years

5-29 years5-30

16.73

15.94

3.05

2.95

16

15

993

658

133

Table continued......


Spousal age difference ≤ 4 years 5-8 years

9 + years

17.4916.44

16.17

3.793.02

2.57

1716

15

283628

740

Body massindex <18.5

18.5-24.9

24.91-37.79

15.78

16.43

18.67

2.41

3.06

3.87

15

16

18

499

998

132

Mass media contactNoYes

15.54

16.71

2.29

3.23

15

16

455

1196

4.1.9. Use of contraception and age at first marriage

It is found that the women who use contraception, their mean and median age at

first marriage are 16.48 and 16 years respectively, which are higher than their non

user counterparts(16.18 and 16 years).

4.1.10. Wealth index and age at first marriage Table 4.1 reveals that the mean and median age at first marriage are lower for the

women, whose Wealth index is low, are 15.24 and 15 years respectively.On the other

hand the mean and median age at first marriage are higher for the women, whose

Wealth index is high, are 17.40 and 17 years respectively. The difference between

mean age at first marriage of low and high Wealth index groups is 2.16 years, this

indicates that high Wealth index women are delayed by about 2.16 years than their

low Wealth index counterparts.

4.1.11. Husband’s age and age at first marriage

134



The result in the table 4.1 exhibits that the higher mean and median age at first

marriage are 17.97 and 17 years respectively for the women whose husband’s age

lies in the age range 33 to 70 years than the others.

4.1.12. Marital duration and age at first marriage Marital duration makes differentials on age at marriage. The higher mean and

median age at first marriage are 16.73 and 16 years respectively for the women

whose marital duration lies in the range 0 to 4 years than the others (15.94 and 15

years), Which suggests that the mean and median age at first marriage are

narrowly increasing trend over the year.

4.1.13. Spousal age difference and age at first marriage The result in the table 4.1 shows that the higher mean and median age at first

marriage are 17.49 and 17 years respectively for the women whose Spousal age

difference lies in the age range 0 to 4 years than the others and lower in the age group

9 years and more.

4.1.14. Body mass index and age at first marriage

Table 5.1 reveals that the mean and median age at first marriage are lower for the

women, whose Body mass index is Below 18.5, are 15.78 and 15 years

respectively. On the other hand the mean and median age at first marriage are

higher for the women, whose Body mass index is 25 and over, are 18.67 and 18

years respectively. This suggests that mean and median ages at first marriage are

increasing as the Body mass index is increasing.

4.1.15. Mass media contact and age at first marriage

Mass media contact has an important effect on mean and median age at first marriage.

The result in the table 4.1 exhibits the higher mean and median age at first

marriage 16.71 and 16 years respectively for the women who have contact with at

least one media than their no-contact counterparts. This tells us that Mass media

contact is helping to increase mean and median age at first marriage.

135


4.2 DIFFERENTIALS OF AGE AT FIRST CONCEPTION AND AGE AT FIRST BIRTH

Age at first conception and Age at first Birth are important in fertility study for any

country, which may be influenced by a host of socio-economic and demographic

factors. Their relative influence may be expected to vary from region to region

even within the same region, religion to religion even within same religion.

Therefore, in this section we have discussed the mean age at first conception and

mean age at first birth by available socio-economic and demographic

characteristics.

4.2.1 Age at first marriage and age at first conception and age at first birth

It is seen from the table 4.2 that as the age at first marriage is increasing mean Age

at first conception and mean age at first birth are also increasing, although the

mean conception wait is found lower while passing through higher group age of at

first marriage (Table 3.1).

4.2.2 Respondent’s education level and age at first conception and age at first birth

Generally education level of respondents is one of the prime factors of marriage

and fertility. Age at first conception as well as age at first birth is the most

important factor for understanding the fertility behavior of Bangladesh. In the

Table 4.2 it has seen that women belong to secondary and higher education level

have higher mean age at first conception (18.23 years) and higher mean age at first

birth(18.98 years) compared to women belong to illiterate and primary education

levels

4.2.3 Husband’s education level and age at first conception and age at first birth

Like wife’s education. The education level of husband is also found as an

important variable in reproduction. Findings from the Table 5.2 indicate that the 136


women whose husband’s are highly educated have the higher mean age at first

conception (18.81 years) than those whose husbands are illiterate (17.03 years)

with a difference of about two years, which is important for fertility in Bangladesh.

4.2.4 Respondent occupation and age at first conception and age at first birth

We have found from the Table 4.2 that the women who are working, their mean

age at first conception and age at first birth (18.09 and 18.85 years) are higher as

compared to their non working counterparts (17.89, 18.65 years), this working

respondents are likely to have more education level and the higher mean age at

first marriage (Table 5.1) consequently the higher mean age at first conception as

well as mean age at first birth. This means that occupation and education are

positively correlated.

4.2.5 Husband’s occupation and age at first conception and age at first birth

The Table 4.2 reveals that the women whose husband’s have no occupation have the

lowest mean age at first conception and age at first birth (17.08 and 17.83 years) as

compared to all other occupational. The women whose husband’s occupation is

agriculture have the second lowest mean age at first conception and age at first birth

(17.16 and 17.91 years) and highest for the Professional worker(18.78 and 19.54 years).

So, we can say that the Professional workers are 1.62 years delayed in conception and

2.03 years delayed in first birth than the women whose husband’s occupation is

Agriculture.

4.2.6 Type of place of residence and age at first conception and age at first birth

It is seen from the Table 4.2 that the women whose living residence is in rural have

the lower mean age at first conception (17.41 years) and age at first birth (18.17

years) as compared to their urban counterparts (18.82 and 19.57 years).Although

about 80% people of Bangladesh are living in rural areas. Therefore to check the

higher growth rate of population the mean age at first conception should be race 137


among the women of rural areas. This can be done to increase the mean age at first

marriage, which can be implemented by increasing the literacy rates of the

females.

4.2.7 Division and age at first conception and age at first birth

Regional differentials of age at first marriage and fertility are pronounced. The

assumption is true in this context. It is seen from the Table 5.2 that the respondents

of Barisal division have the highest mean age at first conception (18.41 years) and

age at first birth (19.16 years), followed by Sylhet division. Whereas the same is

the lowest among the women who live in Rajshahi region (17.38; 18.13

years).Regional differentials regarding mean age at first marriage, conception wait,

mean age at first conception and mean age at first birth can be minimized by

implementation of Government policies uniformly across the regions in

Bangladesh.

4.2.8 Religion and age at first conception and age at first birth The Table 4.2 reveals that the women whose religion is Islam have the lower

mean age at first conception (17.82 years) and mean age at first birth (18.17 years)

as compared to the non-Muslim counterparts (18.82 and 18.57 years). Therefore

emphasize should be given among vast majority of Muslim women to increase age

at first conception for reducing fertility.

4.2.9. Use of contraception and age at first conception and age at first birth

It is seen from the Table 4.2 that the women who do not use contraception have the

higher mean age at first conception and age at first birth (18.12 and 18.88 years) as

compared to their user counterparts. These findings are seemed to be paradoxical

but in the country like ours couples are not interested to use contraception before

138


their first birth. The findings of this variable indicate that couples were use

contraception after their first birth and they are naturally more fertile.

4.2.10. Wealth index and age at first conception and age at first birth

We have found from the Table 4.2 that the women whose Wealth index is low

have the lowest mean age at first conception(16.86 years) and mean age at first

birth (17.61 years) as compared to the others. Probably women belong to relatively

lower group of wealth index are illiterate and having low mean age at first

marriage (16.86 years), and relatively longer first conception wait (25.17 months),

and consequently low mean age at first conception

139

Table 4.2: Mean age at first conception and first birth by background characteristics, Bangladesh, 2007


Categories Mean age at first conception

Minimum Maximum Standard deviation

Mean age at first birth

Standard deviation


Bangladesh 17.94 12.08 39.25 3.39 18.69 3.39 1651Age at first marriage

11 to 15 years

16 to 18 years

19 years and above

15.94

18.06

22.43

12.08

16.08

19.08

39.25

26.00

38.00

2.71

1.50

3.13

16.69

18.81

23.18

2.71

1.50

3.13

731

609

311

Respondent education level

Illiterate

Primary

Secondary & above

17.26

17.77

18.23

12. 08

12.25

12.08

29.83

39.25

38.00

2.76

3.87

3.47

18.02

18.52

18.98

2.76

3.86

3.47

205

397

1049


Illiterate

PrimarySecondary & above

17.03

17.09

18.81

12.08

12.08

12.08

39.25

35.50

38.00

2.99

2.89

3.58

17.85

17.78

19.56

2.99

2.89

3.58

395

427

829

Respondent occupation

Not Working

Working

17.89

18.09

12.08

12.08

39.25

34.67

3.32

3.64

18.65

18.85

3.32

3.64

1293

358

Table continued........



Agriculture

Business

Professional worker

Others

17.16

18.18

18.78

17.08

12.08

12.08

12.08

12.42

29.25

39.25

38.00

34.67

2.74

3.44

3.82

2.75

17.91

18.93

19.54

17.83

2.74

3.44

3.82

2.74

345

369

579

358


Urban

Rural

18.82

17.41

12.08

12.08

38.00

39.25

3.80

3.00

19.57

18.17

3.80

3.00

617

1034

DivisionBarisal

Chittagong

Dhaka

Khulna

Rajshahi

Sylhet

18.41

18.06

18.05

17.68

17.38

18.28

13.08

12.50

12.08

12.42

12.08

13.08

39.25

34.67

37.58

34.17

35.50

30.58

4.12

2.95

3.37

3.26

3.48

3.39

19.16

18.81

18.80

18.43

18.13

19.03

4.12

2.95

3.37

3.26

3.48

3.39

174

348

355

252

305

217Religion Islam

Others

17.82

19.07

12.08

13.17

39.25

38.00

3.35

3.55

18.57

19.82

3.35

3.55

1490161

Use of contraception

Not use Use

18.1217.89

12.2512.08

39.2538.00

3.693.31

18.8818.64

3.693.31

3491302

Table continued.......

141


Wealth index

High

Middle Low

18.85

17.27

16.86

12.08

12.33

12.08

38.00

35.50

39.35

3.67

2.75

2.83

19.60

18.02

17.61

3.66

2.75

2.83

831

312

508Husband’s age 18-26 years

27- 32 years

33-70 years

16.29

17.93

19.94

12.08

12.25

12.08

23.33

34.17

39.25

2.02

2.76

4.38

17.05

18.44

22.55

2.02

2.76

4.38

536

672

443

Marital duration

0-4 years

5-29 years

17.51

18.59

12.08

12.08

31.42

9.25

3.03

3.78

18.26

19.34

3.03

3.78

993

658


≤ 4 years

5-8 years

9 + years

19.35

17.94

17.42

12.58

12.08

12.08

38.00

35.50

36.96

4.03

3.19

3.13

20.10

18.69

18.17

4.03

3.20

3.13

283

628

740

Body mass index

Below 18.5

18.5-24.9

24.91-37.79

17.32

17.91

20.73

12.42

12.08

13.25

34.67

39.25

32.17

2.88

3.35

4.22

18.07

18.66

21.48

2.88

3.35

4.22

499

998

132

Mass media contact

No

Yes

17.24

18.21

12.08

12.08

39.25

38.00

2.97

3.50

17.99

18.96

2.97

3.50

455

1196

Table continued.......

142


143

Chapter Six: Determinant of First Conception Wait A Path Analysis

4.2.11. Husband’s age and age at first conception and age at first birth

We have found from the Table 4.2 that the women whose husband’s age is higher

have higher mean age at first conception (19.94 years) and mean age at first birth

(22.55 years). This implies that husband’s age and wives age are positively

correlated and finally the mean age at first conception is higher among those

respondents whose husband’s age is also higher.

4.2.12. Marital duration and age at first conception and age at first birth

We have got from the Table 4.2 that the women whose marital duration is in the

group 0-4 years have lower mean age at first conception(17.51 years) and age at

first birth (18.26 years) as compared to the women whose marital duration is in the

group 5-29 years. The women whose marital duration are higher probably they are

contraceptive users or may be less fertile than those whose marital duration are

relatively lower.

4.2.13 Spousal age difference and age at first conception and age at first birth

. It is seen from the Table 4.2 that the women whose spousal age difference is (0- 4

years) have the higher mean age at first conception (19.35 years) and age at first

birth (and 20.10 years) as compared to the other women. The results indicate that

spousal age difference is one of the important fertility variables in Bangladesh.

Generally spousal age difference is the highest among Bangladeshi couples (about

9 years). The results reflect that lower spousal age difference is associated with

higher age at first marriage, which is the reason for higher mean age at first

conception and first birth.

4.2.14. Body mass index and age at first conception and age at first birth We have found from the Table 4.2 that the women whose body mass index is

24.91 and over i.e. over weight have the higher mean age at first conception(20.73

years) and age at first birth (21.48 years) as compared to the other women . In

regards to the body mass index of the respondents of this study it has been observed

that the malnourished( less weight) women have lower mean age at first 145


conception than the women who have relatively higher body mass index are found

higher age at conception and age at first birth compared to those who have under

weight. The possible explanation is that respondents having higher social status

and higher education level.

4.2.15. Mass media contact and age at first conception and age at first birth

We have found from the Table 4.2 that the respondents who have contact with at

least one medium have the higher mean age at first conception (18.21 years) and

mean age at first birth (18.96 years) as compared to the other women. The

respondents having contact with any one of the print media or electronics media

have mean age at first conception of 18.21 years which is about 4 months higher

compared to their counterparts who have no mass media contact.

Table 4.3: Also reveals that age at first marriage is highly significantly positively

correlated (0.802) with age at first conception. That means, as the age at first

marriage increases the age at first conception also increases. The significant

Regression coefficient between Age at first marriage and Age at first conception is

146


1.135 which indicates that if Age at first marriage will increase by one year, Age at

first conception will increase by 1.135 years.

Table 4.3:Relationship between ages at first marriage and mean age at first

conception.


Mean age at first conception

Correlation coefficient

Regressioncoefficient

P-value NumberOf respondents

≤13 14.92

0.802 1.135 0.000

23014 15.89 23515 16.85 26616 17.21 24017 18.19 20018 19.12 16919 20.17 10120 21.22 7421 22.34 3922-23 24.04 4424-25 25.46 2626+ 28.76 27

.

To observe the patterns of age at first conception by age at first marriage the trend

values are plotted in figure 4.1. The plotted trend line clearly indicates that with

the increase of age at first marriage the age at first conception consistently

increases.

147


Mean age at first conception at different ages at first marriage

y = 1.1346x + 12.973

R2 = 0.9544

0

5

10

15

20

25

30

35

≤1

3

14

15

16

17

18

19

20

21

22

-23

24

-25

2

6+


Mea

n a

ge

at f

irst

co

nce

pti

on

Mean age at first conception Linear (Mean age at first conception )

Figure-4.1 Mean age at first conception at different ages at first marriage.

148


149

CHAPTERFIVE


CHAPTER FIVE

BIVARIATE ANALYSIS AND COX’S MULTIVARIATE

PROPORTIONAL HAZARD REGRESSION ANALYSIS

5.1. BIVARIATE ANALYSIS

Bivariate analysis is a useful step in studying the relationship between associated

variables. It tells us how important an individual variable is by itself. Moreover, it

helps us to identify those independent variables, which have significant effect on

the conception wait from marriage to first birth. In this analysis, the dependent

variable, the conception wait is categorized into two groups i.e. one is, before and

at mean conception wait (18.32 months) and another one is above 18.32 months. In

this analysis, we use test for the independency of attributes. From the Table-5.1,

it is seen that the independent variables division, husband’s occupation, respondent

education level are found to be significant at 1% level, body mass index, use of

contraception and husband education level are found to be significant at 5% level,

spousal age difference is found to be significant at 10% level, The remaining

independent variables respondent age at first marriage, current age of respondent,

husband’s age and marital duration are found to be highly significant and a brief

description of the Table-5.1 is given below:

It is seen from Table-5.1 that age at first marriage has a strong relationship with

conception wait. age at first marriage has inverse relation with conception wait.

The respondents, whose age at first marriage is 19 years and above (74.9 %)

conceive before and at mean conception wait. Whereas, 58.5% and 73.4%

respondents, whose age at first marriage is in (11 to 15) years and (16 to 18) years

respectively conceive before and at mean conception wait. On the other hand, the

respondents (41.5%) whose age at first marriage is in (11 to 15) years and 26.6%

respondents whose age at first marriage is in (16 to 18) years conceive after the

mean conception wait.

150


Respondents’ education has a significant relation with conception wait.

Respondents, who have secondary and higher education (69.8%), conceive early

than the others. The data shows that the respondents who have primary education

(64.9%) conceive earlier than that of illiterate (61.2%) respondents. Whereas

35.1% illiterate and 38.8% primary educated respondents conceive after the mean

conception wait. Husbands’education is not as strong as women education but has

significant association on conception wait. The respondents whose husbands’

education level are secondary and above (68.9%) conceived on or before mean

conception wait than the respondents whose husbands have primary (68.6%) and

no level of education (62.3%). Because higher educated husbands are likely to

marry educated females as a result early conception occur than the other groups.

Husband’s occupation plays a significant role on conception wait. Husbands who

have better occupation usually marry a woman who is conscious about her life.

The data reveals that 73.7% respondents whose husbands are in business

occupation usually conceive before and at mean conception wait and 36.8%

respondents whose husbands are in agriculture occupation usually conceive after

mean conception wait.

In case of division, we observed that early conception occurred in Sylhet division

i.e.before mean conception wait 74.7% respondents conceive and the next higher

proportion is in Chittagong division (71.6%). Rajshahi (70.2%) Barisal (64.4%),

Khulna (63.1%) and Dhaka (59.7%) are the successive decending order of

proportion of conception before and at mean conception wait. Moreover, we

observed that after mean conception the respondents in Dhaka division have

tendency to conceive later (40.3%) with compare to other divisions namely,

Khulna (36.9%), Barisal (35.6%), Rajshahi (29.8%), Chittagong (28.4%) and

Sylhet (25.3%). This regional differential is observed because of industrialization,

Urbanization and education did not evolve uniformity in all regions.

Current age of respondent has highly significant effect on conception wait. 81.4%

respondents whose age lies between (15 to19) years conceive earlier.65.9% and

49.0% respondents whose age lies between (20 to 22) years and 23 years & above

151


respectively conceive before and at mean conception wait. 51.0% respondents

whose age lies in 23 years & above group conceive later mean conception wait

compared with other groups. Use of contraception plays an important role on

conception wait. 68.6% contraceptive women conceive earlier i.e. before and at

mean conception wait than the others. Whereas 61.6% non contraceptive women

conceive earlier i.e. before and at mean conception wait than the others

contraceptive women conceive earlier i.e. before and at mean conception wait than

the others.31.4% contraceptive women conceive after mean conception wait than

the others.

Husbands’ age has highly significant effect on conception wait. The women whose

husbands’ (73.5%) age lies between 18 to 26 years conceive earlier i.e. before and

at mean conception wait than the other groups. The women whose husbands’

(65.9%) age lies between (26 .1 to 32) years and (61.2%) lies in (above 32 years)

age group are the successive descending order of proportion of conception before

and at mean conception wait. 38.8% respondents lies in the age group above 32

years conceive later than the mean conception wait. Marital duration has highly

significant effect on conception wait. From table it reveals that Marital duration

and conception wait are positively related i.e. conception wait is increasing as

marital duration is lengthen. The respondents (84.9%) whose marital duration (0 to

4) years are likely to conceive earlier i.e. before and at mean conception wait than

the other group. While 59.7% respondents %) whose marital duration more than 4

years are likely to conceive later i.e. after the mean conception wait than the other

group.

Spousal age difference has also significant effect on conception wait. The spouses

whose age difference is more than 9 years and more are likely to conceive earlier

i.e. before and at mean conception wait than the other groups. On the other hand

the spouses (36.7%) whose age difference is less or equal 4 years are likely to

conceive later i.e. after the mean conception wait than the other groups. Body mass

index is another measure of conception wait According to the table, 69.3% normal

weight respondents conceive before and at mean conception wait. While 65.3% of

152


under weight respondents and 57.6% over weight respondents conceive in the

same interval.

Table 5.1: Cross tabulation of conception interval by different background

characteristics of the respondents.


CategoriesPercentage of respondents atConception wait (in month)

Value of Chi-square

P-Value

18.32(mean) >18.32(mean)


11 to 15 years

16 to 19 years

20 years and above

58.5(428)

73.4(447)

74.9(233)

41.5(303)

26.6(162)

25.1(78)

43.77 0.000

Respondent education level

Illiterate

Primary Secondary & above

64.9(133)

61.2(243)

69.8(732)

35.1(72)

38.8(154)

30.2(317)

10.116 0.006


Illiterate

Primary

Secondary & above

62.3(246)

68.6(293)

68.9(569)

37.7(149)

31.4(134)

31.1(260)

5.493 0.064

Respondent occupation

Not Working

Working

67.7(875)

65.1(233)

32.3(418)

34.9(125)0.851 0.356


Agriculture

Business

Professional worker

Others

63.2(218)

73.7(272)

67.4(390)

63.7(228)

36.8(127)

26.3(97)

32.6(189)

36.3(130)

11.608 0.009


UrbanRural

68.9(425)66.1(683)

31.1(192)33.9(351) 1.400 0.237

Division

Barisal

Chittagong

Dhaka

Khulna

Rajshahi

Sylhet

64.4(112)

71.6(249)

59.7(212)

63.1(159)

70.2(214)

74.7(162)

35.6(62)

28.4(99)

40.3(143)

36.9(93)

29.8(91)

25.3(55)

21.216 0.001

153


Religion

Islam

Others

66.9(997)

68.9(111)

33.1(493)

31.1(50)0.272 0.602

Current age of respondent

15 to 19 years

20 to 22 years

23 to 42 years

81.4(495)

65.9(398)

49.0(215)

18.6(113)

34.1(206)

51.0(224)

122.181 0.000

Use of contraception Not use

Use

61.6(215)

68.6(893)

38.4(134)

31.4(409)6.079 0.014

Wealth index HighMiddle Low

68.4(568)66.0(206)

65.7(334)

31.6(263)34.0(106)

34.3(174)

1.173 0.556

Husband’s age 18-26

27- 32

33-70

73.5(394)

65.9(443)

61.2(271)

26.5(142)

34.1(229)

38.8(172)

17.441 0.000

Marital duration (0-4) years

(5-29) years

84.9(843)

40.3(265)

15.1(150)

59.7(393)356.990 0.000


≤ 4

5-8

9 +

63.3(179)

65.4(411)

69.8(518)

36.7(104)

34.6(217)

30.2(222)

6.048 0.109

Body mass index

Below 18.5

18.5-24.9

24.91-37.79

65.3(326)

69.3(692)

57.6(76)

34.7(173)

30.7(306)

42.4(56)8.402 0.015

Mass media contact No

Yes

67.0(305)

67.1(803)

33.0(150)

32.9(393)0.002 0.967

Note: The number in the parentheses is the number of the respondent.

154

Table continued.....


5.2 PROPORTIONAL HAZARD REGRESSION ANALYSIS

This model was first developed by Cox (1972), and has been widely used by

biostatistician, epidemiologists, and demographers (Breslow, 1974, Kalbfleish and

Prentice, 1980; Cox and Okes, 1984) . This model can be used to explain the effect

of covariates on survival times (SAS, 1988, 1992; Colectt, 1994, hess 1994).

Recently this method has been applied in demographic research and work has been

done on marriage dissolution (Menken and others, 1981, Blakrisnan and others,

1987), Contraceptive continuation ( Akhter and Ahmed, 1992) Timing of birth and

birth intervals ( Rodringuez et al.,1984:Rao and Balaksrisnan, 1988, Ahn

Sariff,1993). Some demographers and Statistician (see, for example, Rodrignez

and others, 1984, Newman and McCulloc, 1984) have expressed the view that

hazard technique is particularly well suited for determining the risk of having a

birth. The proportional hazard model is used in the study to investigate the

covariate effects on subsequent fertility in urban and rural Bangladesh. This will

enable evaluation of the probability of having the next birth for a mother in view

of her particular circumstances.

Like the standard life table, it is assumed that there is a hazard (or risk) at each

duration t, of the occurrence of the end-point event (birth). The hazard function is

the product of an underlying duration-dependent risk and covariates (z)

expressed as exp . It is assumed that the duration specific rates or risks for a

given individuals characteristics are proportional. This is defined as

=

Where is the hazard of failure for an individual with covariate z at time t.

is the unspecified base line hazard when z = 0, called the reference group.

155


Z is a row vector of covariates, and is a column vector of unknown parameters

to be estimated in the model.

The term exp is the relative hazard function or relative risk associated with

having the characteristics z. Therefore, the hazard function enables one to estimate

the relative risks of other groups in relation to the base line group (reference

group). When there is no covariate present in the model, then exp is unity.

Values greater then unity indicate that the relative risk of having a conception or

having a birth is greater for that group compared with the reference group. The

parameters in the model (H0:

Can be tested by the Wald statistic:

Where, S.E stands for

standard errors. In using a proportional hazards model, it is assumed that the

hazards associated with covariates are proportional.

In bivariate analysis, we have tried to trace out those independent variables who

have a significant effect on the conception interval. Moreover, before going to the

multivariate analysis by Cox’s proportional hazard regression model, we need

some most independent significant variables to run Cox’s proportional hazard

regression smoothly, which can be found by the bivariate analysis of conception

interval with the various characteristics of the respondent through the cross

tabulation. In Cox’s proportional hazard regression analysis the women are

categorized as zero (0) and (1), where 0 stands for the women having no

conception (1035) as yet and 1 stands for the women having at least one

conception (1651) in last five years preceding the survey and during their fifteen

years of marriage. The respondents who did not conceive during their fifteen years

(180 months) of marriage are considered as primarily sterile and we have excluded

them from this analysis. In the bivariate analysis, we have shown that some

independent variables have very significant effect on the marriage to first

conception interval. However, in multivariate analysis, we would try to find out

156


those independent variables that have a simultaneous significant effect on the

marriage to first conception interval. In such situation, when the independent

variable is a time dependent event, the Proportional hazard model introduced by

Cox in 1972 is the best method to analyze the situation. As fecundability is a time

dependent event, we have tried to find the significant factors affecting on

fecundability through Cox’s regression model.

To fit the model, we have considered 11 socio-economic and demographic

variables, which are found to have significant effect on the conception interval in

bivariate analysis. The variable Husbands age is ignored from the model in order

to avoid the multicollinearity problem. We have found 8 variables having

significant effect on marriage to first conception interval shown in table 4.6.

Statistically significant variables in accordance with their importance are age at

first marriage, spousal age difference, marital duration, use of contraception,

husband’s occupation, division, current age of respondent and body mass index

(missing= 46). The remaining two variables i.e. respondent education level and

husband’s education level do not have any significant effect on marriage to first

conception interval.

It has been observed from the Cox’s Proportional Hazard regression analysis that

age at first marriage is an important factor, which has a great influence on

marriage to first conception interval. It is also seen that women with early age at

marriage are to wait for a longer period of time to be pregnant for the first time

than their higher age at marriage counterparts. This clearly indicates that age at

first marriage has negative influence on marriage to first conception interval and

positive with fecundability. The odds ratio indicates that the women got married at

ages between 16 to18 years are 1.65 times likely to conceive compared to those

who got married at ages less than or equal to 15 years. Again, the odds ratio for

women who got married at ages 19 years and above is 2.35, which indicates that

this group of women are 2.35 times likely to conceive compared to those who got

married at ages less than or equal to 15 years. This is because the early married

women are primarily sterile due to in adolescent sterility and adopting family

157


planning technique or they are reluctant to conceive due to their early age and the

late marrying women are interested to conceive as they lost time due to higher

education.

The Spousal age difference is found to have a statistically significant negative

influence on marriage to first conception interval. The Spousal whose age

difference is 4.1 to 8 years are found to have 1.29 times likely to conceive than the

Spousal whose age difference is less than or equal to 4 years. Again, the Spousal

whose age difference is 9 years and above are found to have 1.62 times likely to

conceive than the Spousal whose age difference is less than or equal to 4 years.

Marital duration is found to have a statistically significant positive influence on

marriage to first conception interval. The women whose marital duration is 5 years

or more are found to have 20 percent lower risk to conceive than the women

whose marital duration is less or equal to 4 years. Use of contraception is found to

have a statistically significant negative influence on marriage to first conception

interval. It is evident from Table 4.6 that the odds ratio for women who Use

contraception is 1.45, which indicates that those women are 1.45 times likely to

conceive, compared to those who do not use contraception. There is a significant

negative effect of husband’s occupation on marriage to first conception interval in

Bangladesh. In the case of Business and others professional husband’s, this factor

is not significant at all. Table 4.6 reveals that the Husbands who are farmer pose a

higher risk of faster conception compared with other Husbands in Bangladesh. The

husbands who are Professional worker pose 16 percent less likely to have a slower

first conception compared with the Husbands who are farmer in Bangladesh.

There is a significant positive regional effect on marriage to first conception

interval in Bangladesh. In Dhaka, Khulna and Rajshahi division, this factor is not

significant at all. Table 4.6 reveals that the respondent who are living in

Chittagong & Sylhet division are found to have 1.56 times and 1.64 times higher

risk of faster conception than the respondent who are living in Barisal division

respectively. Current age of respondent is found to have a statistically significant

158


positive influence on marriage to first conception interval. The respondents who

are in the age group 20 to 22 years and 23 years to above are found to have 19

percent and 63 percent less likely to have a first conception than the respondents

who are in the age group 15 to 19 years respectively. Body mass index is also

found to have a statistically significant positive influence on marriage to first

conception interval in Bangladesh. For the respondents whose body mass index

lies in the group 18.5 to 24.9, this factor is not significant at all. The respondents

whose body mass index lies in the group 24.91 years to higher are found to have

41 percent lower probability to conceive than the respondents who are in the group

lower than 18.50 in Bangladesh.

159


Table 5.2: Cox’s Proportional Hazard Regression of conception waits by different

characteristics of respondents


Coefficient ()

SE P-Value Odds Ratio

Age at first marriage11 to 15 years

16 to 18 years

19 years & above

- -

0.50

0.86

- -

0.064

0.097

0.000 -

0.000

0.000

-1.00

1.65

2.35


≤ 4 years

(5-8) years

9 + years

- -

0.25

0.38

- -

0.078

0.119

0.000 -

0.001

0.000

-1.00

1.29

1.62

Marital duration(0-4) years

(5-29) years

- -

-0.59

- -

0.064

0.000 -

0.000

-1.00

.80

Use of contraceptionNot use Use

- -0.894

- -0.068

0.000 -0.000

-1.001.45

Husband education level IlliteratePrimarySecondary & above

- -0.029

0.041

- -0.074

0.076

0.571 -0.692

0.594

-1.001.030

1.960

Husband’s occupation Agriculture Business

Professional worker Others

- -

-0.001

-0.175

-0.037

- -

0.080

0.074

0.077

0.026 -

0.988

0.018

0.630

-1.00

0.999

0.840

0.764

160

T continued........



Division Barisal Chittagong Dhaka Khulna

Rajshahi Sylhet

- -

0.444

0.073

0.004

0.128

0.493

- -

0.095

0.094

0.100

0.096

0.107

0.000 -

0.000

0.435

0.965

0.185

0.000

-1.00

1.558

1.076

1.004

1.136

1.637

Current age of respondent 15 to 19 years

20 to 22 years

23 to 42 years

- -

-0.216

-0.989

- -

0.070

0.107

0.000 -

0.002

0.000

-1.00

0.806

0.372

Body mass indexBelow 18.5

18.5-24.9

24.91-37.79

-

-0.057

-053

-

0.098

0.093

0.009

0.301

0.000

-

0.945

0.587

161


162

CHAPTERSIX


CHAFTER SIX

DETERMINANT OF FIRST CONCEPTION WAITA PATH ANALYSIS

6.1 INTRODUCTION

The time, a woman takes to conceive for the first time after her marriage is called

the first conception wait or conception delay. A conception delay is defined as the

exposure months preceding, but not including, the month of conception, whereas

the conception wait or the time required to conceive includes that month as well

(Potter and Parker, 1964). To avoid ambiguity between a “conception delay” and

“waiting time of conception” the present analysis has considered the conception

wait. Thus a woman takes several months to conceive after entering the susceptible

state. A woman is assumed have become biologically mature when her menstrual

cycle becomes ovulatory and she continues to ovulate regularly (Pathak, 1978).

She may enter the susceptible state by marriage or resumption of menses after a

live birth while living with her partners. It has been observed that for a

homogeneous group of women, the reciprocal of mean waiting time for first

conception gives the arithmetic mean of fecundability, whereas for a

heterogeneous group of women, the reciprocal of the mean waiting time for the

first conception gives the harmonic mean of fecundability (Henry, 1972; James,

1963; Sheps, 1964). Since fecundability is inversely related to conception wait,

estimating fecundability, conception wait may be calculated either from marriage

to first birth or to first conception or from subsequent birth intervals. In traditional

populations like Bangladesh where marriage is almost universal and early marriage

is vogue, all the female are not exposed to the risk of conception immediately after

marriage either mainly due to biosocial immaturity or due temporary separation

from their spouses.Among females, almost 95 percent of marriages take place

before the end of their second decade of life. This densely populated country is

also characterized by a high population growth rate, high nuptiality and low age at

marriage (Huq and Cleland, 1990; Islam and Islam, 1993). Nonetheless, there has

163


been a clear rising trend towards higher age at marriage over time (Islam and

Islam, 1993; Aziz and Malony, 1985).

Marriage to first conception wait is an important factor of population growth and

development. As such, it is useful information for public policymakers interested

in understanding the connection between Bangladeshi Population and development

problems and in monitoring the progress made in combating those problems. Since

marriage to first conception wait is more important for policymaker and planners,

so we need to understand the impact of various socio-demographic variables on

marriage to first conception wait and also understand the nature of the effect of

these variables on marriage to first conception wait. To understand this Path

Analysis is one such technique of showing causal linkage among interrelated

variables. This technique has been used by much social research quantifying and

interpreting causal linear models.

Path analysis helps in estimating the magnitude of the linkage between

interrelated variables and provides information about the understanding causal

process. This technique explores a chain of relationships among the variables by

using standardized regression coefficient of a set of regression equations.

However, the fundamental task here is to construct a path diagram in which

directions (indicated by arrowheads) should be causally meaningful. This study

employs a recursive path model relating to marriage to first conception wait and

some of its determinants.

6.2 HISTORICAL BACKGROUND

164


Path analysis was developed as a method of decomposing correlations into

different pieces for interpretation of effects (e.g., how does parental education

influence children's income 40 years later?). Path analysis is closely related to

multiple regression; you might say that regression is a special case of path

analysis. Some people call this stuff (path analysis and related techniques) "causal

modeling." The reason for this name is that the techniques allow us to test

theoretical propositions about cause and effect without manipulating variables.

However, the "causal" in "causal modeling" refers to an assumption of the model

rather than a property of the output or consequence of the technique. That is,

people assume some variables are causally related, and test propositions about

them using the techniques. If the propositions are supported, it does not prove that

the causal assumptions are correct.

6.3 ANALYTICAL METHOD

The technique of analysis that employed in this study to examine the effects of the

selected factors on conception wait is Path analysis. Path analysis is a technique of

showing causal linkages among interrelated variables. This technique was

originally formulated by Wright (1921, 1934, and 1960). Recently, its application

is gaining popularity in demography and other social research for quantifying and

interpreting causal linear models (Duncan, 1966; Kendall and Driver, 1973;

Balakrishnan et al., 1980;Ahmed, 1981; Miller and Stocks, 1985; McDonald,

1977). Path analysis helps in estimating the magnitude of the linkages between

interrelated variables and provides information about the under lying causal

processes. The technique explores a chain of relationships among the variables by

using standardized regression coefficient of a set of regression equations. It is a

statistical technique that may conveniently disentangle the specific mechanisms of

the socio-economic factors affecting conception wait by taking into consideration

the variables involved in the analytical system. It also provides a theoretical model

specified, as a system of simultaneous and the equations are linear, additive and

usually recursive (Boyle, 1970:461-480).

165


Moreover, path analysis provides mechanism of decomposing the total effect of an

independent variable into its component parts: indirect and direct effects. A total

effect is that part of the total association or zero order correlation between two

variables in a causal model which is not due to common causes, i.e. due to

correlation between their causes or unanalyzed correlation among the

predetermined variables (Alwin and Hauser, 1975: 38-39). A total effect shows

how much change in dependent variable is brought about a given change in the

independent variable. The indirect effect shows how much of the total effect of an

independent variable on a dependent variable is mediated via a specific intervening

variables in a causal sequences but which remains after the effect of intervening

variables has been removed(Call and Otto, 1977: 72-73).

However, the fundamental task here is to construct a path diagram in which

directions (indicated by arrowheads) should be causally meaningful. Nine

variables have been employed in the path analysis. The name of the variables, their

abbreviations and measurements are given in Table 5.1. The dependent variable is

the Marriage to first Conception wait. The independent variables used in this study

are ‘Division’, ‘Age at first marriage’,’ Body mass index’, ‘Use of contraception’,

‘Current age of respondent’, ‘Marital duration’, ‘Husband education ‘, and

‘Spousal age difference’.

Path coefficients are the standardized regression coefficient in a system of linear

regression equations, usually denoted by Pji. Where j denotes the dependent variable and

i denote to the variables, whose direct effect on the variables is measured. In other words,

Pjis are path coefficient gives the proportion of the standard deviation of the dependent

variable for which the independent variable is directly responsible. In other words, Pji = ,

where and denote the standard deviation of the dependent and independent variables

respectively (See, Chandrasekaran and Harmalin, 1975). Empirical evidence suggests that

166


Socio-demographic determinants have an direct effect and some of them have indirect effect

on marriage to first Conception wait. It is assumed that this variable is the effect of

all other variables in this analysis and that it dose not affect any of the preceding

variables.

Table 6.1: Variables and their Measurements used in the Path Analysis Variables Abbreviation CategoriesX1= Division DIV 1 = Barisal

2=Chittagong 3=Dhaka 4=Khulna 5=Rajshahi 6=Sylhet

X2= Age at first marriage AFM 1=11 to 15 2=16 to 19 years 3= 20 years and abov

X3= Body mass index BMI 1=Below 18.5 2=18.5-24.9 3=24.91-37.7

X4= Use of contraception UOC 0=Not use 1=Use

X5 = Current age of respondent CA 0=15 to 19 years 1=20 to 22 years 2=23 to 42 years

X6 = Marital duration MD 1=(0-4 )Years 2=(5-29)Years

167


X7 = Husband education HEDU 0=Illiterate 1=Primary 2=Secondary & above

X8 = Spousal age difference SAD 0=Less than or equal 4 years 1=(5-8) years 2=(9-12) years 3=(13-49) years

X9= Marriage to first Conception wait CW In completed Variate

168


Figure 6.1: A causal Model for Factors affecting timing of first conception

Div X1

AFM X2

BMIX3

UOCX4

MDX6

CAX5

CW X9

HEDUX7

SAD X8

151


Moreover, under additional assumptions of linearity and additivity, the system of

equations for the model considered may be written as below, which will provide

information on the estimation of influences of variables:

X8 = P87X7 +P8ses

X3 = P31X1 + P3tet

X2 = P23X3 + P21X1 +P2ueu

X6 = P63X3 + P62X2 +P61X1 + P6vev

X4 = P46X6 + P43X3 +P42X2 +P41X1+ P4wew

X5 = P56X6 + P54X4 + P53X3 + P52X2 + P51X1 +P5yey

X9 = P98X8 +P97X7 + P96X8 + P95X5 + P94X4 + P93X3 + P92X2 + P91X1 + P9zez

Where Pji are path coefficients from Xi to Xj in the model and es, et, eu ,ev, ew, ey

and ez are random disturbance terms. All disturbance terms are mutually

independent and are independent of their corresponding explanatory variables. In

the equations, each variable is in the standard form, Xj (j= 2,3,4,5,6,8,9)variables

are considered endogenous and the Xi (i= 1, 7) variables are exogenous.

The expansion of the correlation into path coefficients illustrates the ‘Basic

theorem of Path analysis’ which may be written generally as

Rij =

Where, j and I are two variables and q runs over all variables that have direct paths

to variable j (e.g., Duncun, 1966; Shin, 1977). Path analysis also used to explain

the indirect effect inherent in the model.The correlation between two variables can

be decomposed into a direct effect, indirect effect and the joint effects shared with

other variables in the system (Duncun, 1971).

6.4 RESULTS

152


The computed standardized regression coefficients known as path coefficients are shown

in figure 5.2. Out of 24 hypothesized path coefficients, 14 are found to be statistically

significant. The direct and indirect effects of each of the selected predetermined variable

are presented in Table- 5.3 The model shows that division(X1) has a non-significant

direct negative effect on first conception wait but it has indirect effect on conception wait

through age at first marriage (0.011), Use of contraception (0.013) has found positive

effect and indirect effect through body mass index (-0.001), marital duration (-.004) and

current age of respondent (-0.019) have found negative effect. That means division

through age at first marriage and use of contraception has positive effect on first

conception wait.

153


0.661 -0.473

-0.024 -0.157 0.034 0.532 -0.037 -0.017

-0.112 -0.115

-0.002

-0.023 0.023 -0.038 0.505 0.159

-0.047

-0.037 0.054 0.022

0.097 0.018

0.028

0.054 0.018

Figure-6.2 Path diagram of marriage to first conception wait and predetermined variables

CW X9

Div X1

AFM X2

BMI X3

CAX5

MD X6

HEDUX7

SAD

X8

UOC X4

AFM X2

154


Table 6.2. Path coefficients for specified combinations of variables of Marriage to first conception wait:

Dependent variable

Independent variable X1 X2 X3 X4 X5 X6 X7 X8

X2 -0.024 0.203****X3 -0.038X4 -0.112**** 0.034 0.054** 0.023X5 -0.037** 0.661**** 0.028* -0.017 0.505****X6 -0.023 -0.157**** 0.097****X8 0.018 X9 -0.002 -0.473**** 0.018 -0.115**** 0.532**** 0.159**** -0.022 -0.047**

**** Significant at 0.1%

*** Significant at 1%

** Significant at 5%

*Significant at 10%

155


Age at first marriage(X2) has a significant direct negative effect on first conception

wait. Its total effect on first conception wait is -0.15. Where -0.473 is the direct

negative effect but its indirect effect through Use of contraception (-0.004),

Marital duration (-0.025) has found negative effect and Current age of respondent

(0.352) has found positive effect. Body mass index (X3) has a non-significant

direct positive effect on first conception wait but its indirect effect through Age at

first marriage(-0.096), Use of contraception(-0.006) has found negative effect and

Current age of respondent(0.0149), Marital duration(0.159) has positive effect on

first conception wait. Use of contraception(X4) contributes to the variation on

conception wait. Its total effect on first conception wait is -0.124, of which

92.74% is the direct negative effect but its indirect effect through Current age of

respondent (-0.009) has also negative effect on first conception wait.

Current age of respondent (X5) has a significant effect on first conception wait.

From table it is found that its direct positive effect on first conception wait is

0.532. Marital duration(X6) is found to have significant positive effect on first

conception wait. Its total effect on first conception wait is 0.425, of which 37.41%

is the direct positive effect but its indirect effect through Use of contraception (-

0.003) has negative effect and Current age of respondent (0.269) has positive

effect on first conception wait. Husband education(X7) has a total effect on first

conception wait is- 0.023. Its significant direct negative effect on first conception

wait is -0.022 and its indirect effect through Spousal age difference (-0.001) has

also found negative effect on first conception wait. Spousal age difference(X8) has

a significant direct negative effect (-0.047) on first conception wait.

It is observed from Table 5.4 that the total effects of X1 transmitted via X2,X3,X4,

X5, X6 ,via second order interactions X2X5, X3X2 and via third order interactions

X2X6X5, X3X6X5 on X9. The total effects of X3 transmitted via X2, X4, X5, X6 and

via second order interactions X2X4, X2X5, X2X6, X3X2 and X4X5 on X9. Ideally, a

zero-order correlation coefficient between first conception wait and any

156


predetermined variable should be small difference as the total effect of that

predetermined variable on marriage to first conception wait.

157


Table-6.3: Analysis of the Effects of the variables used in the Path model for explaining Conception wait for women in Bangladesh.

Dependent

variable

Predetermined Variable

Direct effect

Indirect effects throughTotal effect

Zero

order Correlation X2 X3 X4 X5 X6 X8

X2X4 X2X5 X2X6 X3X2 X4X5 X2X6X5 X3X6X5

X9

X1 -.002 .0113 -.001 .013 -.019 -.004 -.008 .004 .001 -.001 -.006 -.017

X2 -.473 -.004 .352 -.025 -.15 -.168

X3 .097 -.096 -.006 .0149 .0154 -.001 .072 -.005 -.005 -.001 .086 .032

X4 -.115 -.009 -.124 -.105

X5 .532 .532 .325

X6 .159 -.003 .269 .425 .445

X7 -.022 -.001 - .023 -.075

X8 -.047 -.047 -.053

158


Table 6.4. Zero-order Correlation Coefficients among the selected socio-economic, cultural and Demographic Variables

Variables X1 X2 X3 X4 X5 X6 X7 X8 X9X1 1.00X2 -.032 1.00X3 -.038 .204** 1.00X4 -.115** .046 .067** 1.00X5 -.067** .597** .196** .034 1.00X6 -.019 -.141** .066** .025 .416** 1.00X7 -.123** .313** .199** .193** .169** -.116** 1.00X8 .039 -.115** .008 -.094** -.130** -.033 .018 1.00X9 -.017 -.168** .032 -.105** .325** .445** -.075** -.053(*) 1.00

** Correlation is significant at the 0.01 level (2-tailed).

* Correlation is significant at the 0.05 level (2-tailed

159


160


CHAPTERSEVEN

161

CHAPTER SEVENSUMMARY, CONCLUSION AND POLICY IMPLICATIONS

7.1 INTRODUCTION

For studying the differentials of marriage to first conception wait and fecundability

among women, which are the important factors of fertility, the data are extracted

from the 2007 Bangladesh Demographic and Health Survey (BDHS). According

to the objectives of the study 1651 ever-married women have been considered out

of 10996 respondents, who have at least one live birth preceding the last five

years of the survey, in order to overcome memory lapse of the respondents.

Marriage to first conception wait plays an important role in population dynamics

and has significant impact on fertility. It also involves biological characteristics,

which are related to the social, economic and in many cases religious aspects.

Despite the tremendous importance of fecundability on fertility, few studies have

been conducted in this area in Bangladesh. In this study, however, different

techniques are employed to intensively investigate the patterns and socio-

economic and demographic differentials of marriage to first conception wait,

fecundability and age at first motherhood. In order to understand the significance

of rich factors for marriage to first conception wait Multivariate Cox’s

Proportional Hazard Regression analysis is employed to estimate the levels, trends

and the differentials of fecundability, Pearson Type I geometric distribution is

fitted and the parameters are estimated by the method of moments. To observe the

differentials of conception wait and fecundability we focused on the differentials

behavior of age at first marriage, age at first conception and age at first birth by

available background characteristics. Path analytical approach has also been used

to examine direct, indirect and joint influences of selected socio-economic and

demographic factors on marriage to first conception wait among women in

Bangladesh.

7.2 SUMMARY OF THE MAJOR FINDINGS


Since age at marriage is the indicator of fertility, the age at first marriage of the

respondents are studied. The mean age at first marriage is found only 16.42 ± 3.04.

The mean age at first marriage is unexpectedly lower in Bangladesh although the

legal age at first marriage is 18 years. Still the mean age at first marriage is about

one and half years lower than the legal age at first marriage for females in

Bangladesh. The age at first birth signals the length of reproductive span. In this

study the age at first motherhood among Bangladeshi women is 18.69 years,

which is closest to legal age at marriage. In South Korea the age at first

motherhood is 25 years in Srilanka 18.5 years and 21.5 years in Ruanda. The

increasing age at marriage in the countries mentioned above was mainly due to

expansion of education and rapid economic growth.

The estimated conception wait (18.32 months) is longer in Bangladesh. This

longer mean conception delay is mainly due to the lower age at first marriage,

health status, malnutrition and finally due to adolescent sub-fecundity. The level of

fecundability is found only 0.055 while estimating by using geometric distribution,

which is the lowest among Bangladeshi women.

For testing the heterogeneity of the sampled women regarding their first

conception, the beta geometric distribution with parameters and b by the method

of moments also has been fitted. According to this mixture distribution, the

harmonic mean and arithmetic mean fecundabilities are found to be only 0.055

and 0.070 respectively by the method of moments. Our study has revealed that

fecundability widely varies with age at first marriage. As the age at first marriage

increases, the mean fecundability also increases. It has also been observed that the

level of fecundability for Bangladeshi women lies between 0.055 and 0.070. The

estimated fecundability among rural Bangladeshi women obtained by the

Meridith, Menken and Alauddin (1987) lies between 0.046 to 0.066. They

obtained an estimate of fecundability for rural Matlab women as 0.060, which

close to our findings.

163


Singh and Saixna (1969) obtained an estimate of fecundability as 0.050 for some

southern women of India. The fecundability estimated from the Banarasi survey

(Singh, 1969) was found 0.269. Majumdar and Sheps (1970) found an estimate of

fecundability as 0.25 for Hutterite women and 0.26 and 0.14 from Princeton

fertility survey data. Jain (1969) found fecundability as 0.163 for Taiwanese

women. Bongaart’s (1975) obtained the average fecundability of 0.37 for five

historical populations ( Crulai, Tourouvre and Perche, Geneva, Tunis and Canada).

Singh and Bhadury (1972) obtained an estimate of fecundability by fitting a

theoretical continuous distribution as 0.22 for Hutterite data by the method of

moments. He further obtained an estimate of fecundability as 0.32 from Princeton

fertility survey data. As compared to other developed and developing countries, it

is seen that the fecundability in Bangladesh is remarkably low. The low average of

fecundability in Bangladesh may be attributed to the different biological,

behavioral and social factors. Among them the lower mean age at marriage,

adolescent infertility, temporary migration, memory lapse, misreporting of age at

first marriage and first pregnancy termination and early spontaneous abortion are

important. For the present data, the mean age at marriage is found to be 16.42

years, which lies in adolescent age interval (10- 19). This mean age at first

marriage is also closest to the mean age at menarche. We have also found 205

women out of 1651 women with no education having mean age at marriage as

15.67 years and consequently they have higher conception interval and lower

fecundability. Moreover, in our analysis, we have found the urban women have

higher mean conception delay (18.67) and the lower fecundability (0.054) than

rural women. This is because the women living in urban areas get greater scope to

do job and higher rate of contraceptive users, the contraceptive prevalence rate

(CPR) in urban areas is high (84.9%) than in rural areas (75.2%).

We have also observed that the women residing in Dhaka division has a greater

mean conception delay (21.23) and lower fecundability (0.047) than the other

region of the country. This is because a considerable number of women are

164


residing in the vast slum areas most of them are illiterate and their mean age at

marriage is lower and lower fecundability. The women of Sylhet division have a

lower mean conception delay (15.44) and higher fecundability (0.065) than the

other divisions. This may happen because the women of Sylhet division were

found with higher mean age at marriage (17.00 years), religious and holding

higher educational status and hence they have low conception wait resulting higher

fecundability.

We have also seen that age at first marriage has negative relation with conception

wait and positive relation with fecundability. It is seen from the study that

conception wait is decreases and fecundability increases with the increasing age at

first marriage whatever be the marital duration. Again the fecundability decreases

with the increasing marital duration whatever be the ages at first marriage. This

indicates that the more the age at first marriage the higher the fecundability and

the less the conception wait and vice-versa. Moreover, the more the marital

duration the less the fecundability and higher the conception wait and vice-versa.

We have found that the significant correlation coefficient between Age at first

marriage and conception wait (-0.154), which discloses that age at first marriage is

negatively correlated with first conception wait. We also get the significant

regression coefficient between age at first marriage and conception wait is -1.252

which indicates that if age at first marriage will increase by one year, conception

wait will decrease by -1.252 months.

The study also reveals that age at first marriage is highly significantly positively

correlated (0.802) with age at first conception. This implies that as the age at first

marriage increases the age at first conception also increases. The significant

regression coefficient between age at first marriage and age at first conception is

1.135 which indicates that if age at first marriage will increase by one year, age at

first conception will increase by 1.135 years. We have found from the differentials

of age at first marriage, the lower mean age at first marriage (15.46) for the

165


women who have no education than the others. We have also got that secondary

and higher educated women are delayed by about 1.94 years than their illiterate

counterparts.

The higher mean and median age at first marriage for the women whose husbands

have Secondary and above level of education is 17.40 years and 17 years

respectively than the others. From respondents occupation it is clear that non-

working women got married early than the working women. The data shows that

the women are more likely to have married lately when their husbands are

professional worker and their mean and median age at first marriage are 17.25 and

17 years respectively. The data exhibits the lower mean and median age at first

marriage 15.91 and 16 years respectively for the women who are living in rural

areas than their urban counterparts.

The findings show that the highest mean age at first marriage among women

occurred in Sylhet division (17 years) and the lowest mean age at first marriage is

experienced in Rajshahi division. The result indicates that Non-Muslim women

are delayed by about 1.3 years than their Muslim counterparts. It is found that the

women who use contraception, their mean and median age at first marriage are

16.48 and 16 years respectively, which are higher than their non user

counterparts(16.18 and 16 years). We have also found that whose wealth index is

high women are delayed in marriage by about 2.16 years than their low wealth

index counterparts.

It is clear that the mean and median age at first marriage are higher for the

women, whose body mass index is 25 and over, are 18.67 and 18 years

respectively. This suggests that mean and median ages at first marriage are

increasing as the body mass index is increasing. The higher mean and median age

at first marriage 16.71 and 16 years respectively for the women who have contact

with at least one media than their no-contact counterparts. This tells us that mass

media contact is helping to increase mean and median age at first marriage. From

the differentials of age at first conception and age at first birth we have found that 166


as the age at first marriage is increasing mean age at first conception and age at

first birth are also increasing.

The women who are illiterate have the lowest mean age at first conception and age

at first birth (17.26 and 18.02 years) as compared to the others. The women whose

husband’s education level is primary have the lowest mean age at first conception

and age at first birth (17.03 and 17.78 years) as compared to the others. The

women who are working their mean age at first conception and age at first birth

(18.09 and 18.85 years) are higher as compared to the not working counterparts.

We have found that the professional workers are 1.62 years delayed in conception

and 2.03 years delayed in first birth than the women whose husband’s occupation

is agriculture.

The women whose living residence is in rural have the lower mean age at first

conception and age at first birth (17.41 and 18.17 years) as compared to the urban

counterparts (18.82 and 19.57 years. The women whose living residence is in

Barisal division have the highest mean age at first conception and age at first birth

(18.41 and 19.16 years) as compared to the other divisions and the second highest

mean age at first conception and age at first birth (18.28 and 19.03 years) as

compared to the other divisions in Sylhet division. The women whose religion is

Islam have the lower mean age at first conception and age at first birth (17.82 and

18.17 years) as compared to the Non-Muslim counterparts (18.82 and 18.57 years

The women who do not use contraception have the higher mean age at first

conception and age at first birth (18.12 and 18.88 years) as compared to the others.

The women whose wealth index is low have the lowest mean age at first

conception and age at first birth (16.86 and 17.61 years) as compared to the others.

The women whose body mass index is 24.91 and over i.e. over weight have the

higher mean age at first conception and age at first birth (20.73 and 21.48 years) as

compared to the other women. The women who have contact with at least one

media have the higher mean age at first conception and age at first birth (18.21 and

18.96 years) as compared to the other women

167


The multivariate analysis through the Cox’s proportional Hazard Regression

model shows that the respondent age at first marriage, spousal age difference,

marital duration, use of contraception, husband’s occupation, division, current age

of respondent and body mass index are found to have statistically significant

association with the marriage to first conception wait.

Finally we have drawn a path diagram with the variables which are significant in

Cox’s proportional Hazard Regression Analysis. From this path analysis, we have

found that significant effect of the variables on conception wait are age at first

marriage, current age of respondent, husband education, marital duration, spousal

age difference. age at first marriage (X2) has a significant direct negative effect on

first conception wait. Its total effect on first conception wait is -0.15. Where -0.473

is the direct negative effect but its indirect effect through use of contraception (-

0.004), marital duration (-0.025) has found negative effect and current age of

respondent (0.352) has found positive effect. Current age of respondent (X5) has a

significant direct positive effect on first conception wait is 0.532. Marital

duration(X6) is found to have significant positive effect on first conception wait.

Its total effect on first conception wait is 0.425, of which 37.41% is the direct

positive effect but its indirect effect through use of contraception (-0.003) has

negative effect and current age of respondent (0.269) has positive effect on first

conception wait. Husband’s education(X7) has a total effect on first conception

wait is- 0.023. Its significant direct negative effect on first conception wait is -

0.022 and its indirect effect through spousal age difference (-0.001) has also found

negative effect on first conception wait. Spousal age difference(X8) has a

significant direct negative effect (-0.047) on first conception wait.

7.3 CONCLUSION

Based on the findings the study concludes that as the age at first marriage is

increasing the age at first conception and age at first birth are also increasing and

conception wait is decreasing as a result fecundability is increasing. The women

whose husband’s education level is secondary and above their mean age at first

168


marriage is high for that the mean age at first conception and mean age at first

birth are also high and mean conception wait is low as a result mean fecundability

is high. Mean age at first marriage, mean conception wait, mean age at first

conception and mean age at first motherhood are found higher and fecundability

lower among working women. The reason for higher conception wait and lower

fecundability for working women is probably due to the use of contraception

before first conception. Similarly, mean age at first marriage, age at first

conception timing of first parenthood and fecundability are found relatively higher

and conception wait lower among the respondents whose husbands are

professionals.

Rural urban differentials regarding the components of human reproduction

behavior studied in this work are not found pronounced because first conception is

an welcome event both for urban and rural families. The women who are living in

Sylhet division their mean age at first marriage, age at first conception and age at

first birth are high and conception wait is low and fecundability is high. Mean age

at first marriage of Non-Muslim is high for that the mean age at first conception

and mean age at first birth are also high and conception wait is low, as a result

fecundability is high. In regards to current age of the respondents it has been

observed that the respondents whose current age is higher their mean age at first

conception, age at first motherhood and fecundability are higher. One of the

probable reasons for higher fecundability of this cohort is that they want to

conceive as early as possible because of time loss due to late marriage. The

women who use contraception though their age at first marriage is high their age at

first conception and age at first birth are low and conception wait is low as a result

fecundability is high.

Socio-economic status of the respondents, which is considered here as the wealth

index, is found as one of the important predictors in explaining the dynamics of

the reproductive behavior. Based on the findings it can be concluded that

respondents belong to high social status are high mean age at first marriage, age at

169


first conception and relatively higher fecundability. The study reflects that

husband’s age is positively associated with age at first marriage and age at first

conception of the respondents and negatively related with fecundability level. The

variable marital duration is found negatively associated with age at first marriage

and fecundability and positively related with mean age at first conception and

mean age at first motherhood. This result indicates that the mean age at first

marriage is increasing and level of fecundability is also increasing.

As the Spousal age difference is increasing mean age at first marriage, age at first

conception and age at first birth of the women are decreasing but conception wait

is decreasing as a result level of fecundability is also increasing. Late marriage of

husband’s and polygamy may be the main reason for higher level of fecundability

of the higher spousal age difference group. Nutritional status of the respondents

converted here as the body mass index (BMI) is found one of the important

variables in explaining the different segments of the reproductive span analyzed in

the present study. The findings show that age at first marriage and age at first

conception are positively associated with BMI of the concern respondents. But

conception wait of the overweight women is found positively associated with BMI

and negatively associated with fecundability while the scenario have been

observed for the women belong normal BMI. Among the variables considered in

this study mass media contact is found very important. The women who have no

contact with electronics and print media have the lower age at first marriage and

also the lower age at first motherhood and lower fecundability.

7.4 POLICY IMPLICATIONS

The findings of this study may have some policy implications that would help the

planners and policy-makers to take necessary steps in increasing fecundability by

decreasing marriage to first conception wait among women in Bangladesh. The

factors that are found to have significant effect on these vital events may have

important policy implications. Therefore, the following recommendations can be

suggested for fruitful policy implications:170


The study noticed that age at first marriage is an important predictor of marriage to

first conception wait and fecundability. To decrease marriage to first conception

wait and to increase fecundability, age at first conception and age at first

motherhood, age at first marriage should be increased because the adolescents who

have a lower fecundity (lower monthly probability of conception) they also have

lower mean age at first marriage and higher conception wait, which are related

with adolescent sterility. Therefore, the government should ensure the existing

marriage act where age at first marriage for the females is eighteen years. In this

context campaign should be launched about marriage act through mass media. In

addition to mass media elected members of the local government, Imams and

marriage registrar (Kazi) can also be involved. Priority should be given in

education, particularly for females. Females should be aware about negative health

consequence for early marriage and early conception. Job opportunities must be

increased inorder to delay the first marriage as well as age at first motherhood.

To check the higher growth rate of population the mean age at first conception

should be raise among the women of rural areas. This can be done to increase the

mean age at first marriage, which can be implemented by increasing the literacy

rates of the females. Regional differentials of age at first marriage and fertility are

pronounced. Regional differentials regarding mean age at first marriage,

conception wait, mean age at first conception and mean age at first birth can be

minimized by implementation of Government policies uniformly across the

regions in Bangladesh. Emphasize should be given among vast majority of

Muslim women to increase age at first marriage as well as age at first conception

for reducing fertility. With the fruitful implementation of the aforesaid suggestions

the changes in timing of marriage, timing of first conception, level of fecundability

and age at first motherhood will contribute to decline of fertility in Bangladesh.

7.5 SUGGESTIONS FOR THE FURTHER RESEARCH

171


This study is a modest attempt to estimate the “Estimation of first Conception

Waits, Fecundability and age at first Conception among Women in Bangladesh”.

Undoubtedly this task is complicated because we have considered here the

retrospective (birth history) data, which is not free from errors. Therefore

longitudinal data can be used for further studies in order to in-depth analysis of the

reproductive behavior of the women of Bangladesh. This can be possible with the

proper implementation of the vital registration system across the country.

APPENDIX-1172


PROGRAM FOR FINDING THE ESTIMATED VALUE OF TYPE-I GEOMETIC DISTIBUTION

DIMENSION P(2000) OPEN(5,FILE='rattd.out') A=4.195 B=55.32 D1=A D2=A+B P(1)=D1/D2 P2=P(1) DO 10 J=2,111 D1=(B+J-2) D2=(A+B+J-1) D3=D1/D2 P(J)=P(J-1)*D3 P2=P2+P(J) 10 CONTINUE DO 20 J=2,111 P(J)=P(J)/P2 P(J)=100*P(J) WRITE(5,1) P(J) 20 CONTINUE 1 FORMAT(3X,F7.2) STOP END

REFERENCE

173


1. Gini, C.(1924). “Premiers recherches Sur la fecondabilite de la femme, Procedings of the International Mathematical Congress, Toronto,Pp. 889-892.

2. Henry, L (1957). “Fecondit et Female: Models Mathematigues” Population, 16:24-48 and 261-282.

3.United Nations(1958). Multilingual Demographic Dictionary. Population Studies,No.29 New York: United Nations.

4. M atsummoto,Nogami,S. Y. and Ohkuri, S(1962). “Statistical Studies on Mensmatra.” 11:294-318.

5. Vollmen, R. F. (1556).The degree of vasiability of the lengh of the Menstual Cycle in correlation with age of women. Gynaeeologia 142:310-315.

6. Bongaarts J (1975). “A method for the estimation of fecondability.” Demography 12: 645-60.

7. Leridon, H.(1977).“Human Fertility. The basic compounts Chicago : The University of Chicago Press, Chicago.

8. Perrin E.B. and Sleps M. C. (1964). “Humen Pepndation” A Statistic Process,” Biometrics, 20:28-45.

9. Wilson, C. (1987). “The Proximate Determinamts of Mantal Fertility in Eyld 1600-1799” Pp.203-230 in the world We Have Gained. Edited by Llonyd Bonfield, Richard N. Smith, and Keith Wrightson. New York: Basil Blankwell.

10. Larsen,U. and T. W. Vaupel(1993).“Hattesite Fecundability by age and parity:Strategies for Fraity Modeling of Event Histories.” Demography 30:81-102. 11. Potter, R. G. and Parker, M P. (1964): Predicting time required to Conceive, Population Studies, 18:99-116.

12.Westoeff, C. F., Potter, R. G., Sagi, P. G., and Misnler, E. G.(1961):Family growth in metropolitan America, Princeton University Press, Princeton.

13. Jain A. K.(1969). Fecundability and its relatiion to age in as Sample of Tawiwanese Women, Population Studies,23:69-74.

14. Henry,L. (1964a). Mortality intra-uterin et fecundabilite, Population 19:899-940.

15. Glass, D. V. and Grehenik, E. (1959). The Trend and Pattern of Fertility in Great Britain: A report on the Famly Census of 1946.London: H. M. Statiny office.

16.Lachenbruch, Peter A (1967). Frequency and timing of intercourse: Its Relation to the Probability of conception. Population Studies 21:23-31.

174


17. Tictze, C. (1960). Probability of Pregnency. Resulting from Single unpretected Coitus. Fertility and Sterility 11:485-488.

18. Barrett, J. C. (1969). A Montocarlo Simulation of Human Reproduction. Genes 25:1-22.

19. Barrett, J. C. (1971). Fecundability and coitus Frequency, Population Studies 25:309-313.

20. Henry, Louis, (1953). Fecundabilities des Marriages. Paris; Peesses Universities de Frace.

21. Potter, R. G.(1961). Length of the Fertile period. The Milbank Memorial Fund Quartery 39:132-162.

22. Sheps, (1965). An Analysis of Reproductive Patterns in an American Isolate. Population Studies 19:65-80.

23. Majumder, H, and Sheps, M. C. (1970), Estimation of a type-1 Geometric distribution from observations on Conception Times. Demography 7:349-360.

24. Sheps M. C. (1964). On the time required to Conceive, Population Studies 18:85-97.

25. Sheps M. C., and Menken, J. A. (1972). On estimateky the link of Conception from Censord data , Pp 167-200 in T. N. E. Greville (ed.), Population Dynamics, New York : Academic Press.

26. Sheps M. C. and Menken, J. A.(1973): Mathematical Models of conception and Brith. Chacago University of Chacago Press.

27. D. Sougs, Srinvab(1973). Interlive birth intervals of Non-Contraceptive Populations:A data Analysis Study. Social Actim 20:404-425.

28. Srinivasion, K. (1966). An Application of a Probability Model. Application to the study of Inter-live birth Intervals, Sankha, Seris B 28: 175-185.

29. Srinivasion, K. (1967). A Probability Model. Application in the Study of inter-live births Intervals and Random Segments of the same Population Studies 21. 63-70.

30. Brass, W.(1958). The distribution of births in Human Populations 12:51-72.

31. Jame W. H. (1963). Estimation of Fecundability Population Studies 24:341-345.

32. Singh S. N. (1963). Probability Models for the variation in the Number of births per Couple. Journal of America Statistics Association 38:711-727.

175


33. Singh S. N. (1969). Low Estimates of Fecundability P. 108 in Preceeding of the International Population Conference vol. 1, London 1969. Leige Belgium. International union for the Science Study of Population.

34. Islam M. M. and Yadava R. C. (1997). Estimation of fecundability: Levels and Trends in Bangladesh. The Journal of Family Welfare 43(3):13-21.

35. Holmberg,I.(1970).Fecundability, Fertility and Farly pling, Gothenburg, Sweden. University of Gothenburg, John, M. A.

36. Men Kew, J. and Chowdhury A. K. MA(1987). The Effect of Breastfeeding and Nutrition on fecundability in Rural Bangladesh: A Hazard Model Analysis. Population Studies 41:433-446.

37. Potter, R.G(1960). Length of the observation period as a factor effecting the

conception failure rates. Milbank Memorial Fund Quarterly, 38:

142-144

38. Mode, C.J. (1985). Fecundability. Stochastic Process in Demography and their

Computer Implication, Spinger Verlag, New York.

39. Balakrishnan,T.R. (1972). Probability of Conception, Conception delay, and

Estimation of Fecundability In Rural and Semiurban Aeas of

Certain Ltin

American countries. Social Biology 26: 226-131.

40. Weinberg, C.R. and Gladen, B.C. (1986). The Beta geometric distribution

applied to

comparative fecundability studies, Biometrics, 42 : 547-560.

41. Vincent, P. (1961). Recherches Sur la Fecondite Biologique Institute National

D’

Etudes Demographic Parts, France, Press Universitaires DE Frace.

176


42. Weinberg, CR. and Gladen, B. C. (1986). The Beta geometric distribution

applied to

comparitive fecundability Studies, Biometrics,42:547-560.

43. Rao, CR. (1952). Advanced Statistical Method in Biometric Research,

Newyork:Wiley.

44. Kandell M.G. and Stust (1958). The Advanced Thory of Statistics, vol. 1,

Newyork:

Hafner.

45.Gautier, E and Henry, L. “La population De Crulai, Parosse Normanda:Etdue

Historique”. Truvaux et Document 33:I, N, E, D, Paris, 1958.

46. Islam, M. M. and Islam M. M. “Birth spacing in Bangladesh patterns,

Differentials

and components”. The Bangladesh Journal of Scientific Research,

13(2).

47. Henripin J. “La Fecundability manages Canadien an debut du 18-eve Siecle”

Travaus

et Documents 22:I. N. E. D., Paris.

48. Palhak, K. B. An Extension of the waiting time Distribution of First

Conception,

Journal of Fi 10:231-34,1978.

49. Meridith, J.A., Menken, J.A. and Chowdhury, A.K.MA.A. The Effects of

Breastfeeding ans Nutrition on Fecundability in Rural Bangladesh: A

Hazard-

177


Model Analysis. Population Studies, 41: 433-466, 1987.

50. Goldmen, N. and Montgomery, M. “ Fecundability and Husband’s Age”.

Social

Biology. 36: 146-166, 1990.

51. Balakrisnan, T.T. probability of Conceptions, Conception Delay, and

Estimation of

Fecundability in Rural and Semiurban Areas of Certain Latin American

Countries. Social Biology. 26(3):226-231, 1988?

52. Islam, M.M. AND Yadava. Estmation of Fecundability: Levels and Trends in

Bangladesh. The Journal of Family Welfare.43(3): 13-21. 1997.

53. Singh, S.N. Probability models for the variation in the number of births per

couples.

Journal of American Statistical Association. 58: 721-727, 1963.

54. Singh, S.N., Bhattacharya, B.N. AND Yadava, R.C. “ On some distribution,

for

couple fertility and their applications”. In s.n Singh(eds) Proceedings of All

India

Seminar on Demography, 76-94, 1997.

55. Henry, L. “ French statistical research in natural fertility”, Public Health and

Population change,edited by M.C. Sheps and J.C. Ridley ,p.342. , 1965.

56.Pathak, K.B., and C.V.S. Prasad. Amodel for estimating adolescent sterility

among

married women. Demography,14: 103-194, 1997.

178


57.Bongaarts, J and Potter, R.G.. “Fertility Biology and Behavior. An Analysis of

the

Proximate Determinants of Fertility.” New York. Academic Press, 1983.

58. Shella, J. (2001) Factors influencing the time taken to initiate women’s

marriage after menarche in Tamil Nadu.Demography, India Vol. 30(1):113-125.

59.Gray, R.H. (1977); Biological factors other than nutrition and lactation which

may influence natural fertility: a review. In natural fertility. Edited by Henry

Leridon and Jane Menken. Ordinn, Belgium.

60.Wood, James W. (1994). Dynamics of Human Reproduction: Biology,

Biometry, Demography. New York, NY: Aldine De. Gruyter

61. Heter,C.L.(1970). The fertility of sterility and subfecund Women in New

Orleans, Socila Biology, 17(30: 195-206.

62.Weir, W.C. and D,R, Weir (1961). The Natural History of Infertility. Fertile

Sterile, 12: 443-141.

63. Abramson, F.D.(1973).” Spontaneous fetal death in man.” Social Biology, 20

(4): 375-403.

65. Potter et. al., (1865). “Fetal Wastage in Eleven Punjab Villages”. Human

Biology, 37 : 262-273.

66.Titeze, C. (1960). “ Probability of pregnancy resulting from a single Act of

intercourse” Fertility and Sterility”, 11 : 485-488.

179


67. Hartman,C.C. (1962). “ Science and the safe period” Williams and Wilkins ,

Baltomore.

68. James, W.H. (1979). The causes of the decline in fecundability with age.

Social Biology, 26:330-334.

69. Robert, G AND Potter Jr. ( 1961). “ Length of the Fertile Period”. The

MilbankMemorial Fund Quarterly. 39 (1): 132-162.

70. Vandar Vliet, W.L. and Hafez (1974). Survival and Aging of Spermatozoa: A

Review. American Journal of Obstetrics and Gynecology, 118 (7): 1006-1015.

71. Barrett, J.C. AND John Marshall (1969). :” The Risk of Conception on

Different Days of Menstrual Cycle” Population Studies, 23 : 455—461.

72. United Nations (1959). “Multilingual Demographic Dictionary.” Population

Studies, No.29, United Nations, New York.

73. McDonald, P.F. and Abdurahman. (1974). ;Marriage and Divorce in West

Java: An example of the Effective use of Mrital Histories”. Jakarta: Lembaga,

Demografa, Fakultus Ekonomi, Universitas, Indonesia (Mimeo).

74. Bangladesh Bureau of Statistics (BBS) , 1991. Bangladesh Population Census

Report. National volume , Dhaka, Bangladesh, 1994.

75. Balakrishnan, T.R., K.V.Rao., E Lapierrie-Adamcyk and K.T.Krotki

(1987)..”Ahazard model analysis of the covarates of the marriage dissolution in

Canada”. Demography 24(3) :395-407.

180


76. BANGLADESH BUREAUE OF STATISTICS (BBS) Bangladesh Population

Census 2001, National Report (Provisional) , Planning Division. Ministry of

Planning Government of the Peoples Republic of Bangladesh, 2001.

77. S.N. MITRA., A.A.SABIR, T. SAHA AND S.KUMAR. Bangladesh

Demographic and Health Survey 2007. National Institute of Population Research

and Tanning (NIPORT).Miytra and Asoociates, Dhaka, ORC Macro: Calverton,

U.S.A.,2009

78.Dr. M. Mazharul Islam is Associate Professor in the Department of Statistics, University of Dhaka, Dhaka 1000, Bangladesh. Dr. R. C. Yadava is a Professor in the Department of Statistics, Banaras Hindu University, Varanasi 221005, India.

79. Quamrun Nahar( AUGUST 2009) ICDDR,B, Mohakhali, Dhaka 1212, Bangladesh;

Trends and Determinants of Adolescent Childbearing in Bangladesh

80. USAID Country Health Statistical Report, Bangladesh May 2009

81. Jeffrey Kallan and (Feb., 1986), J. Richard Udry Source: Demography, Vol. 23, No. 1 pp. 53-66

The Determinants of Effective Fecundability Based on the First Birth Interval Author(s):

82. Anrudh Kumar jain, Fecundability and its relation to age in sample of

82. Perrin EB and Sheps MC: 'Human reproduction: A stochastic process', Biometrics, 20:28-45 (1964).

83. Bongaarts J: 'A method for the estimation of fecundability', Demography, 12:645-660 (1975).

84. Leridon H: 'Human fertility: The basic components', translated in English by Judith F Helzner, The University of Chicago and London (1977).

85.Potter RG and Parker MP: 'Predicting the time required to conceive', Population Studies, 18:1 9-116 (1964).

86.Sheps MC: 'On the time required from for conception", Population Studies, 18:85-97 (1964).

87.Jain AK: 'Fecundability and its relation to age in a sample of Taiwanese women', Population Studies, 23: 6985 (1969).

181


182

final thesis complete

Documents

estimation of conception

differentials of conception

fecundability level109chapter

department of statistics

mean conception delay

differentials of age

thesis work

entire research work