final thesis complete
TRANSCRIPT
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
99
DedicatedTo
My BelovedParents
AndEldest Brother
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
110
CHAPTERONE
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
152
CHAPTERTWO
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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....
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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....
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
(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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
; 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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
, 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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
= -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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
=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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
≤ 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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Age at first marriage
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
0
5
10
15
20
25
30
≤13 14 15 16 17 18 19 20 21 22-23
24-25
26+
Age at first marriage
Me
an
co
nc
ep
tio
n w
ait
Mean conception wait
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
209
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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+
Age at first marriage
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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.
3.10.13 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
Variance = S = (ab(a+b-1))/((a-1) .(a-2)) is not
defined unless a>2.
3.10.14 Spousal age difference, conception wait and fecundability
221
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
223
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
98
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Spousal age difference
≤ 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
Chapter Three: Estimation of Conception Waits and Fecundability: Levels, Trends and Differentials
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Mean conception wait
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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( )
Number of respondents
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
128
CHAPTERFOUR
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Number of respondents
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
. 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......
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Table continued......
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Background characteristics
Categories Mean age at first conception
Minimum Maximum Standard deviation
Mean age at first birth
Standard deviation
Number of respondents
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
Husband education level
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........
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
Husband’s occupation
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
Type of place of residence
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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Spousal age difference
≤ 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
Chapter Four: Differentials of Age at First Marriage, Age at First Conception and Age at First Birth
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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.
Age at first marriage
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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+
Age at first marriage
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
149
CHAPTERFIVE
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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.
Background characteristics
CategoriesPercentage of respondents atConception wait (in month)
Value of Chi-square
P-Value
18.32(mean) >18.32(mean)
Age at first marriage
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
Husband education level
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
Husband’s occupation
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
Type of place of residence
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Spousal age difference
≤ 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.....
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
Table 5.2: Cox’s Proportional Hazard Regression of conception waits by different
characteristics of respondents
Background characteristics
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
Spousal age difference
≤ 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........
Table continued......
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
162
CHAPTERSIX
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
predetermined variable should be small difference as the total effect of that
predetermined variable on marriage to first conception wait.
157
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
160
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
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
Chapter Six: Determinant of First Conception Wait A Path Analysis
182