FERTILITY AND CHILDHOOD MORTALITY ESTIMATES

FOR BANGLADESH BASED ON PREGNANCY HISTORY DATA

by

S.M. Shafiqul Islam

A thesis submitted in partial fulfilment of the requirements for the degree of Master

of Arts in Demography at the Australian National University

Canberra, March 1981

D E C L A R A T I O N

Except where otherwise indicated, this thesis is my own work.

(S.M. Shafiqul Islam)

March, 1981.

i

ACKNOWLEDGEMENTS

I wish to express my profound gratitude to Dr S.K.Jain, my supervisor,

for his invaluable advice and constructive comments throughout the study.

I am grateful to Dr P.F. McDonald and Dr L.T.Ruzicka for sparing some of their

valuable time in helping to solve some problems posed in the course of this

study.

I would like to thank the Australian Government for offering me a 'Colombo Plan' Scholarship, and the University of Cittagong, Bangladesh for granting me

study leave. I also thank the Government of Bangladesh for according me permission to use the computer tape of the Bangladesh Fertility Survey, 1975-76.

I am indebted to Mrs Jenny Widdowson and the staff of the Coombs

Computing Services for their help in processing the data. I owe a great debt

to Mr Peter Waddell-Wood who went through the draft and to Mrs Anne Cappello who efficiently typed the thesis. The secretarial assistance provided by

Mrs Sylvia Boyle is gratefully acknowledged.

I am thankful to Dr D.W.Lucas and fellow students of the M.A. Program in Demography for their encouragement.

Finally, my heartfelt thanks go to my wife, Shakila, for her assistance

in preparing the bibliography and proofreading and also for her encouragement

and patients throughout the study. My children, Shami and Samina made my

stay in Australia so interesting.

S.M.Shafiqul Islam.

ii

ABSTRACT

The main purpose of this study is to estimate fertility and childhood

mortality rates for Bangladesh based on the pregnancy history method using

the data from the Bangladesh Fertility Survey, 1975-76. Fertility remained

high in Bangladesh up to 1970. A substantial decline in fertility was found

to have occurred during the period 1971-75. The observed dip in fertility

has reflected the effects of the 1971 war of liberation and the 1974 widespread

famine in Bangladesh. This dip in fertility was probably temporary and it

may revert to its previous level. The total fertility rate for Bangladesh

was estimated to be 5.9 per female during the period 1971-75. Observed

fertility in the 5-10 years before the date of the survey was found to be

exaggerated which perhaps could be associated with the errors in the data

resulting from misplacement of events in time. Observed fertility beyond the 15 years preceding the survey was found to be considerably affected by

response errors, especially misplacement and omission of events. Fertility rates were generally higher in rural areas than in urban areas.

Mortality under age 5 (l^o and 5 o) has, in general, declined during the

period 1955-69. However, no apparent decline in mortality of children aged 1-4 was noticed for Bangladesh. Estimates of l^o, 4^1 and 5* 0 were found to

be 133, 87 and 208 per thousand live births respectively during the period

1965-69. Infant mortality was found to have risen during the period 1970-73.

This rise was perhaps the reflection of the effects of the devastating cyclone

in 1970 and the subsequent war of liberation in 1971. Childhood mortality

(4* 1 and 5 o) could not be estimated because of the fact that some births of

children would not have been exposed to the full risks of dying during this

period. Infant mortality of 139 per thousand live births during the period

iii

1970-73 was about 9 to 14 per cent lower than the corresponding estimates

for this period available from other studies based on different methods of

estimation. Mortality of children under age 5 during the period 1965-69 was

also under-reported by about 6 to 13 per cent as compared to the estimates

derived in other studies for the comparable period of time. Infant and

childhood mortality rates were generally higher in rural areas than in urban

areas. In general, 4^1 estimates were higher for females while l^o and 5^0 estimates were higher for males.

iv

LIST OF CONTENTS

PageACKNOWLEDGEMENTS i

ABSTRACT iiLIST OF TABLES vi

viiiLIST OF FIGURES

CHAPTER

1 INTRODUCTION

1.1 Background and Objectives 1

1.2 The Study Area and the People 2

1.3 Review of Literature in the Context of the Present Study 6

1.4 Data Source 8

1.5 Organization of the Study 11

2 METHODOLOGY OF ANALYSING PREGNANCY HISTORIES AND BASIC DATA FOR THE STUDY

2.1 Introduction 122.2 The Mechanics of Pregnancy History Analysis 13

2.2.1 Errors and Biases in Pregnancy History Data andTheir Effects on the Estimates of Fertility and Childhood Mortality 16

2.3 Description of Data for the Study 213 FERTILITY

3.1 Introduction 29

3.2 Some Determinants of Fertility 313.3 Life-time Fertility 35

3.4 Current Levels of Fertility 37

3.5 Trends in Fertility 41

3.6 Age Patterns of Fertility Observed in Various Studies 47

3.7 Effects of Possible Sources of Errors on Fertility Estimates 52

3.8 Adjustment of Fertility Rates 58

V

CHAPTER Page

3.8.1 Brass's First Birth Procedure 58

3.8.2 P/F Ratio Method 60' 3.8.3 Gompertz Relational Model 63

4 INFANT AND CHILDHOOD MORTALITY

4.1 Introduction 68

4.2 Infant Mortality 70

4.3 Comparison of Infant Mortality Rates With Those Obtained in Other Investigations 73

4.4 Childhood Mortality 77

4.4.1 Mortality in the Age Group 1-4 (4qi) 77

4.4.2 Mortality Between Birth and Exact Age 5 (5qo) 83

5 SUMMARY AND CONCLUSION 87APPENDIXAl Pregnancy History Questionnaire 92A2 Single Year Age Distribution of the Respondents by Area in

Bangladesh, 1975-76 95A3 Live Births for 5 Year Age Groups of the Respondents

Calendar Year in Rural Bangladesh, 1941-76by Single

96A4 Live Births for 5 Year Age Groups of th e Respondents

Calendar Year in Urban Bangladesh, 1941-76by Single

97

A5 Live Births for 5 Year Age Groups of the Respondents Calendar Year in Bangladesh (Total Weighted Sample)

by Single , 1941-76 98

Bl Percentage Age Distribution of Ever-Married Females Bangladesh

in99

Cl P/F Ratios for Different Periods Preceeding the Survey under theAssumptions of Constant and Changing Fertility: Bangladesh 100

C2 P/F Ratios by Birth Order for Recent Periods: Bangladesh 101

D1 Estimates of Probability of Dying Between Birth and x, xq0 (x = 2, 3, 10) by Area for Bangladesh

Exact Age102

REFERENCES 103

LIST OF TABLESTable Title Page1‘1 Summary Measures of Fertility and Infant Mortality for

Bangladesh Derived from Analyzing Pregnancy History DataCollected in Various Studies. 8A

2.1 Percentage Distribution of Reporting of Data of SomeEvents in the Bangladesh Fertility Survey, 1975-76 25

2.2 Sex Ratio and Birth Per 100 Females for Bangladesh,1941-75 27

3.1 Summary Measures of Fertility for Bangladesh, 1901-74 303.2 Percentage of Women in Each 5 Year Age Groups who are Ever-

Married in Bangladesh, 1951-76. 33

3.3 Mean Number of Children Bom Alive per Ever-MarriedWoman in Bangladesh, 1961-76 36

3.4 Age Specific Fertility Rates for Bangladesh, 1971-75 39

3.5 Age Specific Marital Fertility for Ever-Married Women and Changes in these Rates for Selected Years Between1966 and 1975 in Bangladesh 42

3.6 Age Cohort and Cross-Sectional Age Specific MaritalFertility Rates for Ever-Married Women in Bangladesh 44

3.7 Total Marital Fertility Rates for Ever-Married WomenAged 15-44 Years 46

3.8 Percentage Distribution of Fertility by Age Groups andTotal Fertility Rates for Bangladesh : BSF and Other Sources 48

3.9 Proportion of Total Fertility in Specified Age Groups :Pavlik's Model and United Nations' Model 50

3.10 Classification of Types of Fertility Patterns in Bangladesh According to the Proportion of Total FertilityFalling into Specified Age Groups by Pavlik's and U.N. Models 51

3.11 Sex Ratio at Birth by Age of Mother and Years Before theSurvey per 100 Female Births in Bangladesh. 56

4.1 Estimated Crude Birth Rates for Bangladesh, 1901-74 69

4.2 Infant Mortality Rates ('000 Live Births) by Area and bySex for Bangladesh, 1955-73 71

4.3 Sex Ratio at Infant Death ('00 Females) by Area forBangladesh, 1955-73 73

4.4 Estimates of Infant Mortality ('000 Live Births) forBangladesh from Different Sources, 1957-75 75

4.5 Estimates of Probability of Dying Between Ages 1 and 4 (4 1) by Area and by Sex for Bangladesh, 1955-69

vi

81

Vii

Table Title Page

q4.6 Estimates of Probability of Dying Between Ages 1 and 4 (4 1)for Bangladesh from Different Sources 82

4.7 Estimates of Probability of Dying Between Birth and ExactAge 5 (5 o) by Area and by Sex for Bangladesh, 1955-69 84

4.8 Estimates of Probability of Dying Between Birth and ExactAge 5 for Bangladesh from Different Sources 85

viii

LIST OF FIGURES

Figure Title Page

1.1 Bangladesh Fertility Survey : Scatter of Sample Points 10

2.1 Birth Cohort and Cross-Sectional Fertility Using theLexis Diagram 15

2.2 Infant and Childhood Mortality Using the Lexis Diagram 17

3.1 Age Specific Fertility Rates for Bangladesh fromDifferent Sources 53

4.1 Probabilities of Dying Between Birth and Exact Age x, x^ofor Bangladesh, 1955-73 78

4.2 Probabilities of Dying Between Ages 1 and 4, 4^1 and Between Birth and Exact Age x, x\> for Bangladesh, 1955-69 79

CHAPTER 1

INTRODUCTION

1.1 Background and Objectives

Bangladesh is faced with a serious population problem. Even with substantial fertility control efforts, the 1975 population of Bangladesh

will nearly double by the turn of this century to about 150 millions

(Frejka, 1974 ; Revelle et al., 1972). Khan (1973b : 188-196) has

estimated that the population of Bangladesh has doubled during the 43 years

between 1931 and 1974. This alarming rate of population growth (estimated

at about 3 per cent per annum) has been threatening the social and economic

development of Bangladesh. The Government of Bangladesh has recognised that if the present rate of population growth is sustained, all the efforts

towards development will be frustrated. In its Five Year Plan (1973-78), the Government of Bangladesh has emphasised the need for an intensive

delivery of contraceptive devices and family planning services in arresting the galloping rate of population growth and its consequent economic and

social evils (Planning Commission, 1973). Recognising the implications of the higher rate of population growth upon the process of socio-economic

development of the country, the government has termed the population problem as the "number one problem of the country" in 1976.

The system of vital events registration which can provide detailed

fertility and mortality data representative of the whole national population at any point of time is still non-existent in Bangladesh. Consequently,

attempts have been made to derive the estimates of fertility and mortality

through indirect methods using the data from censuses and specially designed

sample surveys. Although several studies have examined the levels and trends

in fertility and mortality in Bangladesh, some of them revealed contradictory

2

results while some of them were of limited value due to the inadequacy of

data (Khan 1973b : 192; Sirageldin et al., 1975 : 207; PCFP Division, 1978 :

1; Cain et al., 1979 : 5-6; Amin et al., 1980 : 1-2). Moreover, most of

the studies are out of date to reflect the changes that have occurred,

particularly due to the social disruptions caused by the war of liberation

in 1971 and the widespread famine in 1974.

National estimates of fertility and mortality have great policy

significance. These can provide the basis for formulating various national

policies. Accurate measurement of the levels and trends in fertility and

infant and childhood mortality can serve as the basis for establishing

benchmark estimates against which targets could be set. They could also

be used for evaluating the achievement of the government family planning

program. Thus the main objective of this study will be to estimate the levels

and trends in fertility and infant and childhood mortality for Bangladesh

based on the direct method of pregnancy history analysis. This study will

also explore the applicability of the pregnancy history method in the context

of Bangladesh.

1.2 The Study Area and the People

Bangladesh lies approximately between 20°30' and 26°45' North Latitudes

and 88°00' and 29°56' East Longitudes. Except for a short border with Burma

in the southeast, the country is bounded by a long land border with India on

the West, North and East and by the Bay of Bengal on the South. The low-

lying riverine country has a total area of 55,598 square miles including river

surface. More than 85 per cent of the total land is flat alluvial plain

transversed by three major rivers, the Ganges-Padma, the Brahmaputra-Jamuna

and the Megna, and their innumerable tributaries. Bangladesh has a humid

sub-tropical climate with three seasons: a hot summer with high humidity

from March to June, a moderate hot and humid monsoon from June to early October,

3

and a cool drier winter from mid-October to early March. Average annual

temperatures vary between 57°F and 80°F. January is the coolest month while

the hottest months are April and May. The humidity is high throughout the

year. During the most humid months of June and July, the relative humidity

across the country ranges from 84 to 90 per cent. Mean annual rainfall

varies from 50 inches in the west to 100 inches or more elsewhere. The

north-eastern hilly region,which is one of the heaviest rainfall areas in

the world, has a mean annual precipitation of 200 inches or more. This

heavy rainfall spreads over the whole of monsoon season and causes frequent

floods. Tropical cyclones often accompanied by tidal bore from the Bay of

Bengal and floods have, on many occasions, brought about havoc in the country

resulting in considerable loss of human lives and properties (Nyrop et al.,

1975 : 56-65; PCFP Division, 1978 : 3-4).

The area now encompassed by Bangladesh has been known by a number of names. Before August 14, 1947, it was known as East Bengal, a part of the British Indian state of Bengal. Between 1947 and December 1971, it was the eastern province of Pakistan called East Pakistan. During the 24 years of union, the two wings of Pakistan (east and west) separated by 1500 miles of

foreign territory had an uneasy partnership with occasional regional,

economic, political and cultural conflicts (Curlin et al., 1976 : 88). A

political debacle escalated into a civil war in March 1971. The civil war

culminated in an all-out war of liberation against West Pakistan. After a

tremendous sacrifice of life and prox^erty, Bangladesh emerged as a sovereign

and independent state on December 16, 1971.

Bangladesh is the eighth most populous country in the world. According

to the 1974 census, the population was 76.2 million. Density of population

is one of the highest in the world with an average of 1,283 persons per

square mile. With an estimated growth rate of 2.9 per cent per annum, the

population will double in 24 years (PCFP Division, 1978 : 8). The population

4

was fairly stable at about 17 million during the eighteenth century (Mosley and Hossain, 1973:8). Since then it began to grow slowly reaching 29 million

in 1901 and approximately 55 million in 1961. The rate of growth has increased

considerably in the decade of sixties (Mosley and Hossain, 1973 : 8-9). The

population of Bangladesh is characterised by a youthful age structure with about

48 per cent below age 15, and only about 3 per cent aged 65 and over (Moseley

and Hossain, 1973 : 10; Bangladesh Bureau of Statistics, 1977 : 13).

Bangladesh is predominantly a rural country with 91 per cent of the

population living in rural areas (Bangladesh Bureau of Statistics, 1977 : 14).

With the exception of some tribal population in the hilly regions, the popul­

ation is ethnically homogeneous. The major religious community in the

country is formed by the Muslims. More than 85 per cent of the population

are Muslims who have largely accounted for the overall growth of the population. Among the other religious communities, Hindus comprised of

about 14 per cent and Christians and Buddists constitute only 1 per cent of the total population (Bangladesh Bureau of Statistics, 1977 : 23).

Marriage patterns in Bangladesh are characterised by near universality and

early age at marriage for women. According to the 1974 census, only 1.1 per cent of males and 0.3 per cent of females reported to have never married in the age group 45-49. Singulate mean age at marriage based on the 1974 census

data were estimated at about 16 years for females and 24 years for males.

Both for males and females, the values of the singulate mean age at marriage

have increased by about 1.5 years since 1951 (Bangladesh Bureau of Statistics,

1977 : 18-20). The practices of near universality of marriage and low age

at marriage are associated with the prevailing socio—religious customs in

Bangladesh (Ahmed, 1978 : 3).

The literacy rate in Bangladesh is very low. It is lower among females

than males. Only 27.6 per cent of males and 12.2 per cent of females were

able to both read and write in any language according to the 1974 census.

5

Slightly less than one-fourth of the population aged 5 years and over were

literate. The average literacy rate increased by 0.26 per cent per annum in

rural areas while that for urban areas increased by only 0.02 per cent

during the intercensal period between 1961 and 1974. This lower increase in

the literacy rate for urban areas was probably due to a large influx of

illiterate males from rural areas in search of work (Bangladesh Bureau of

Statistics, 1977 : 27).

The economy of Bangladesh is chiefly dominated by agriculture which

contributes about two-thirds of the gross national product (Chen and

Choudhury, 1975 : 201). Agricultural land in Bangladesh has been

fertile and about 80 per cent of the total cultivable land provides for

food production. Despite the predominance of agriculture, food supply hasbeen inadequate for the needs of the population even in the periods of good

harvest (Rosenberg, 1973 : 31). According to Khan (1972 : 25-27)."... nearly half the population are on a starvation diet, nearly 90 per cent have some kind of deficiency in food, and that a vast majority are ill-clad and nearly shelterless".

Further, the agricultural production deteriorated in recent years due to

cyclone, war and monsoon floods resulting in a decline in the per capita

availability of food for consumption (Chen and Choudhury, 1975 : 207). By

1975, the population was economically worse off than it had been before

independence."Average per capita income was the equivalent of about US$70 a year; the poor majority of the population had substantially less on which to live".(Nyrop et al., 1975 : 209).

Crude activity rate in Bangladesh is low. The percentage of the total population economically active has declined from 34.3 in 1961 to 28.7 in

1974. More than three-fourths of the economically active population 10 years

and over were engaged in agriculture. Only 4 per cent of the females aged

6

10 and over were economically active according to the 1974 census (Bangladesh

Bureau of Statistics, 1977 : 35-41). This low activity rate among the

females may be due to the prevailing social customs that discourage

participation by women in outside work for wages with their male counterparts.

Bangladesh has a high dependency ratio of 105 persons which is

mainly due to an exceptionally high proportion of young persons in its age

structure (Bangladesh Bureau of Statistics, 1977 : 13).

1.3 Review of Literature in the Context of the Present Study

A number of studies have been conducted both within and outside the

country in order to estimate the levels and trends in fertility and infant

and childhood mortality based on pregnancy history data. In this section,

only those studies undertaken in Bangladesh that have used the pregnancy

histories will be reviewed briefly. However, other relevant studies will be

referred to at appropriate places in the text of this thesis.

The earliest demographic survey which collected pregnancy histories was conducted in Bangladesh in 1961-52 by the Statistical Survey Research Unit in collaboration with the Institute of Statistical Research and Training, University of Dacca. The survey covered approximately one-sixth of the

entire area of Bangladesh and slightly more than one-fourth of the total population of the country at that time. Pregnancy histories of a sample

4,200 ever married females both from rural and urban areas were collected.

Since the pregnancy history data were collected by male interviewers, the

quality of the data might have been impaired. Schultz and DaVanzo (1970)

using the pregnancy history data from this survey estimated various vital

rates for the period 1952-61 and found these estimates consistent with other

sources of demographic information. They observed general declines in

fertility and infant and childhood mortality in the 1950's. In this study, they have attempted to develop multivariate statistical models to determine

whether the trends in vital rates were likely to have continued in the 1960s.

7

Schultz (1972) summarized in a paper the findings of the above study.

The National Impact Survey, 1968-69 conducted in Bangladesh was the

second survey which collected detailed pregnancy histories. This survey was

sponsored by the then Pakistan Family Planning Council. A complete pregnancy

history was obtained from all ever-married women aged 55 years or less in a

sample of about 3,000 households representative of all Bangladesh. Sirageldin

et al., (1975) using pregnancy histories calculated the age specific marital

fertility rates for the three periods 1960-62, 1963-65 and 1966-68 and

found fertility declines for all 5-year age groups of women aged 15-44. The

declines were largest between the periods 1963-65 and 1966-68 and, on the

whole, the total marital fertility rate declined by about one-fifth in the

period 1960-68. Sirageldin et al, (1975) also estimated the cumulativefertility of currently married women in 1959, 1964 and 1969 by birth cohort

and observed that more recent cohorts experienced higher fertility at younger ages and lower fertility at older ages than their predecessors. Infant

mortality estimates for the 3-year periods between 1960 and 1968 also showed a declining trend in this study, although the rates were somewhat lower than

those obtained in other studies.

The third sample survey in which pregnancy histories were used was

conducted in Kotwali Thana under Comilla district of Bangladesh in 1968.

The survey was carried out under the auspices of the Bangladesh Academy for

Rural Development. Pregnancy histories were collected from a sample of 1,600

currently married women. One of the main objectives of the survey was to

study the trends in fertility and infant mortality rates in a particular area

of the country where an experimental family planning program had previously

been launched. By applying Bogue's program of pregnancy history analysis, Stoeckel and Choudhury (1973) estimated the age-specific marital fertility

and infant mortality rates for the periods 1958-59, 1963-64, 1964-65 and 1966-67.

They observed that the total marital fertility rate declined by approximately

8

27 per cent and infant mortality rate fell slightly over one-fifth between the years 1958 and 1967. As the sample size was relatively small and the

length of the periods for which the above rates were estimated was short, the

rates were subject to considerable sampling variations. Nevertheless, in

this study, the estimates of age-specific marital fertility rate were lower

and infant mortality rates were higher than those obtained in the two other surveys based on pregnancy history analysis.

Bangladesh Fertility Survey, 1975-76 is the fourth survey in which

detailed pregnancy histories from nationally representative sample of ever-

married women were collected. Various studies using pregnancy history data from this survey are being conducted. Bradd and Rashad (1980) have calculated

the number of births per woman and the cumulated fertility rates for the

5-year periods from 1936-40 to 1971-75 and for various age cohorts of women. They have also attempted to adjust the levels and trends in fertility in Bangladesh using the P/F Ratios and the Gompertz Relational Model.Arriaga (1980) estimated the infant mortality rates for 5-year periods before the survey and also studied infant mortality differentials according

to age of mother at the time of survey, rural-urban residence and education of mothers based on the pregnancy history data from nine Latin American

and Asian countries including Bangladesh. The sources of data for this

study were the surveys conducted under the auspices of the World Fertility Survey. Table 1.1 summarizes the measures of fertility and infant mortality

for Bangladesh derived from the earlier studies that used pregnancy history data.

1.4 Data Source

The present study is based on the data collected in the Bangladesh

Fertility Survey (BFS) in 1975-76. This survey was carried out by the

Population Control and Family Planning Division, Ministry of Health and

Population Control, Government of the People's Republic of Bangladesh in

collaboration with the International Statistical Institute, the Hague. A

8a

wwHQDE-*

K C/3coW w

Pm CP

Pm U O 2

HCO CO

CO H<PwQ

ror~

o cn or - in XS rH O CO

X3 cn r - P '— 00 cnP i—i P cn fd cn rM

03 fd -- - •H i—1 >1 XS rH -—1U XJ rH P P -- -P N 0 rH 03 P fd cdP P N 03 • M X XJ Cn0 i—i p Cn rH Ü XJ cn fd fdCO P (d fd fd 03 p in X ■H

X > P O 0 fd cn Pu (d •H P P X p fd Pco Q CO 03 CO U CQ CP 5

>1p-H

<dppO£

Paid

4MpH

O cn 40 ro 43 CD CO ^m n h h t-~ ^ m rorH r—I i-M I—I I—I I—I I—1 | —I

+

oo m m cn co rorH r—I l-M I—I

03P<dcP

> ip•H

rH -H id -P -P P O 03 Em Pm

ro 00 cn n 4 ) v f H40 <n <n tn 03 H CN

co r - co co io h 4 )I

>1 p •H I—I •H -pp0) 03 Pm -P

fd-H (P fd -poEM

in ro CO CN CNro ^ m o o

co t"~ r - 40

<d 43 ^ e-- ro ro

co mI I I I

X) CD O CN in 00 cn •P1 in r - in0 m CD ID CD CD m 40 ID CD 0-

•H 1 1 1 i 1 i 1 1 1 iP ro !-~ O ro ID CO ro 40 rH in o rH CN ro0) m m CD CD CD in cO CD 40 t - r^- e'­ r - e '­ e'­PM cn cn cn <n cn cn cn cn cn cn cn en (Ti en en

rH i—i rH i—! rH . rH i—1 r—1 rH rM rH rH rH i—i rH

cnX -P P

•rMXI

03>

•H

03rH T)■a §rH tn •H Pcd o > J3 (d -P

I -H

0)-Pos

00 I—I

CNO'cnrH

N-prMP

x :u

CO

rHCDI

r -mcnI—I

XJO

•HP03a

03X-P

o-p

inp03

4m03od

p03eo

XJ03

•Hppcde

3>irH-Pp03PPPU

+ * * *

XlPfd

XJO

X-P03e

o•HP<dp XJ

0IP X\ PC4 03

e0)

A 0P ■H

PCn cdP P

•HCD CmP \

CP00ro 03

• Xr ' P

X) CnP Pcd •H

cno Pm

• 40f ' ro

0) e'­p0) en

cd5

in03 03p Pcd cdp p

>iP P•H •HrH rH•H kO "HP rM PP 03 P03 > 0)

4M •H 4MP

rM O rMcd 03 cdP 0-t P0 cn op 03 P

pXJ XJ03 rM 0)P 03 Pcn xJ cnP O P

•i—i E - nxi XJcd rH cd

cdCn P CnP O P

•H •H "HXJ P xJP cd pO rH OC4 03 Q,cn P cn03 03P N pP p p0 P 0u 03 Ü

CP03 E <D

X O rPEm O EM

cd JQ

9

detailed description of the methodology of the data collection including

the sample design for the survey is given in PCFP Division (1978). However,

it seems pertinent to discuss here briefly the sample design, sample size

and criteria for selection of the respondents for the survey.

The survey was based on a three-stage sample design of non-institutional

households for both the rural and urban strata. A sample was drawn independ­

ently in all three stages. In the first stage, 160 rural and 80 urban primary

sampling units (PSU) were selected with probability proportional to size.

Each PSU consisted of a few contiguous villages in the rural stratum and

a few census blocks in the urban stratum. From each selected PSU, one village

in the rural areas and one census block in the urban areas was selected again

with probability proportional to size. These were the secondary sampling

units or the intermediate sampling units (ISU) and each consisted of

approximately 50 households. Figure 1.1 shows the distribution of these

240 villages and census blocks. Complete listing of households in each

selected IUS was made. The households (ultimate sampling units) were then

selected from each ISU with probability inversely proportional to the

measure of size used in the selection of the respective PSU and ISU. The

number of selected households varied between 15 and 45 in a rural ISU and

between 5 and 29 in an urban ISU. Both rural and urban samples were

self-weighting. In order to obtain national estimates, weighting was

necessary.

The total number of households selected in the sample were 6,145 out of

which 4,626 were rural and 1,519 were urban. However, 4,437 households were

successfully contacted in the rural areas and 1,418 in the urban areas.The non-contact rates were 4.1 per cent in rural areas and 6.6 per cent

in urban areas. 'Dwelling vacant' was the main reason for non-contact

in both rural and urban areas (PCFP Division, 1978 : 28). Details of

residence, marital status, age, sex and educational attainment of all

members of the selected households were recorded on the household schedule

of the Bangladesh Fertility Survey. After the household interview, all

(INDIA

)FIGURE 1.1 10

BANGLADESH FERTILITY SURVEY SCATTER OF SAMPLE POINTS

District Boundary : URBAN SAMPLE : .RURAL SAMPLE : *

(INDIA)

A S S A M

* ( B0GRA. L* I , V»S> *

* D \CCftl * / * TRIPURA^ PABNA

I K i r v = - '4

H, x> ■

i

BAY OF BENGAL

Source: Reproduced from PCFP Division (1978 : 24)

11

the night preceding the interview were eligible for an individual interview.

The individual questionnaire was thus based on a de-facto system and contained

11 sections in the following order: respondent's background, marriage

history, total pregnancy history. Knowledge and use of contraception,

fertility regulation, work history, current (last) husband's background,

children's education and work, assets and expenditure, abortion, and

height and weight. Out of 6,648 eligible respondents identified in the

household interview, 6,513 were successfully interviewed in the individual

interview with a non-response rate of 2.0 per cent. Among the successfully

interviewed females, 5,024 were from the rural areas and 1,489 from the

urban areas.

1.5 Organization of the Study

The study is organized into five chapters. Following this introductory

chapter, the methodology of analysing pregnancy histories including the

errors and biases of the method and the description of data used for this

study are presented in Chapter 2. Chapter 3 deals with the levels, trends

and patterns of fertility. Infant and childhood mortality levels and

trends are presented in Chapter 4. Finally, Chapter 5 summarises and draws

conclusions for the whole study.

CHAPTER 2 12

METHODOLOGY OF ANALYSING PREGNANCY HISTORIES AND BASIC DATA FOR THE STUDY

2.1 Introduction

In most of the developing countries where demographic data for measuring

fertility and mortality levels and patterns are inadequate or unreliable,

a wide range of techniques have been developed over the last three decades

using retrospective reports of births and deaths at censuses and surveys.

One such technique employed in the present study is the technique of

pregnancy history analysis. This technique "...not only is correct in terms

of demographic theory, but also seems to be practicable when put to use under

the conditions that exist in the developing countries ..." (Bogue and Bogue,1970 : 1). According to Brass (1977 : 89-90), the data obtained frompregnancy histories might be the most promising means for studying thetrends in fertility in the face of inadequate data of traditional type

and this technique could be superior in reliability than the repeated surveysto derive the level of fertility at intervals of a few years. He cited anexample in the context of Bangladesh where a series of surveys were conducted

which failed to provide a definite trend in fertility due to varying amount

of biases in the surveys. Shryock and Seigel (1973 : 832) suggested that

"... a successful attempt to record pregnancy histories from a survey would not only substitute for what traditional systems obtain from a combination of vital statistics registration and census but would also present analytical possibilities far superior to the traditional system, which is subject to inconsistencies in the classificiation of flows and stocks by various demographic criteria".

These are probably the reasons why the program of the World Fertility Survey,

for example, has employed the pregnancy history approach to determine the

levels and trends in fertility and to analyse fertility differentials for a

number of countries.

13

A pioneering example in which the pregnancy history data were collected is the 1946 Family Census of Great Britain. Glass and Grebenik (1954) in

their report on the family census presented precise measurements of fertility

for different periods and the completed family size for marriage cohorts

from 1910 to 1934. Later the technique of pregnancy history analysis has

been elaborated upon by Bogue and Bogue (1970), Brass (1971, 1975, 1977 and

1979), Brass and Rashad (1980), and Potter (1975, 1977a and 1977b). However,

a brief overview of the pregnancy history method and its biases are

presented in this chapter. The chapter also describes the basic data for

analysing fertility and childhood mortality in the context of Bangladesh.

2.2 The Mechanics of Pregnancy History AnalysisThe pregnancy history analysis is based on the chronological record of

each pregnancy experienced by each woman of childbearing age. Besides the

pregnancy itself, information about the date of birth of mother, year and month of birth of child, sex of the live-born child, whether the pregnancy resulted in a live birth or still birth, whether the pregnancy terminated in

spontaneous or induced abortion, whether the live-born child was still living or dead, and age at death of deceased child in year and month are recorded for each pregnancy (Bogue and Bogue, 1970 : 3-4).

To get the fertility estimates, information obtained from pregnancy

histories can be transformed into two matrices, one for the numerator and

the other for the denominator. The numerator matrix is formed for each

live birth classified simultaneously according to the calendar year in which

it occurred and the age of mother at the time of birth. Each row of the matrix indicates one calendar year and each column indicates one year of age

of mother. The denominator matrix is formed for each woman classified

simultaneously according to the calendar year and the person years by age

of women. In the denominator matrix, each row represents one calendar

year and each column represents one person year by age of women. Thus from

the above two matrices age specific fertility rates can be calculated for a

long period (for example, 35 years if the age of the oldest respondents is 49 years at the time of survey) in the past. However, as the analysis is

extended back into the past, fertility experience is only available for women at young ages. This procedure provides fertility rates both in the cohort

and cross-sectional contexts that may be explained with the help of the

Lexis diagram (Figure 2.1) as follows.

The exact ages of women separated into 5 year cohorts determined by age

at interview are represented on the vertical axis. Periods before the survey

relative to the date of interview in 5 year intervals are represented on the

horizontal axis. Fertility rates for each cohort of women are allocated

to different periods before the date of survey. Thus each cell in the form

of parallelogram centered at exact ages 15, 20, ..... 50 in the Lexis

diagram (Figure 2.1) represents fertility rates specific to both the cohort

and the period.Cross-sectional fertility rates over different periods preceding the

date of survey can be obtained by reading through the columns. Diagonals

give the cohort fertility rates. For example, the uppermost diagonal indicates the fertility rates for the cohort of women aged 45-49 years at the time of interview. To study the change in cross-sectional fertility,

different cohorts can be compared for the same age of women at the time of interview. For instance, to examine the change in fertility of women aged

35-39 (age at end of period) in the 15 years before the date of survey, the

shaded parallelogram in the Lexis diagram (Fig. 2.1) corresponding to women

aged 35-39, 40-44, and 45-49 years at interview in 0-5, 5-10 and 10-15 years

respectively before the survey are to be compared. This ability to provide

an estimate of the change in fertility over time is the main advantage of

pregnancy history analysis (Potter, 1975 : 4-5).

Infant and childhood mortality estimates can also be derived directly

from the mortality information contained in pregnancy histories. Cross­

tabulation of the year of birth of child by age at death in completed years

15

FIGURE 2.1

BIRTH COHORT AND CROSS-SECTIONAL FERTILITY USING THE LEXIS DIAGRAM

Years before the Survey

50

45

40

35

30

25

20

15

10

Exact Age of Women at the Time of Survey

Shaded Parallelograms to be compared in measuring the change in fertility of women aged 35-39 (age at end of period) in the 15 years before the date of Survey.

16

together with the information of the total live births for each year provides

the current as well as past estimates of infant and childhood mortality.

The following Lexis diagram (Figure 2.2) illustrates the procedure.

The deaths occurring at ages 0, 1, 2, 3, and 4 in completed years among

those born in the calendar year t are located in the parallelograms A , A ,o 1A^, A^ and A^ respectively.

If the number of deaths in the parallelogram A is denoted by d , ino othe parallelogram A^ is denoted by d^, ......... , in the parallelogram A^

is denoted by d^ and if B is the total number of children born in the

calendar year t, then the estimates of infant and childood mortality can

be obtained as follows:d +d +d +d d +d +d +d ,+d,,1 2 3 4 _q o l 2 3 4— ; 5 o = ------ -------- whereiqo = -2 ; 4qi

ID B-d Bo

nc3x is the probability of dying between exact ages x and n+x. it is to be

noted that dx (x = o, 1, 2, ....) is analogous to the d^ column of the

cohort life table with radix (i.e., £o) equal to B. However, if the numbers

of deaths in completed years are not known, the above procedure becomes

slightly complicated.

2.2.1 Errors and Biases in Pregnancy History Data and Their Effects on the Estimates of Fertility and Childhood Mortality

The pregnancy history data may be affected by a number of response errors,

particularly in developing countries. These errors may be classified into

three main categories:

(i) Mis-statement of age.

(ii) Omission of events;and

(iii) Misplacement of events over time.

The nature of these errors and their effect on the estimates of

fertility and childhood mortality are discussed below.

17FIGURE 2.2

INFANT AND CHILDHOOD MORTALITY USING THE LEXIS DIAGRAM

tnUrtfCL)>hC■ H

rtfaQ-prtfa)<

\\

\N

/ // A 4 /

/ V/ /

\ > 3 // /

/ /

/ A 2 // /

/ // A

/ // /

/ . // 0 /

4 /

t t+1 t+2 t+3 t+4

Calendar year of Dsath for those Born in the Year t

18

Mis-statement of current age/date of birth in months and year of women (respondents) has serious implications in the estimation of fertility. It creates problem in the definition of cohort and results in the false

decline or exaggeration of cohort fertility. If, for example, the cohort of

women aged 15-19 at the time of interview includes women beyond age 20 years,

the cohort fertility will be spuriously overstated because of the inclusion

of larger number of births. Similarly, the fertility for the cohort aged

15-19 years would be low if it includes women below age 15 years by age

misreporting. Further, if the cohort of women aged 30-34 at interview

includes a considerable number of women who are actually over 35 years old

and none of those below age 30 years, then, in the absence of any other

errors the fertility rate for the cohort may be an underestimate.

Age mis-statement also introduces biases in the study of the change in fertility. If all the cohorts of women were subject to similar age mis­reporting by the respondents, the consequences in the measurement of fertility change over time would not be so serious. However, when the two adjacent cohorts are subject to differential age misreporting, the problem arises. For instance, if the women in the two adjacent cohorts between the exact ages 35 and 45 are intermixed as a result of misreporting of age/

the older cohort will tend to be younger than its actual age and the

younger cohort will be older. Fertility rate for the most recent 5 year

period for the younger of the two cohorts will be underestimated and the

rate for the older of the two cohorts with reported age 40-44 will be over­

estimated in the period 5-10 years precedinq the survey if fertility, in fact, declines after age 30.

The second type of error in pregnancy history data arises due to

omission of births and deaths of children. Available evidence suggests that certain events, such as, children who have died, children who moved

away from home, infants, female and illegitimate children are more likely

19

to be omitted than others from pregnancy histories. As a result, the total number of children ever born is usually understated. Omission of

birth is directly related to the time elapsed since the birth of child

(Brass, 1979 : 16; Potter, 1975 : 6). This may be due to memory lapse

associated with larger number of births, longer intervals and absences

of children through death or migration. The effect of omission of births

is largely dependent on the distribution of omitted births by period and

by cohort. If, for example, the respondents of all ages have a considerable

tendency to omit very young children, fertility in the most recent period

would be underestimated and "the bias would be greater, the shorter the

length of the period chosen" (Potter, 1975 : 6). It is usually suggested

that the omission of births is more common among older women. This results in the underestimation of the level of fertility for earlier periods and

for older cohorts. Consequently, the measurement of the trends in fertility may also be biased by omission of births. This is probably the reason why

the pregnancy history analysis is considered as a "conservative evaluation technique" for studying the decline in fertility (Bogue and Bogue, 1970 : 25).

"An unfortunate practice, no doubt related to this reasoning, is that very often analysts do not present results for more than the last 10 years of

experience covered in a survey" (Potter, 1977a : 337). Omission of children who had died, particularly infants, combined with the underreporting of live

births cause errors in the estimation of childhood mortality levels and trends.

The third type of error occurs when the dates of births and deaths ofchildren are wrongly placed in the different time periods before the

survey. Brass (1979 : 17) explains this error "in terms of a distortion of

the time scale". Some births/deaths that actually occurred in the period

0-5 years preceeding the survey are misplaced in the period 5-10 years

preceeding the survey and events that ought to be reported in the 5-10 years

before the survey are misallocated in the next 5 year period or preceeding

20

5 year period and so on. This distortion of the time scale is called the

reference size bias. Generally, the longer the reference period, the

longer the reference size bias. According to Brass (1979 : 17),the level of

period fertility would be exaggerated if the reference size bias is longer

while the shorter reference size bias will result in the understatement of

the level of period fertility. Reference size error may also cause a

misleading conclusion in studying the trends in fertility between two periods.

For example, if an upwards reference size bias in the earlier period is

followed by a downwards or zero reference size bias in the latter period,

the conclusion of a decline in fertility between the two periods will be

false.

Another kind of event misplacement occurs when the bias in the date of

births of children is associated with the age of mother. The older women

tend to misplace the date of births of children, putting it further back

in time than it actually belongs.

"This error results in an overestimation of the level of fertility for the earliest period preceeding the survey and implies a change in the age pattern owing to a false decline in fertility in young age groups for more recent cohorts" (Brass, 1979 : 17).

Potter (1975 : 13) discusses that there is also a tendency among women

to misplace the initial event closer to the date of interview and to over­

state or underestimate the length of interval between births. These event

misplacement errors depend on whether the questionnaire begins with questions

about the earliest event or the most recent event. According to Potter

(1975 :12-14), if the questionnaire begins with the earliest event and the

respondents have a tendency to misplace the date of birth of the first child

towards the date of interview and exaggerate the length of interval between

births, some of the recent births would be pushed into the future, unless

a constraint is introduced by interviewers or respondents. On the other

hand, if the interview begins with the most recent event and the respond­

ents tend to misallocate it further back in time and overstate the length of

21

inter-live birth intervals, then this would result in the wrong distribution

of the first birth at impossibly young ages of respondents. However, a good

field work can help minimizing such gross errors in reporting events.

In many developing countries, a considerable proportion of dates of

birth of mothers and children as well as the date of death of children cannot

be obtained in months and years. A large proportion of respondents report

the dates in terms of 'years ago'. These dates are estimated in months and

years, either in the field or by imputation when processing the data. This

may introduce complicated misplacement errors in the allocation of events

over time and may distort the real fertility situation.

Misallocation of deaths of children over time, as pointed out earlier,

affects the level and trends in infant and childhood mortality rates. For instance, this error may result in the higher or lower childhood mortality

level than the real situation in the recent period depending on whether a considerable number of reported deaths of children are misallocated towards

or away from the period.

2.3 Description of Data for the StudyThe information obtained from the pregnancy history section of the

individual questionnaire of the Bangladesh Fertility Survey, 1975-76 was the main source of data for studying fertility and childhood mortality. The pregnancy history questionnaire was designed to obtain the information as

reliably and completely as possible. Some filter type questions were asked

at the beginning of the pregnancy history section in order to introduce the

topic and determine the eligibility of the respondents for this section.

Details of all pregnancies experienced by the women were recorded in a

chronological order. The pregnancy history table consisted of two parts: a recording of pregnancies that resulted in live-births followed by a listing

of other births terminating in non-live-births. A single integrated history

of all pregnancies was obtained probing for pregnancies not resulting in

22

l i v e - b i r t h s i n e a c h i n t e r v a l b o u n d e d , a s a p p r o p r i a t e , b y m a r r i a g e and f i r s t

l i v e - b i r t h , f i r s t l i v e - b i r t h and s e c o n d l i v e - b i r t h , an d s o on u n t i l t h e open

i n t e r v a l . T h i s s e g m e n t a l a p p r o a c h was d e s i g n e d t o h e l p t h e r e s p o n d e n t s m in im iz e

memory l a p s e a n d m i s p la c e m e n t i n t h e l o c a t i o n o f e v e n t s o f t h e i r r e p r o d u c t i v e

l i f e . I t i s t o b e n o t e d t h a t t h e p r e g n a n c y h i s t o r y p r o v i d e s i n f o r m a t i o n o f t h e

d a t e s o f b i r t h s i n m o n th s and y e a r s , s e x , s u r v i v a l s t a t u s o f a l l l i v e - b o m

c h i l d r e n , an d a l s o t h e ag e a t d e a t h i n m o n th s and y e a r s o f c h i l d r e n who w e re

r e p o r t e d d e a d . I n c a s e s w h e re t h e r e s p o n d e n t s w e re u n a b l e t o r e p o r t t h e d a t e s

o f e v e n t s i n c a l e n d a r m on ths an d y e a r s , t h e y w e re a s k e d 'How many y e a r s ago d i d

t h e e v e n t o c c u r ? ' A copy o f t h e E n g l i s h v e r s i o n o f t h e p r e g n a n c y h i s t o r y

q u e s t i o n n a i r e h a s b e e n r e p r o d u c e d i n A p p e n d ix A 1 .

The s t a n d a r d r e c o d e f i l e p r e p a r e d a t t h e WFS H e a d q u a r t e r s f rom t h e i n d i v i d u a l

q u e s t i o n n a i r e o f t h e B a n g la d e s h F e r t i l i t y S u r v e y , 1 9 7 5 -7 6 was u s e d f o r o b t a i n i n g

t h e d a t a n e e d e d f o r t h i s s t u d y . The f i l e was c r e a t e d i n t h e s t a n d a r d WFS f o r m a t

f ro m t h e e d i t e d q u e s t i o n n a i r e s a n d t h e o r i g i n a l B a n g la d e s h r e c o d e f i l e . Some

i n c o n s i s t e n c i e s d e t e c t e d i n t h e o r i g i n a l r e c o d e f i l e w e re c o r r e c t e d i n t h e

s t a n d a r d r e c o d e f i l e . T h i s m i g h t s l i g h t l y a f f e c t c o m p a r i s o n o f t h e f i n d i n g s

w i t h t h o s e p r e v i o u s l y p u b l i s h e d . F o r e x a m p le , t h e num ber o f s u c c e s s f u l l y i n t e r ­

v ie w e d u r b a n r e s p o n d e n t s w e re 1 ,4 8 9 a c c o r d i n g t o t h e o r i g i n a l r e c o d e f i l e ( s e e

PCFP D i v i s o n , 1 9 7 8 :3 0 ) w h i l e a c c o r d i n g t o t h e s t a n d a r d r e c o d e f i l e , t h e c o r r e s ­

p o n d i n g num ber o f u r b a n r e s p o n d e n t s w e re 1 ,4 9 7 . The s t a n d a r d r e c o d e f i f e c o m p r i s e d

o f a s i n g l e r e c o r d f o r e a c h r e s p o n d e n t . A d a t a d i c t i o n a r y d e s c r i b i n g e a c h

v a r i a b l e i n t h e s t a n d a r d r e c o d e f i l e can b e fo u n d i n WFS/Tech. 1065 ( 1 9 7 9 ) . The

d a t a d i c t i o n a r y c o n t a i n s l i s t i n g , l o c a t i o n i n t h e f i l e , maximum and minimum c o d e s ,

n o t a p p l i c a b l e / n o t s t a t e d c o d e s w h e re n e e d e d , an d t h e v a l u e l a b e l s f o r a l l t h e

v a r i a b l e s i n t h e f i l e .

From t h e s t a n d a r d r e c o d e f i l e , two s u b f i l e s w e re c r e a t e d f o r t h i s s t u d y .

One was t h e m o t h e r s ' s u b f i l e an d t h e o t h e r was t h e b i r t h s u b f i l e . M o t h e r s '

s u b f i l e c o n t a i n e d v a r i a b l e s V006 (sa m p le w e i g h t ) , V007 ( d a t e o f i n t e r v i e w ) ,

V008 ( d a t e o f b i r t h o f r e s p o n d e n t s ) an d V103 ( e v e r m a r r i e d r e s p o n d e n t s ) .

B i r t h s u b f i l e i n c l u d e d t h e ab o v e v a r i a b l e s an d t h e v a r i a b l e B012 ( d a t e o f

b i r t h o f c h i l d ) , B013 ( s e x o f c h i l d ) and B014 (a q e a t d e a t h o f c h i l d i n

23

completed years). The variables V006, V007, V008 and V103 in both the

subfiles had column specifications as 1-4, 5-8, 9-12 and 13-14 respectively. Those variables remaining in the birth subfile were located in column

numbers 15-18, 19 and 20-21 respectively. Information in the mothers' subfile was taken from the respondent's background and marriage history

sections and the birth subfile contained information from these sections

as well as from the pregnancy history section of the individual questionnaire.

Sample weights were estimated with a view to allow for oversampling

of the urban women. It is to be noted that oversampling in the urban areas

was purposely done to obtain a reasonably large sample required for carrying out a separate study for urban women. Details of the calculations of weighting

factors can be found in PCFP Division (1978 : 31). Weights assigned to the

rural and urban samples were 1.194 and 0.347 respectively. In order to obtain the national estimates of fertility and childhood mortality, these

weights were used. Date of interview, date of births of respondents and their

children were recorded in the standard recode file in the century month code, that is, January 1900 was coded as 1, February 1900 as 2, and so on. Dates

of births of the respondents and the children were imputed in cases where these were not available in months and years. The imputation procedure

assumed that the years of births of the respondents and the children were

known in some form (calendar year stated and month not stated, years ago or

age of respondents at the time of birth or the estimated year by examining

relevant information in the questionnaire where it was not stated). This

procedure for imputation of dates of births of the respondents and the

children in century month form is described in PCFP Division (1978 : 45-46).

From the mothers' subfile, the age distributions of the respondents by

single years were obtained for rural and urban areas as well as for the

total (weighted) sample by taking appropriate weights. The age distribution

24

was derived by subtracting the dates of births of respondents from the date

of interview and dividing by 12; fractions were not taken into account when

compiling age in completed number of years. The single year age distribution

was used to compute the number of person years lived by the female population

for each calendar year from 1941 to the date of interview. Other variables

needed for estimating the levels and trends in fertility were derived by

using the birth subfile. These were the calendar year of birth of child

and the age of mother at birth of child. It is to be noted that the age of

mother at birth of child was computed by subtracting the date of birth of

mother from the date of birth of child and dividing the result by 12.

Fractions were rejected as in the case of age distribution of respondents.

Infant and childhood mortality data were based on the calendar year of birth of child, the age at death of child in completed years and the total number of births in each calendar year. These were derived from the birth subfile. Appropriate weights were used to get rural, urban and national estimates of

infant and childhood mortality. Since the reported number of deaths of children per calendar before 1955 were small, the analysis of infant and childhood mortality was carried out from the year 1955 to the calendar year for which the children were exposed to the risk of dying before the date

of the survey.

The accuracy of the estimates of levels and trends in fertility and

childhood mortality based on pregnancy history data depends on whether

the dates of birth of respondents and their children as well as the ages at

death of children were correctly reported. The accuracy also depends on

whether the respondents remember all the pregnancies that resulted in live-

births. Available evidence suggests that retrospective survey data of WFS

type are subject to errors of various types.

25

"The high standard set by WPS for data collection operation are expected to result in better quality data than typically obtained in the past but this expectation in no way obviates the need for a detailed assessment of the quality of data"(0'Muircheartaigh et al., 1980 : 1).

In a traditional society like Bangladesh, where a vast majority of the

respondents are illiterate and are not calendar-conscious, it is very

difficult to obtain correct reporting of dates of events. Intensive training

was given to the interviewers in the use of probes, local calendar and

historical events with a view to obtain the data as accurately as possible.

Unfortunately, a great majority of the respondents were unable to provide

the dates of events in calendar months and years. Table 2.1 shows that

about 99 and 88 per cent of the respondents could not report their own dates

of birth and the dates of birth of their children,respectively, in calendar

months and years. However, at the time of recoding, the dates stated in

terms of 'years only'and 'years ago' were imputed into calendar months and

years using the WFS imputation program. The imputation of dates affects the

TABLE 2.1

PERCENTAGE DISTRIBUTION OF REPORTING OF DATES OF SOME EVENTS IN THE BANGLADESH FERTILITY SURVEY, 1975-76

Calendar Month and Year

Calendar Year Only

Years Ago Total

Respondent's Date of Birtha 1.4 1.4 97.0 99.8

Date of Last Live-Birth 33.0 4.0 63.0 100.0

Date of All Live-Birth^* 12.0 3.0 85.0 100.0

3 WFS/Tech. 1065 (1979 : 3)

k Chidambaram et al., (1980 : 30).

quality of data in situations where'the proportions of dates of births to be

imputed are high and rise sharply from the recent to more distant past (see

Chidambaram et.al. , 1980 : 31).

A re-interview survey was conducted in May 1976 after the completion of

the main field work. The sample was designed to over-represent urban areas.

A shortened version of the original questionnaire of the Bangladesh Fertility

Survey (Section 1 through 4) was used in the re-interview survey. Out of 430 respondents selected for re-interview, 322 were successfully re-interviewed.

The response rate was much lower than that for the main survey. The unedited

raw-data re-interview file was matched with the partially edited raw-data

main survey file. The results were reported in a draft document by Ahmad

(cited in O'Muircheartaigh et al., 1980 : 8). Based on the data from the

draft document, 0'Muircheartaigh et al., (1980 : 42-43) computed the means

and levels of discrepancies for selected vairables. They found, for

example, that the mean current age of respondents was 28.7 years in both the main survey and the re-interview survey, while about 80 per cent of the

respondents stated their ages differently in the two surveys. It was, however, found that 40 per cent of the respondents stated their ages differently

with respect to 5 year age groups. Also in the dating of first births discrepancies of two or more years were found to have occurred among approximately one-third of the respondents. 0'Muircheartaigh et al.,(1980 : 43) noted that the reporting of age at marriage was substantially more

reliable than that of current age of the respondents.

Marginals of the variables used in this study revealed a number of

biases and inconsistencies. Out of 25,825 reported births, sex of 317 births were not stated. No attempt was made to adjust these 317 births (about

1.2 per cent of total reported births) with sex not stated, and these 317 birthsaffected by the reporting errors were excluded from all analyses. The cases of

births with sex not stated were observed relatively more frequently among theolder rural respondents. Of the remaining 25,508 births, 13,040 were males and

12,468 were females, giving an overall sex ratio at birth of 105 per 100 females.

Available evidence suggests that male births are better remembered than female

births and this has been particularly observed in case of Bangladesh where the

prevalence of a patriarchal system places a higher value on males than females

(Rukanuddin, 1967 : 159). Thus there may be some sex-selectiveness in the

reporting of births. Moreover, some time trend in the sex ratio of reportedbirths might be expected if the reporting is associated with the length ofthe recall period (Sirajeldin et al., 1975 : 213). According to Rukanuddin

(1967 : 158), the sex ratio at birth per 100 females in Bangladesh is expected to lie in the range 102-107. Table 2.2 reveals that there were selective

over-reporting of male births in rural and all Bangladesh, in particular,

beyond 25 years before the date of the survey. The apparent under-reportingof male births in urban areas in the early years might be due to the smaller

number of urban respondents.Some discrepancies were observed in the marginals of date of birth and

age at death of children. In 45 cases, dates of birth of children were

recorded earlier than those of their mothers or beyond the date of interview.

The ages at death of children in completed years were recorded for births at

impossibly young age of the respondents or earlier than their dates of birth

in 41 cases. All these 86 cases of dates of birth of children and the age at death of children affected by the reporting errors were deleted from all analyses. However, deletion of these cases would not affect the conclusions

of this study substantially.TABLE 2.2

SEX RATIO AT BIRTH PER 100 FEMALES FOR BANGLADESH,1941-75

Year Rural Urban Total (Weighted)

1941-45 126 90 1231946-50 125 100 1231951-55 117 106 1161956-60 100 108 1011961-65 103 109 1041966-70 102 106 1031971-75 101 101 101

Source: Pregnancy history data, Bangladesh Fertility Survey, 1975 76.

Although there was no basis for assuming the reported ages of the

respondents were free from errors, no attempt was made to adjust their age distribution . However, a comparison of the 5-year age distribution of the

28

respondents with that of the ever-married female population according to the

1974 Bangladesh Population census is made in Appendix Table B1 . The observed

age distributions of the respondents up to age group 20-24 both in rural and

urban Bangaldeshwere characterised by the relatively older age structure as

compared to those of the census ever married females. In general, the age

structures of the survey and census ever married females did not show much

difference. Nevertheless, it is expected that the age reporting in the

Bangladesh Fertility Survey would be relatively more accurate because an indirect

procedure (establishing age with reference to historical or local important

events) was used by the well-educated and well-trained interviewers to

determine the ages of the respondents who were unable to report their exact

ages.

CHAPTER 3 29

FERTILITY

3.1 INTRODUCTION

Fertility in Bangladesh is high even by the standards of the developing

nations (Census Commission, 1977 : 76). Various nationally representative

estimates of summary measures of fertility (Table 3.1) from different sources

show that there has been no marked change in fertility in Bangladesh during

the period 1901-74. These estimates, though not strictly comparable because

of their different techniques of estimation, indicate that fertility in

Bangladesh has been considerably high. The estimated crude birth rate has remained at a level of around 50 births per thousand population during the

period 1901-61 (Table 3.1). Since 1960, whether there was a real long-term fertility decline remains somewhat controversial. "Most retrospective

fertility surveys document a decline, but methodological constraints preclude from conclusions" (Chen et al., 1979 : 104). The Chandrasekar-Deming estimates of the crude birth rate and total fertility rate were probably high (Ahmed,1970), while those based on the pregnancy history data from the National

Impact Survey were low, which might possibly be due to under-reporting of births (Sirageldin et al.,1975 : 214-215). Recent decline in fertility has

been documented from the longitudinal vital registration data collected by

the former Cholera Research Laboratory in rural Matlab Bazar Thana of

Comilla district, Bangladesh. A modest decline in crude birth rate was observed in 1971-72, one year following the 1971 war of liberation. The birth rate

fell again in 1974-75, the year of the 1974 famine and declined sharply one

year after the famine (Curlin et al., 1976 : 91; Chen and Chowdhury, 1977 :

411-412; Chowdhury and Curlin, 1979 : 117; Chen et al., 1979 : 103).

There are two approaches to the measurement of fertility using the pregnancy

histories. These are the cross-sectional or period approach and the cohort or

generation approach. The former refers to the measurement of fertility by the

SUMM

ARY

MEAS

URES

OF

FERT

ILIT

Y FO

R BA

NGLA

DESH

, 19

01-7

4

30

CoPPUPtoopft0)ftcnU)op0

>ip-HI—I■HPP01ftPrdPOEh

X-PP•Hm01X!ppu

On on on1 1 O IDi—1 P CN TO —.Cn on *• .. r- 01 ft.. .. lo r- cn p 0i—) i—i r- r- •• cn fton On on p —on on p P C" •n,—, i—i P — on TO Cont"- ^ .0>c ci rd 0pi—I b c 1 0 01 p. • rd P P X rd• • i—1 i—1 to • P grd rd P C/1 P Prd 01 P rd Ti PCO P P E G cnl-' 01 01 G g P rd fton P O 01i—1 P P TO 0 rd X •-— P P P P P P inft1 ft1 01 cn g rd 2 0rd O O cn P rd TO O iG O 0 rd cn 43 P CNrd P P P C 43 01 0 0X rd (d P 01 rd Cn onft ft ft 0 0 ft rd G POcn P vP P pcn rd GG P 0101 P g0 ft P

I I I I I I

I I I I I I

co on -sr ^ co oO M O N O I—I O in m m in ^ in in

X0Pp01 ft

oOnr—I

<—I CNIr—!POnP

co co m rn* *

^ n O p iD

h ID ID > ID■K *

O O O co P’o o m co h in ^ ' j

rd0 0 00 0 0 • •1 1 1 cn

CN (N 0 I f ^ 010 0 0 r-' t'' pOn On on On On 0i—1 P i—1 P P 2

TO01PcnP

•r~iTOf00)XPg0ppTOd)cprd-p4300)P0)2rHC OIPino nPXOnP0PxPrHrH1rH0O ni—Icn0)TOrdÜ(DTO01 X PP0Pcn01-pid(4

TO0 •H P01 f t

01X4-)P0p01 p cd ft

PJ£01gpp01ftXw

p01ftXw

gopPidg•HpcnftXp2op0£OPPrdPdft0ft01 x p

g0pp01 p cd g p P cn 01

On£pg01QIpcdX01cnrdpTO£rdX0 Lo

ngit

udin

al r

egis

trat

ion

esti

mate

s fr

om t

he P

opul

atio

n Gr

owth

Est

imat

ion

(PGE)

Expe

rime

nt,

1962

-65.

De

rive

d fr

om t

he N

atio

nal

Impa

ct S

urve

y (N

IS), 19

68-6

9.De

rive

d fr

om t

he B

angl

ades

h Re

tros

pect

ive

Surv

ey o

f Fe

rtil

ity

and

Mort

alit

y (B

RSFM

), 19

74.

Deri

ved

from

the

19

74 B

angl

ades

h Po

pula

tion

Cen

sus.

J1

number of live-births to women of different ages during a reference period and

the latter involves the measurement of fertility by the cumulative number of

live-births to a group of women with common characteristics (born at a given

time or married at a given time) as they progress through their reproductive life span. The cross-sectional approach is employed to study the levels

of fertility at a given time while the cohort approach gives no indication

of current levels of fertility. Both approaches are used to study trends

in fertility. Since the cohort approach takes into consideration the

same group of women through reproductive life from menarche to menopause, it is

found to be superior to the cross-sectional approach in studying the past trends

in fertility (PCFP Division, 1978 : 62). However, if there has been no change in fertility during the past 30-40 years, the cross-sectional fertility can be

translated into the cohort fertility.

This chapter attempts to analyse the cross-sectional and cohort fertility based on pregnancy history data from the Bangladesh Fertility Survey, 1975-76.

The purpose is to find out the levels, trends and patterns of fertility in Bangladesh. This chapter also includes a brief description of some determinants of fertility. Effects of possible sources of errors on the estimates of fertility will also be discussed. Finally, some of the suggested adjustment

procedures for fertility will be reviewed in this chapter.

3.2 SOME DETERMINANTS OF FERTILITY

Freedman (1961-62 : 38) points out that "the variables that influence human fertility are numerous and complex in their intractions". Davis and

Blake (1956 : 211-231) have proposed a set of eleven variables, called

intermediate variables that directly affect fertility. They suggested that

these intermediate variables are related to the three steps of the process of

reproduction :

(1) exposure to intercourse affected mainly by nuptiality and

duration of marriage;

32

(2) exposure to conception affected mainly by use of contraception;

and (3) gestation and successful parturition affected by foetal

mortality from involuntary and voluntary causes.

Although the Davis and Blake framework has found wide acceptance, quantification

of the link between a set of intermediate variables and fertility has proved

difficult. For simple quantification, the eleven intermediate variables have

been modified into seven factors, namely, (1) proportion married among females,

(2) contraceptive use, (3) prevalence of induced abortion, (4) duration of

postpartum infecundability, (5) fecundability (or frequency of intercourse),

(6) spontaneous intrauterine mortality rate, and (7) prevalance of permanent

sterility (Bongaarts, 1980 : 2). Each of these factors influences fertility

directly and together they determine the levels of fertility. Only the first

three factors will be discussed briefly in this Section as the empirical data

for the remaining four factors are not available in the context of Bangladesh.

Proportion Married Among Females: In Bangladesh the incidence of premarital

conception has been rare and most females remain in a married state during

the reproductive period. Shahidullah (1979 : 23) found that about 87 and 82

per cent of the women aged 10-49 were currently married according to the 1974

Population Census and the Bangladesh Fertility Survey, 1975-76 respectively.

Here the proportion ever-married among females will be discussed. Table 3.3

shows that, in general, there has been a gradual change in the proportion of

ever married women, particularly below age 20 during the period 1951-76. It

is revealing that the age at marriage has been rising in Bangladesh. Also it

is clear that most marriages in Bangladesh take place at an early age. About

99 per cent of the females below age 30 were ever married and the short span

time during which most marriages occurred can be noted from Table 3.2. The

proportions of ever-married women in various 5 year age groups of women were

higher in rural areas than those in urban areas as found in the 1974 Population Censu:

and the Bangladesh Fertility Survey, 1975-76. It should be pointed out that the

IN

BA

NG

LA

DE

SH,

19

51

-76

33

2H

g

§ftO

ftOc

2wuftwft

Cb

oPo(DCTOC

in CO f ' lO in i—1 on i—1 00 f t i—iro • • • • • • • • • • i1 cn cn cn cn on on cn cn cn cn M4 in

O cn on m cn cn cn cn cn On cn O Hm 1—1

i>iP ••>P cn

cn cn in 00 cn CM On H CM CO o P O'CM . • • • • • • • CO cni CO cn on 00 cn O' cn cn CO cn H

in on cn cn on cn cn cn cn cn cn p ’CM >

•H X4-1 pO HP r—1

o O' ■r m cn 00 CO cn p* Cb PCM . . • . • • • • • • cn X!i O' 00 CO in O' On lO CO cn m 0 •r4

o cn cn cn cn on CO cn cn 00 cn p XCM 4-1 p

P Xf t CO

X — £cn r - O' o O' CO in CM On CM cn *x oi—i . • . . . • • • • ■ p o On Pi CO i—i O' O' O' in m p* CO o X • 1 44

in 00 cn 00 lO O' in O' O' in O' P o CO•—i H o cn X

cn fH Pp ..P >i PCQ rH O' Cb

p* m *o o o o CO in o Ml* 00 p r - £•—i • • . • • • • • • . • • 4J On Oi CO CM CO CO o cn CM O' CO P H O

o CM rn 1—1 H H 2 E —

•—I rtf Vi P f t

PrtfXPD

r—IItfPPft

PP

XIPD

rHP4->oEi

inO'»rH

rHX)(nI—I

cnPWppu

inCOOnI—I

O' HOn OnH H

X)r -i

inO'o nI—I

L•HHP4-1POsX!pp

> 14->•HH•HVPP

ftaDQ

-PPPe•HppCbXftpo•H4->pe

Xipp

coH

•HX0pCbCbpppp•xTm1

oro

Cb >» P

COe'­enH

Pu

•H w •P X in cn p

•h p o4-1 X) -H

<n P P in4-1 1—I H CO tn >

•• P H*4-4 p Q

r o o mcn P P H P P — p XI

P p p p p O CQ

•H TO W X P

•H CO 4-1W f t 4->

P O'»

>1 p

x h-p -H

•H X) P 0

4-1O P P P X) CO f t

P Xo in

•H

cn p > p

•r| t f HQ P P

rH HCn P P P PCQ P

O£ 44O P

ft ftoft£op

<44 144

ftftuftBop

Ip

XP4->PCb£0U

PPP

P P 4J t f 4J P P P

H H P P Cn O Cb P P O P P f t CQ f t

X) X P P 4-1 4-1 4-> P P P P Cb Cb U E £ P O O P U U f t

U) p

rä p

f t CO P U f t f t CQ

* X o

Bangladesh Fertility Survey household data were not edited (Ruzicka, 1980).

Hence the proportion ever married among females found from the 1974 Bangladesh

Population Census data will be used in this investigation.

Contraceptive Use: The use of contraception among couples is very low in

Bangladesh. According to the Bangladesh Fertility Survey, 1975-76, only

about 14 per cent of ever married females reported to have ever used any methodof contraception while 10 per cent had ever used any 'efficient''*" method (PCFP Division, 1978 : 77). Ever-use of contraception was the highest

(20 per cent) among women aged 35-39, and about 12 per cent of rural

women reported to have ever used any methods of contraception compared

with 28 per cent of urban women (PCFP Division, 1978 : 77-80). According

to the Bangladesh Fertility Survey, 1975-76, both husband's and wife's

education were positively associated with the use of contraception. About

43 per cent of the respondents with husband's education above secondary levelwere the ever users of contraception while only about 10 per cent of therespondents with husband's no formal education ever used any methods ofcontraception (PCFP Division, 1978 : 80-81). Sirageldin et al., (1975 : 7)found from the data collected by the National Impact Survey, 1968-69 that

only 6.4 per cent of the married couples had ever used any contraceptivedevice. Although the findings of the two nationally representative surveys

are not totally comparable because of differences in methodology and quality,

it appears that contraceptive use in Bangladesh has increased over the period1968-76. Nevertheless, despite the introduction of the National Family

Planning Program by the Government in 1965, this increase in the use ofcontraception is not quite encouraging. This Program had a marked rural focus

and "Its basic strategy was to develop an intensive infrastructure of family

planning personnel and facilities in order to make family planning information,services and contraceptive supplies readily available to the population"

(Sirageldin et al., 1975 : 2).Prevalence of Induced Abortion: Studies concerning the prevalence of induced

abortion are rare in Bangladesh. The practice of abortion is not permissible except for saving the life of the expectant mother according to the law in the

country. Self-induced or other cases of abortion are treated as a criminal

1 The methods defined as efficient were: pill, interuterine device (IUD),condom, sterilization and use of foam tablets, ielly and rroann

35

offence. Socio-religious customs also place a stigma on abortion. "Despite

these restrictions, there are indications that abortion is practiced in

Bangladesh. ... the Government's current national population policy has hinted

towards the liberalization of the law on abortion" (PCFP Division, 1978 : 81-82).

According to the Bangladesh Fertility Survey, 1975-76, only 2.5 per cent

of all pregnancy losses were reported to have resulted in an induced abortion,

and 4.3 per cent of all pregnancies terminated at less than 7 months gestation

were due to an induced abortion. The actual incidence of abortion was probably

under-reported (PCFP Division, 1978 : 82).

3.3 LIFE-TIME FERTILITY

The mean number of children born alive (mean parity) to women in the

reproductive ages is usually employed as an index of life-time fertility. Table3.3 compares the mean number of live births per ever married woman obtained from

various nationally representative sources in Bangladesh during the period 1961-76.

It is to be noted that the mean parities reported in the Bangladesh Fertility

Survey, 1975-76 was uniformly higher than that found in other sources forall age groups of ever married women. Although the estimates of mean parity found ii

the 1961 and 1974 censuses were almost similar, these were probably low because

the census data in Bangladesh have suffered from various reporting errors

(Khan, 1973a: 175). The mean parities reported in the Bangladesh Retrospective Survey of Fertility and Mortality, 1974 was higher than that in the 1961 and 1974 censuses for all ever married women aged above 24. Available

evidence suggests that the data on the mean number of children ever born derived

from detailed pregnancy histories are comparatively more reliable than those

obtained by single retrospective questions (Mukerjee, 1975 : 141). Thus it is expected that the mean parity distribution obtained from the pregnancy history

data in the Bangladesh Fertility Survey was of better quality as compared to

that found from other investigations in Bangladesh. Moreover, the data were

collected by a segmental approach with detailed probing for events in each

MEAN

NUMBE

R OF C

HILDREN

BORN A

LIVE P

ER E

VER-MARRIED

WOMEN

IN B

ANGLADESH, 1

961-1976

36

cnIinh *

Io

aP0Sho

3Cn<

CnroIinro

roIoro

COCMIinCM

CMIOCM0> I—IIini— I

I— iIoI—I

uIT)a)>h\0)uMP0CQ

31 in rH m 00 O ro CD CDr~ CO rH m 00 CD G> 3 cn 3• ■ • • 3 • • 3 .— , 44m in CD CD — - CD — ■ CD — CD 3CD 443•• 3.— . /—V .— . UCTi o i—l 00 3 <3 CO in in r"- 43CM <n CO 3 ro cn 3 t-*• • • • 3 • rH i CD cn Gin in CD CD •— CD —' CD — 3 •H3

c 3,—v 0 Unr CTi CO 3 CTi 3 rH 3 3 •H PCM 3* CO CD rH 3 LO CD 3 3 Cn• • • • in • rH • CD 3 •Hm in m CD v—' CD CD '—" •H Cm

0 ,—. .— . u 3nr CO 3 O 3 3 CT> C3 3 4-1CD in cn c- o in O CD cn 3 3t • • • CD • CM • 3 3 33nr 3* 3. in — ' in '—" in — 3

c >i3 ShU 0*-s .—s 44i—i CTi CM cd 3 in v£> 3 00 • • 3in CM m CM in Cn U0 3 o •H• • • • CO • CM • rH 3 rGCO CO CO 3 ~—- 3 3 1—1 3— cn Chr—i uc-—. -—. 3nT 1—1 3 in cn t"~ CM U0 3 Ch GCM CTi Cn 3 CM 3 3 3 4-1 Cn• • • t o • ro • 3 •H 3CM i—1 3 CM i—i CM ^ ' 3 i—1 i—1 S4— — 3 0444u e.—. .— . O 0r-' C" CO CM O in in S ur- CD in CO 3 O vD CO 3 <H• • • • CTi • CM • o 33o O o o — - i—1 1— ’ o rH G 33'— ' 3 344>i P4-) a.3 m O i—1 00 i—1 •H e •

O i—I CM o t—l 0 Gi 1 i • CM • • 3 •H u 3— s— — 44 eU •• 03 £Cn CDr'' 3344 i 30 in •Hr- H>i cn Sh3 rH 3> Eu 1p U33 3 CO 3 33 3 > >-3 3 Sh 34G • • > P

Cn •H CO 44■H cn 44 03 3 U £>H. cn 3 44 )H■—' i—i Cl< -H 3— 3 rH n

r—1 3 <—1 0 •H3 3 3 - Sm 44 3Sh rQ 4-) • 44 Sh G3 3 O 3 rH 3 3CQ D Eh rH 3 Dh Cm 3

rQ 43 4-> 43 43 44H 3 3 3CD •H 3 3 Or- 3 •H 3) 33 44

i > g 3 3i—i 3 3 in 3 3 rH t-1 ShCD r- e'­ A tn Cn 3cn (3i cn en 44 A G C 44rH i—1 . i—l 3 0 3 3 3 3

33333u

2; « CQ CQ sh I * 3 4Q

37

segment of the reproductive life of a woman (see Chapter 2, Section 2.3). The

higher national (weighted) estimates of the mean parity reported in the

Bangladesh Fertility Survey, 1975-76 as compared to those in the 1974 Census

of Bangladesh and in the 1974 Retrospective Survey of Fertility and Mortality

might be accounted for by the better coverage of live-births (PCFP Division,1978 : 63).

. The estimates of mean parity for different age groups of women in the rural areas

were generally higher than those in the urban areas of Bangladesh (Table 3.3).

The relatively higher mean parity for urban ever married women below age 20

may be attributed partly to the parity selectivity arising out of sampling

variations. The completed family size in the rural areas appears to be nearly

the same as that in the urban areas (Table 3.3). However, the completed family

size in the rural areas seems to be an underestimate of the actual fertility performance. The lower mean parity of the rural women in the age group 45-49 as compared to that in the age group 40-44 might be attributed to the following possible causes: (1) the women in the age group 45-49 might have failed toreport some of their children due to memory lapse; (2) the women aged 45-49 years might have experienced a lower fertility; (3) there might have been age misreporting among the older women; and (4) there might have been a selective

survival of women in the age group 45-49 with relatively few live-born children.

3.4 CURRENT LEVELS OF FERTILITY

Current levels of fertility can be studied by age specific fertility rates

and total fertility rates based on the period approach for the period of 5 years

preceding the survey. Table 3.4 presents age specific fertility rates and

total fertility rates for rural and urban areas as well as for the total weighted

sample by single years from 1971 to 1975. In order to minimize the effects of

sampling variation and to examine the overall levels of current fertility, the rates were also computed for the 5 year period 1971-75.

38

It can be noted from Table 3.4 that the age specific fertility rates for urban women were generally lower than those for their rural counterparts

in all age groups as revealed by the 5 year (1971-75) average rates. Relatively

higher age at marriage for urban women (see Section 3.2) may partly account for

this. Hbwever, a considerable departure from the higher rural fertility situation was observed in the case of single year rates, particularly in the

age group 20-24. Fertility of urban women in the age group 15-19 in different

single years from 1971 to 1975 was found to be lower which was in contradiction

with the corresponding finding from the data on the mean number of children

ever born. The total fertility rates for rural women were consistently higher

than those for urban women in all the single years from 1971 to 1975. It is

to be noted that the total fertility rate for Bangladesh has declined from

7.3 to 5.2 over the period of 5 years from 1971 to 1975. The decline, though not regular is found from year to year.

The marked decline in fertility from 1971 to 1972 may be explained by a

number of factors. The 1971 social disruption caused by the war of liberation affected the reproductive performance of couples considerably. Due to fear, anxiety or the desire to postpone pregnancy coital frequency might have been

reduced. A large exodus of about 16 per cent of the population (Curlin et al.,

1976 : 88) to neighbouring India as refugees might also have contributed to the reduction in conception either by voluntary or involuntary separation of

spouses. As a result of the postponement of marriages during the period of

social disruption, the number of persons entering reproductive life might have

diminished. Reduction in the number of births.one year after the war of

liberation might be the reflection of the reduced number of conceptions during

the period of war (Chen and Chowdhury, 1977 : 422). A similar decline in

fertility one year after the period of liberation war was also found from the

reasonably complete vital registration data in Matlab Thana (Curlin et al.,

1976 : 94; Chen and Chowdhury, 1977 : 410-411). Although this finding was

of

Wom

en

39

CDtn<

>14-1•HI—I •P 4-> P CD Ua

rHo4-) O Eh

04orI

LT>oT

ororIoor

n ro

O' d o-

O (M O l

o in ino o 04

o in ini f l o ID

in in inn o in in m m

o rrffi id in in

no o CO CN O i—I 0 0 . 0

^ in in <D ID <D O O O

no o CO (N O i-H O O O

O O O O O O O O O

uo o inrH O rHo o o

^ in id O rH O O O O

CO ID COo o o o o o

o <d o in no in o o o

o no O in or or o o o

D CN <D no ^ no O O O

n o n no ^ no O O O

ID ID ID or or or o o o

in O' ro ID no or rH CO04 in t—1 D 00 in in H CNUO04 0) 10 >44no 00 O' 04 no o no 04 00 04 rH O' o 04 or 04 no 04 04 > 101 rH rH o O o o o o H QJ rd ■P

in dno P QJ rd

0 P u<44 fd •H

>4410 • H '•j IDQJ 10 O' £ 0-

or O' COCO ID CO 1--1 ro o uo D 04 ro or H or or 04 4-1 d 04 S' 1ro in COin »—1 ID rH o 04 o rH 00 H 00 in CO H 04 H id to H mi <04 04 04 04 rH 04 04 rH 04 04 H 04 H H 1—1 04 H Ol p d 10 [■».o Q) d „ 04no >1 O •p ^ rH

4-1•P d d OH 0 0) ^ >1■H•p h rr-, ^4-1 4-1 0 TJ >

O in 04 04 00 ID or O' ro <D or i—1 ro o no o H ID H P id $ ?s HCN CNO' 04 in O ID COO' 00 or ro or in uo in o- ID O' QJ H d1 no 04 no 04 no 04 CM04 04 04 CNCN 04 04 04 04 04 04 <44 d <44 d c/)in 04 O rp04 H o r! ^fd cu 04 U 4J

4-1 or ^ -p•r| x i $ rPP 10 in O -Pfd QJ or - 4-1

or CO04 CO 04 CDo ID no CO o COH CO(N no in O- in g TJ m UCN uo D ID rH rH 04 or rH o H o O' ro 00 H O H id cui no no no no 04 ro ro 04 ro ro ro ro 04 ro CN ro ro ro o H d £ Pmo •H 04 0 D)04 <4-1 d p SC

•H id 04 too CQ > QJQJ QJ S T!D a or 04 S~ rdto O' rd n H

o 00 1—1 O UOro O 04 COID in O' O 04 COor or O- CO 041—1 04 in 04 COor 04 O or 04 04 or 04 04 or 04 o or 04 QJ H QJ m1 IN rH 04 rH i—1 rH 04 rH rH H H ■—1 1-1 H H 04 •—1 H 04 SC % ^

in id QJ 4-1 DQrH SC P

O' 4-1 Pd 0•rl d <44 4-1

-P •'j id1--1 to rH rj-f

or in no in in 04 or or or or O' or ID <DID <D in ro in D a ts QJ •Hi—1 o O o o O o o o o o o O O O O O O o •p d 4-1 £ >4i o o o o O o o o o o o O O O O O o o 4-1 d fd ^ po H 0 P 0t—1 d <44 >p +j

g TJ r> 10TS QJ S -p

>< QJ 4-1 If1, scS3 •P QJ <44

P TJ M >4n P QJ O u uX—V QJ id 4-1 fd S d3 3 3 3 3 3 4-1 g SC P

fd ---- — d O' S3 co c<D D a P •H O'P i—1 d rH rH d i—1 rH d rH rH d 1—1 H d H H d H g QJ Q) Q)

rd fd rd td rd fd rd fd id fd fd fd fd fd fd fd fd fd 0 > 3 SC S' UP X 4-) P S 3 4-1 p S3 4-1 p S3 4-1 p S3 4-1 p S3 4-1 o QJ Eh C cud p n d P n d P O d p n d P 0 d P 0 n

PC D Eh PC D Eh PC D EH PC D Eh PC D H PC D gHH 3

in • •o- •• QJl 10 O

p rH 04 ro or m i—1 QJ Prd O' O' 4-1 dcd 04 04 04 04 04 04 0 0>n rH rH rH H H H 3 in

co•H4-1P0oopD a

(1)X4-1

X4-1•Hscqjgo

TJ0•rlPPfdg

CNO'04H

d•rl

<DinoX4-1

inid

<D

§io

<1)rC4->

(1)XO4-1

TJQJ

CO•rl4-1dQ)>CoorHid

0d

qjX4-1

tmd

•HS0HrHo

144

inIDPdo>•rl

40

limited to a relatively smaller area and might not be representative of all

Bangladesh, it might be taken to support the evidence of the marked decline in

fertility one year after the war of liberation.

The post-conflict 'compensatory' rise in fertility was found in Matlab

Thana in 1973 resulting from the desire to replace the lost children and the

clustering of the postponed marriages (Chen and Chowdhury, 1977 : 422-423).

However, the Bangladesh Fertility Survey data did not show any such rise in

fertility. Event misplacement in pregnancy history data produces an exaggerated

decline towards the date of survey (Potter, 1977a : 356). This is found to be

more pronounced one year before the date of survey (see for example McDonald

et al., 1976 : 12-13; Asare, 1980 ; 49). The apparent decline in fertility in Bangladesh in the year 1974 may possibly be accounted for by this.

The 1974 famine in Bangladesh caused by severe monsoon flooding which

destroyed most of the standing crops might have been the reason for the highest decline in fertility in the year 1975. Moreover, "...the shortage in availability of food probably decreased the nutritional status of women which

caused a reduction in fecundity and lengthen the period of temporary

infertility" (Chowdhury and Curlin, 1979 : 114). Chen and Chowdhury (1977 :

410-411) also found a. sharp decline in fertility in Matlab Thana one year

after the 1974 famine which may be taken to support the decline in fertility

in Bangladesh in the year 1975.

The total fertility rate for Bangladesh during the period 1971-75 was

found to be 5.9 which was not out of line with those of 6.3 (longitudinal

registration estimate) in 1963-65 obtained in the Population Growth Estimation

Experiment (Farooqui et al., 1971 : 93) and of 6.0 in 1966-68 derived from the

National Impact Survey data (Sirageldin et al., 1975 : 209). However, the estimate of total fertility for the year 1974 in this study was considerably

lower than that of 7.1 obtained from the 1974 Retrospective Survey of Fertility

41

and Mortality data using the Coale- Trussell fertility model based on the

assumption of constant fertility in the past (Census Commission, 1977 : 76).

This assumption though generally valid for Bangladesh might be less realistic

for recent years (PCFP Division, 1978 : 74).

3.5 TRENDS. IN FERTILITY

The age specific marital fertility rates for ever married women are used

in' this section in order to study the trends in fertility. These rates for the

10 year period preceeding the date of survey are presented in Table 3.5. In

order to minimize the effects of sampling variations the rates have been computed for the selected periods 1966-68, 1969-71 and 1972-75.

The total marital fertility rate for ever married women has declined from

8.7 to 6.2 in Bangladesh during the 10 year period between 1966 and 1975. A modest decline in fertility was observed between the periods 1966-68 and

1969-71 whereas a rapid decline was noticed between the periods 1969-71 and 1972-75. As pointed out in Section 3.4, the marked decline in fertility between the periods 1969-71 and 1972-75 might have been due to a combination of factors - such as the social disruptions caused by the 1971 war of liberation and the

1974 famine. Total marital fertility rates for ever married rural women were higher than those for their urban counterparts in all the selected periods

between 1966 and 1975 (Table 3.5). It is to be noted that the decline in total

marital fertility in the urban areas was slower when compared to that in the

rural areas. This feature is difficult to interpret without further investig­

ations .

Fertility decline between the periods 1969-71 and 1972-75 was observed

for all age groups of ever married women above age 14 both in the rural and urban

areas as well as for all Bangladesh. The percentage decline in marital fertility

rates was found to increase with the increase in age of women between the periods 1969-71 and 1972-75. Overall, the total marital fertility declined by about

TABLE 3.542

AGE SPECIFIC MARITAL FERTILITY RATES1 FOR EVER-MARRIED WOMEN AND CHANGES

IN THESE RATES FOR SELECTED YEARS BETWEEN 1966 AND 1975 IN BANGLADESH

Age Period Percentage Change

1966-68 1969-71 1972-75 1966--68 to 1969-71 1969- 71 to 1972-75 1966- 68 to 1972-75

Rural Urban Total(W)

Rural Urban Total(W)

Rural Urban Total(W)

Rural Urban Total(W)

Rural Urban Total(W)

Rural Urban Total(W)

10-14 .052 .058 .053 .053 .057 .053 .051 .070 .053 + 1.9 - 1.7 0.0 - 3.8 +22.8 0.0 - 1.9 +20.7 0.0

15-19 . 331 .259 .325 . 306 .282 . 304 .254 .265 .255 - 7.6 + 8.9 - 6.5 -17.0 - 6.0 -16.1 -23.3 + 2.3 -16.1

20-24 . 385 .353 . 382 .386 .385 .384 .312 . 326 .313 + 0.3 + 1.4 + 0.5 -19.2 - 8.9 -18.5 -19.0 - 7.6 -18.1

25-29 . 360 .336 .385 .336 .311 .334 .261 .271 .262 - 6.7 - 7.4 - 6.7 -22.3 -12.9 -21.6 -27.5 -19.3 -26.8

30-34 .307 .277 . 305 .273 .268 .272 .205 .176 .203 -11.1 - 3.2 -10.8 -24.9 -34.3 -25.4 -33.2 -36.5 -33.4

35-39 .193 .147 .191 .180 .153 .179 .117 .078 .113 - 6.7 + 4.1 - 6.3 -35.0 -49.0 -36.9 -39.4 -46.9 -40.8

40-44 (.119) (.114) (.119) .079 .071 .078 .042 .043 .042 -33.6 -37.7 -34.5 -48.8 -39.4 -46.2 -64.7 -62.3 -64.7

TotalMaritalFertilityfor ever-MarriedWomen(10-44) 8.7 7.7 8.7 8.1 7.5 8.0 6.2 6.1 6.2 - 7.7 - 2.8 - 7.4 -23.0 -18.1 -22.6 -28.9 -20.4 -28.4

Notes: W = Weighted.1 Marital Fertility rates for ever married women rather than for currently married women have been calculated because about

90 per cent of both the rural and urban respondents (ever-married) have spent their reproductive lives in married state (Shalidullah, 1977 : 77-78). However, marital fertility rates for currently married women can be calculated by using the following table of the proportion of respondents who are currently married by age:

Age Group10-14 15-19 20-24 25-29 30-34 35-39 40-44

Rural .92 .92 .95 .93 .90 .84 .77Urban .95 .94 .89 .92 .87 .80 .76Total .92 .93 .95 .93 .90 .84 .77

Source: Bangladesh Fertility Survey, 1975 (PCFP Division 1978

Source:

Bracketed rates are based on two-year average 1967 and 1968.

Pregnancy history data, Bangladesh Fertility Survey, 1935-76.

43

29 per cent in the rural areas and about 20 per cent in the urban areas during the decade before the survey.

The age specific marital fertility rates in 5 year periods for the 35 years preceeding the date of survey for ever married women are presented in Table 3.6.

The table provides both the cross-sectional and birth cohort fertility rates

as explained in Chapter 2 (Section 2.2). In the absence of errors due to age misreporting, omission and misplacement of events, change in fertility over a

long period of time can also be studied from the table. It can be observed

that both the cross-sectional and birth cohort marital fertility for ever married

women in Bangladesh follow generally the same pattern as those for ever married

women in the rural areas. The cross-sectional marital fertility for the rural

women has shown a sharp decline in the 0-5 years before the survey as found earlier.

However, the cross-sectional fertility for the urban ever married women in all age groups has declined in the 10 year period before the survey.Table 3.6 also shows that cross-sectional marital fertility for rural women

was the highest except for the women aged 30-34 years at interview during the period 5-10 years before the survey. On the other hand, the cross-sectional

marital fertility for urban women was the highest in the period 10-15 years

prior to the survey except for the women aged 25-29 and 30-34 years at interview. Probably due to reporting errors, cross-sectional fertility for ever

married women could not reveal a clear time trend, particularly beyond 15 years

before the survey.

Both for rural and urban Bangladesh, the cohort aged 40-44 at interview

(women born in the period 1931-35) had the lowest fertility in all age groups

except for the age groups 30-34 in the rural areas and 20-24 in the urban areas

during the period 5 to 25 years before the survey. This feature seems to be

implausible as there is no external evidence that fertility contributed partly

TABLE 3.6 44

AGE-COHORT AND CROSS-SECTIONAL AGE SPECIFIC MARITAL

FERTILITY RATES FOR EVER-MARRIED WOMEN IN BANGLADESH

.008

.046.099• Age Specific Marital Fertility Rates for Rural Women .133.

. 297 . 215.242 .278

. 352 . 273 /.297 . 342. 312,

. 341. 317 . 323. 312. . 309,

.236, . 248, .262.283. . 32 3.

before the SurveyYears

.006

Exact Age of Women at the Time of Survey

Exact Age of Women at the Time of Survey

45

by this cohort of ever married women was lower in the period 5 to 25 years

prior to the survey. On the other hand, for the cohort born in the period

1926-30 (aged 45-49 at interview) marital fertility was highest in the urban

areas for all but the age groups 15-19 and 20-24 while in the rural areas

fertility was highest in the age group 35-39. This suggests that there was

a substantial amount of reporting errors by the older women, particularly

those from rural areas. It can be noted from Table 3.6 that for the three

oldest cohorts (aged 45-49, 40-44 and 35-39 years at interview) both in the

rural and urban areas, the younger the cohort the lower was the fertility for

women only in the age groups 35-39 and 40-44.

A comparison of adjacent age cohorts to detect the change in marital fertility reveals that in the 35 years before the date of survey the rural women in the age group 15-19 belonging to the youngest cohort (aged 15-19 at interview) have experienced a lower fertility. For other age cohorts in the age group 15-19, the older the cohort the lower has been the fertility for the rural women. This feature seems to be quite contradictory to what one would

expect. The apparent lower marital fertility in the age group 15-19 for various cohorts other than the youngest cohort indicates a considerable amount of response errors, especially by the older respondents in the rural areas.

A similar feature was also observed in the case of urban respondents except

for the cohorts aged 20-24 and 30-34 years at the time of interview. Table 3.6 shows that in the different 5 year periods before the survey, the marital

fertility rates for various age cohorts of rural women were generally higher

than those for their urban counterparts with the exception of fertility rates

for various age cohorts in the age groups 20-24 and 25-29.

Table 3.7 presents the birth cohort and cross-sectional total marital

fertility rates for ever married women aged 15-44 years. It is worthwhile to

TABLE 3.7 46

TOTAL MARITAL FERTILITY RATES FOR EVER-MARRIED WOMEN AGED 15-44 YEARS

Birth Cohort Cross-■Section

Year of Birth Area Total Marital Fertility Rates

(15-44)

Period Area Total Marital Fertility Rates

(15-44)

Rural 7.0 Rural 8.21926-30 Urban 7.7 1966-70 Urban 7.2

Total (W) 7.1 Total (W) 8.1

Rural 6.8 Rural 6.3

1931-35 Urban 6.6 1971-75 Urban 6.1Total (W) 6.8 Total (W) 6.2

W = Weighted

Source: Computed from Table 3.6

note that total marital fertility rates for the 1926-30 birth cohort were higher

than those for the 1931-35 birth cohort in the rural and urban areas as well as

for total Bangladesh. Total marital fertility rates for both the cohorts compare well with the mean number of children born to ever married women aged 40-44 years

except for the 1926-30 birth cohort in the urban areas. The apparent higher total marital fertility rate for the 1926-30 birth cohort in the urban areas might

possibly be due to the higher fertility reported by this cohort at the ages 25-29 and 30-34 years (see Table 3.6). The cross-sectional total marital fertility for the

rural women in the period 5-10 years before the date of survey also seems to be

an overestimate of the actual fertility situation. This may be explained by the

tendency of pushing the earlier births forward or the most recent birth backwards

in the 5-10 year period before the date of survey. The plausible reasons for the

decline in cross-sectional marital fertility in Bangladesh during the period of

47

5 years preceeding the survey were explained earlier in Section 3.4.

3.6 AGE PATTERNS OF FERTILITY OBSERVED IN VARIOUS STUDIES

The relative contributions of women in each age group of the reproductive

span to total fertility in Bangladesh according to the Bangladesh Fertility

Survey and other nationally representative sources are shown in Table 3.8. It

can be observed from the table that the shapes of the majority of the fertility

schedules are characterized by the peak at age group 20-24. For the remaining

few fertility schedules the peak is found at the age group 25-29. The contrib­

utions to total fertility of about 12-21 per cent by the age group 15-19 from

different sources might be explained by the early age at marriage. On the other

hand, women aged above 39 years contribute only 4 to 8 per cent to the total fertility.

The patterns of various fertility schedules can be examined by comparing the relative shares of the total fertility in the various age groups of the reproductive period of women. Pavlik (Cited in Nassef and Gaafar, 1971 : 17; Kpedekpo, 1976 : 10-13; Asare, 1980 : 54-56) in his study of the empirical typology of the age patterns of fertility classified five different types of

fertility curves according to the proportion of total fertility to different age groups of women. The classification is shown in Table 3.9. The table also

shows the United Nations'(1963 : 106) classification of the age patterns of

fertility based on the study of the age specific fertility rates from 72

countries. The United Nations' classification of the age patterns of fertility

takes into account the 5 year age group of women at which the age specific

fertility rate is at a maximum, and "the degree of concentration of fertility

in age groups at or near the peak" (United Nations, 1963 : 106). On the other

hand, the typology due to Pavlik makes it possible to arrange different fertility

schedules according to the age patterns of women in the reproductive period.

TABLE 3.8 48

PERCENTAGE DISTRIBUTION OF FERTILITY BY AGE GROUPS AND TOTAL FERTILITY RATES FOR BANGLADESH : BSF AND OTHER SOURCES

Year Area Per Cent Age Distribution of Fertility Total MeanFertility Age

Rate(15-49) m

Age Groups

15-19 20-24 25-29 30-34 35-39 40-44 45-49 Total

BFS5

Rural 15.6 25.2 22.2 17.6 13.5 4.4 (1.6) 100.0 7.3 27.91971 Urban 11.5 28.0 20.6 21.9 13.1 4.9 (0.0) 100.0 6.6 28.1

Total (W) 15.2 25.4 22.2 17.8 13.5 4.5 (1.2) 100.0 7.2 27.9

Rural 16.4 26.3 21.8 18.2 11.2 4.2 1.9 100.0 6.0 27.61972 Urban 13.9 26.8 29.7 16.1 10.1 3.5 0.0 100.0 5.2 27.1

Total (W) 16.2 26.4 22.4 18.1 11.1 4.3 0.0 100.0 5.9 27.5

Rural 16.9 27.3 24.0 16.8 10.7 4.2 0.0 100.0 6.0 27.01973 Urban 15.0 24.6 27.6 19.5 9.0 4.3 0.0 100.0 4.9 27.3

Total (W) 16.7 27.1 24.4 17.0 10.6 4.2 0.0 100.0 5.9 27.0

Rural 17.5 26.9 21.9 19.3 9.9 3.2 1.3 100.0 5.6 27.11974 Urban 14.7 31.8 23.1 18.6 7.6 4.2 0.0 100.0 5.0 26.8

Total (W) 17.2 27.2 22.0 19.2 9.8 3.3 1.4 100.0 5.5 27.1

Rural 19.1 26.6 23.9 17.5 9.1 3.4 0.4 100.0 5.2 26.61975 Urban 14.9 33.3 25.4 15.5 4.5 4.9 1.5 100.0 5.0 26.9

Total (W) 18.7 27.2 24.0 17.4 8.8 3.4 0.6 100.0 5.2 26.6

Rural 17.1 26.5 22.8 18.0 11.1 3.9 0.7 100.0 6.0 27.21971-75 Urban 13.9 28.9 25.1 18.3 9.0 4.3 0.6 100.0 5.3 27.2

Total (W) 16.8 26.7 23.0 18.0 10.9 3.9 0.7 100.0 5.9 27.2

BRSFMb

1974 Total 14.0 23.8 22.0 18.5 13.9 6.7 1.0 100.0 7.1 28.4

NISC

1966-68 Total 21.6 25.8 21.4 17.0 10.8 3.2 0.3 100.0 5.8 26.5

PG1E (Chandrasekar-Deming)

1963-65 Total 18.7 24.3 24.1 17.6 10.1 3.8 1.5 100.0 7.4 27.2

PGE^ (Longitudinal Registration)

1963-65 Total 18.3 24.1 24.9 17.6 10.3 3.8 1.0 100.0 6. 3 27.1

Note: The percentages in brackets are calculated on the assumption that the age specific fertility ratesin the age group 45-49 for the year 1971 are the same as those for 1972.W = Weighted.

Sources: a pregnancy history data: Bangladesh Fertility Survey, 1975. ASFR's are computed bymultiplying age specific marital fertility rates for ever married women with proportions ever married found in the 1974 Bangladesh Population Census.

b Bangladesh Retrospective Survey of Fertility and Mortality, 1974 : Census Commission,(1977 : 77).

c National Impact Survey, 1968-69 : Sirageldin et al. (1975 : 209). d Population Growth Estimation Experiment, 1962-65 : Farooqui et al. (1971 : 95).

49

Different fertility schedules for Bangladesh are converted in Table 3.10

according to Pavlik's and United Nations' classifications. The 'CDDC' pattern

of the age curve of fertility is depicted by the three out of five single

calendar year fertility schedules for rural Bangladesh. The age patterns of

fertility followed by the fertility schedules for all Bangladesh are the same

as those followed by the fertility schedules for rural Bangladesh except for

the year 1971. As classified according to Pavlik's model, urban fertility

schedules show different patterns in different single calendar years from 1971 to 1975. This feature is difficult to explain. Relatively smaller numbers of

respondents in the urban areas and the shorter period for which the rates were calculated, may partly account for this. However, the urban fertility schedule

based on 5 year (1971-75) average has not differed much from the 'CDDC' pattern

revealed by the corresponding fertility schedules for rural and all Bangladesh.

It can be noted from Table 3.10 that the Chandrasekar-Deming and longitudinal registration estimates of the fertility schedules from the Population Growth

Estimation Experiment have depicted the similar pattern as shown by the 5 year (1971-75) average fertility schedule for Bangladesh based on the data from the

Bangladesh Fertility Survey according to Pavlik's classification of the age

pattern of fertility. Although it is claimed that in countries with little use

of contraception and high total fertility rate the normal age curve of fertility is characterised by Pavlik's 'BCCA' pattern (Kpedekpo, 1976 : 13), only the

fertility schedule derived from the Bangladesh Retrospective Survey of Fertility

and Mortality data is found to approximately follow this pattern.

With the exception of the fertility schedules for urban Bangladesh in the

years 1972 and 1973 obtained from the Bangladesh Fertility Survey data and those

obtained from the Population Growth Estimation Experiment, all the fertility

schedules have followed the 'early peak' pattern. Thus the fertility schedules

for Bangladesh may generally be characterised by Pavlik's 'CDDC' pattern and

the United Nations' 'early peak' pattern. The mean of fertility schedule has

TABLE 3.9 50

PROPORTION OF TOTAL FERTILITY IN SPECIFIED AGE GROUPS :

PAVLIK'S MODEL

Type Age Group of Women

15-29 15-24 15-19 40-49

A 41-50 11-20 0-5 9+

B 51-60 21-30 6-10 7-8

C 61-70 31-40 11-15 5-6

D 71-80 41-50 16-20 3-4

E 81-90 51-60 21-25 0-2

Source: Kpedekpo, 1976 : 12

United Nations11 Mode1

Early Peak Type Maximum fertility at age group 20-24

Late Peak Type Maximum fertility at age group 25-29

Broad Peak Type Age specific fertility rates at ages20-24 and 25-29 differ only slightly

Source: United Nations, 1963 : 106

TABLE 3.10 51

CLASSIFICATION OF TYPES OF FERTILITY PATTERNS IN BANGLADESH

ACCORDING TO THE PROPORTION OF TOTAL FERTILITY FALLING

INTO SPECIFIED AGE GROUPS I3Y PAVLIK' S AND U.N. MODELS

Year Area Age Group of Women Model

15-29 15-24 15-19 40-49 Pavlik U.N.

BFS3Rural 63.0 40.8 15.6 6.0 CDDC EP

1971 Urban 60.1 39.5 11.5 4.9 BCCC EPTotal (W) 62.8 40.6 15.2 5.7 CDCC EPRural 64.5 42.7 16.4 6.1 CDDC EP

1972 Urban 70.4 40.7 13.9 3.5 CDCD LPTotal (W) 65.0 42.6 16.2 5.8 CDDC EPRural 68.2 44.2 16.9 4.2 CDDD EP

1973 Urban 67.2 39.6 15.0 4.3 CCCD LPTotal (W) 68.2 43.8 16.7 4.2 CDDD EPRural 66.3 44.4 17.5 4.5 CDDC EP

1974 Urban 69.6 46.5 14.7 4.2 CDCD EPTotal (W) 66.4 44.4 17.2 4.7 CDDC EPRural 69.6 45.7 19.1 3.8 CDDD EP

1975 Urban 73.6 48.2 14.9 6.4 DDCC EPTotal (W) 69.9 45.9 18.7 4.0 CDDD EPRural 66.4 43.6 17.1 4.6 CDDC EP

1971-75 Urban 67.9 42.8 13.9 4.9 CDCC EPTotal (W) 66.5 43.5 16.8 4.6 CDDC EP

BRSFMb

1974 Total 59.8 37.8 14.0 7.7 BCCB EP

NISC

1966-68 Total 68.8 47.4 21.6 3.5 CDED EPdPGE (Chandrasekar-Deming)

1963-65 Total 67.1 43.0 18.7 5.3 CDDC BP

PGE (Longitudinal Registration)

1963-65 Total 67.3 42.4 18.3 4.8 CDDC BP

Note: EP (Early Peak); LP (Late Peak); BP (Broad Peak).W = Weighted

Sources: a Bangladesh Fertility Survey, 1975 Pregnancy history datab Bangladesh Retrospective Survey of Fertility and Mortality,

1974 : Census Commission (1977 : 77).c National Impact Survey, 1968-69 : Sirageldin et al. (1975 : 209)

52

been found to be about 27 years with a few exceptions (Table 3.8).

The age curves of fertility for Bangladesh from various nationally represent

ative sources are plotted in Figure 3.1. The shape of the fertility curves

drawn from the Bangladesh Fertility Survey data and the National Impact Survey

data is found to be almost similar except for the age group 15-19. It may be

pointed out that both the surveys used pregnancy history data. The discrepancy

in- the age group 15-19 may be explained in part by the proportion married

employed in the two surveys. The fertility curve from the Bangladesh

Retrospective Survey of Fertility and Mortality has tapered off abruptly from

age group 25-29 and has depicted an 'older' fertility schedule. On the other

hand, the Chandrasekar-Deming estimate from the Population Growth Estimation

Experiment has demonstrated a 'younger' fertility schedule. Thus it appears that different estimation procedures employed in various surveys,in addition

to the differential inaccuracy of data,might have caused differences in the shape of the age curves of fertility in Bangladesh.

3.7 EFFECTS OF POSSIBLE SOURCES OF ERROR ON FERTILITY ESTIMATESIn Bangladesh where a vast majority of the respondents are not calendar

year conscious and are even ignorant of their own date of birth (see Chapter 2),

fertility rates derived from pregnancy history data are likely to suffer from

the response errors in the form of misstatement of age of mothers and mis­

placement of births in time. Omission of births due to recall lapse also might

have a considerable effect on the fertility estimates. Effects of age mis­

statement, omission and misplacement of events on estimating fertility rates derived from pregnancy history data were discussed broadly in Chapter 2. The

purpose of this section is to examine the possible effects of these errors on

the observed fertility estimates for Bangladesh.

Until now, no direct method for detecting the effects of various response

errors (misstatement of age, omission and misplacement of events) on fertility

AGE

SPEC

IFIC

FER

TILI

TY R

ATES

FOR

BAN

GLAD

ESH

FROM

DIF

FERE

NT S

OURC

ES

53

in inIT) COI ICM CM

>1 CO ID-P (Ti cr>•H r—I rHi—I•H -4-1 4-1 -Uu a aas a) cdID [m E Er- *h *H

I 4-4 L 5-1m o a) a)CL O-iCTi > i CTi X XI—I (D i£) W W

> I- l co a a —>, P ID O O COS CO O *H -HOP> <—I 4-1 4-S *HL CD ccS aS 4->P > " E E aSCO -H D>~i "H -—- -H P-l

>i U4J as r - l w -H w•H Oj cn PrH US

4-) Q 4-1•H OL 4-1 >i U O L O

4-> aS L OS Las asPm « •h a, es ^ esas h a us c

us usas as V4 I—I -H L "H

0 OS 4-> T! 4-)2 C as C aJrS -H D 3a a c 4J a, u aas as as aS O

CQ CQ

CO CO CM CM I—I rH

asCnC

serpen Ä ^ T iT rp a e j O T jx o a d s sßv

Grou

p of

Wom

en

54

rates is available. The alternative is to inspect the observed fertility

rate for the apparent distortions in the light of the available evidence.

Internal consistency checks are also sometimes helpful. In addition to the

response errors, it is suspected that the observed fertility rates might have

been affected by the procedure of imputing the dates of birth of a vast

majority of respondents and their children that were not reported in calendar

years and months (see Chapter 2). It may be pointed out that the response errors affect fertility rates either independently or in combination with one another.

It is sometimes very difficult to isolate the individual effects of these errors

on fertility estimates. Thus attempts will be made to focus on the combined

effects or individual effects (where possible) of these response errors on the

observed fertility estimates for Bangladesh.

According to Potter (1977b : 108), fertility rates for various age cohorts

of women in the age group 15-19 are sensitive to age at first marriage, age mis-reporting, omissions and event misplacement. In Table 3.6, it was observed

that in age groups 15-19 corresponding to various age cohorts of women other than the youngest cohort (aged 15-19 years at interview), the older the cohort

the lower was the fertility in rural and all Bangladesh. For urban women a similar situation was also observed with the exception of the cohorts aged 20-24 and 30-34 years at interview. As there has been no evidence of declining age at

marriage in Bangladesh during the 35 years before the survey, one would normally

expect that the fertility rates for various age cohorts in the age group 15-19 would be higher the older the cohort. The contradictory situation observed for

all age cohorts other than the youngest cohort in Bangladesh indicates that a

substantial amount of response errors have crept into the reported fertility

rates.

Effects of omission of births may be examined from the observed fertility

rates, particularly for the older age cohorts of women. The cohort of women

born in the period 1931-35 (aged 40-44 years at interview) generally reported

55

the lower fertility rates as compared to those reported by other age cohorts

of women in the rural areas (Table 3.6). This provides evidence of a

substantial omission of births by the older women. Age misreporting and event

misplacement among the older rural women might also partly account for the reported lower fertility.

One of the internal consistency checks for the possible effects of the

response errors on the observed fertility rates is to examine the patterns of

the reported sex ratio at birth. As discussed in Chapter 2, the reported sex

ratios at birth per 100 females in 5 year periods have revealed a sex-selective

under or over reporting of births, particularly beyond 25 years before the

survey date. Hence, the cross-sectional fertility rates might not show the real fertility situations in Bangladesh during the period 25-35 years before the date of the survey. However, a study of the sex ratio at birth by various age cohorts of women might focus on the possible effects of response errors on

the observed fertility rates. The sex-ratios at birth for various age cohorts of rural women in the age group 15-19 have shown a clear time trend beyond the period 15 years before the survey (Table 3.11). This is indicative of possible

under-reporting of female births in the early years of childbearing. As pointed

out earlier, the sex ratio at birth per 100 females in Bangladesh is assumed to lie somewhere between 102 and 107 (see Chapter 2). The sex-ratios at birth in

Table 3.1i have shown marked deviations from the expected range with few

exceptions both for rural and urban age cohorts of women. Misplacement of births

in time may also be attributed to the observed patterns of sex ratios at birth.

For example, the cohort of rural women aged 35-39 years at the time of

interview reported a sex ratio of 92 births in the 15-20 years before the

survey while the sex ratio in the period 20-25 years before the survey was

reported to be 116 (Table 3.11). It seems that either some male births

were shifted from the period 15-20 years preceding the survey to the previous 5 year period or some female births were shifted from the period

20-25 years to the period 15-20 years before the survey. Nevertheless,

as pointed out in Chapter 2, the 317 births with sex not stated might

TABLE 3.11 56

SEX RATIO AT BIRTH BY AGE OF MOTHER AND YEARS BEFORE THE SURVEY PER 100 FEMALE BIRTHS IN BANGLADESH

50 Exact Age of Women at the

** Time of Survey 45

Sex Ratios for Rural Areas

Years before the Survey50 Exact Age of

Women at the ** Time of Survey

Sex Ratios for Urban Areas

Years before the Survey50 Exact Age of

Women at the ** Time of Survey

Sex Ratios for Bangladesh (total weighted sample)

Years before the Survey

* Less than 30 birthsSex Ratios for the oldest birth cohort (aged 45-49 at interview) are based on 4-year average.Pregnancy history data, Bangladesh Fertility Survey, 1975-76.

57

somewhat affect the sex ratio at birth, particularly for rural Bangladesh in the relatively distant past.

Event misplacement in the form of pushing the earlier births forward and

the recent births backward in time affects the fertility rates derived from

pregnancy histories. If such misplacement of live births should occur, the

cross-sectional fertility rates in the most recent period (0-5 years before

the survey) would be understated and the rates corresponding to the period

5-10 years prior to the survey would be inflated (Cain et al., 1979 : 5; Brass

and Rashad, 1980 : 16). A decline in cross-sectional fertility in the period

0-5 years before the survey and an apparent overstatement of fertility in the

period 5-10 years preceeding the survey for rural and all Bangladesh (Table 3.6)

indicate the existence of event misplacement errors. However, as discussed earlier, there might have been a fertility decline in Bangladesh during the

5 year period immediately preceeding the survey due to the social disruptions caused by the war of liberation in 1971 and the 1974 famine. A change in the proportion of ever-married, particularly before age 20, as found in Table 3.2 might also contribute to a reduction in fertility in the 0-5 years before the

survey. The cross-sectional marital fertility rates in the period 10-15 years

before the survey for urban Bangladesh seem to have been affected by event

misplacement error coupled with age misstatement, particularly in the younger

age groups of mothers (see Table 3.6). There has been no evidence of a

substantial fertility decline in Bangladesh during the 1940's"*’ and 1950's.

The observed lower cross-sectional marital fertility rates, in general,

corresponding to the different 5 year periods beyond the 15 years preceeding

the survey both in rural and urban Bangladesh indicate that these rates were

affected by errors due to omission of births, misreporting the age of mothers

1 There was a slight fertility decline in Bangladesh during the 1940's due to the 1943 Bengal famine (Khan, 1973a: 179).

58

and misplacement of births in time. Thus, for a developing country like

Bangladesh, it may be suggested that the analysis of fertility based on

pregnancy history data should not extend beyond the last 15 years preceeding

the survey.

3.8 ADJUSTMENT OF FERTILITY RATES

As observed in Section 3.7, fertility rates derived from pregnancy

histories were affected by errors due to misstatement of age, omission and

misplacement of events. These errors were found to distort the real fertility

situation either individually or in combination with one another. Unfortunately,

no reliable technique has so far been developed for detecting and correcting

the effects and interactions of these errors on fertility estimates. However,

Brass (1971 : 10-15 and 1975 : 44-49) and Brass and Rashad (1980 : 12-28)attempted to develop procedures for adjusting fertility rates using internal

evidence from the survey. This section discusses the relevance of these adjustment procedures in the context of the fertility situation in Bangladesh.

3.8.1 Brass's First Birth ProcedureBrass (1971 and 1975) has devised a procedure for correcting and adjusting

fertility rates for event misplacement error in pregnancy history data with

the distribution of first births. He assumes that first births are invariant

in time and "... the timing biases computed for first births apply to all births" (Brass, 1975 : 44). Under this assumption, the total births for each

age cohort of women are allocated in accordance with the correction needed to

bring the observed distribution of first births in line with the standard

distribution of first births. The distribution of age-specific first births

per 1,000 women in the most recent time period (0-4 years before the survey)

is used as the standard for deriving the correction factors for the observed

first births and all order births. It is to be noted that the totals of births

reported by each age cohort of women remain the same, only the births are

redistributed to different time periods.

59

The proportions of the cumulative standard distribution of first births

per 1,000 women for each age group of women are calculated for logit transform­

ations. The proportion for the total standard first births corresponding to

the last age group of women is arbitrarily taken as 0.98. The logit of

proportion, p, is calculated for all age intervals from the start of reproduction

(5, 10, 15, .... 30 years) by 0.5 log^ [p/(1-p)].The next step of the procedure is the calculation of the proportions for the cumulative distribution of first

births and total births for each age cohort of women. In each case, the

proportion for the last age group of women is taken from the corresponding

proportion as calculated for the standard distribution of first births. The

logits of proportions for first births and all order births are then calculated.

The logits of the standard distribution of the cumulated first births are

plotted as ordinates against the age interval points as abscissa. The plotted points are joined with a freehand curve. For any age cohort of women, the logits of the cumulated first births are then plotted at points where the

ordinate values fall on the standard curve. The difference between the points for the age cohort and the corresponding points for the standard distribution

of first births can be read off from the abscissa. This difference provides an

estimate of the event misplacement error. The points for the age cohorts are referred to as end points of the corrected age intervals and these can be used

to plot the end points on the curve drawn for all order births for any

specified age cohort of women. Reading off from the ordinate at these end

points provides the logit values that can be transformed back to proportions

and finally to adjusted age-specific fertility rates. This procedure is

illustrated in Brass (1975 : 44-49).

Potter (1977a : 347) has demonstrated that the first birth procedure

developed by Brass is inapplicable for adjusting fertility rates for event mis­

placement error. He has found that the errors in reporting of first births are not the same as those for all order births. In particular, Brass's assumption

60

of the distribution of first births in the period 0-4 years before the survey

as the standard may not be applicable in case of Bangladesh because of social

disruptions caused by the 1971 war of liberation and the 1974 widespread

famine in the 5 year period proceeding the survey. For these reasons, the

first birth procedure suggested by Brass is not used for adjusting the fertility

rates derived in this study.

3.8.2 P/F Ratio Method

P/F ratio method was originally developed for application to the data

from censuses and surveys, where the total number of children ever born and

the births in the past 12 months were reported. Recently, Brass and Rashad

(1980 : 12-18) have applied the P/F ratio method for adjusting fertility rates

derived from pregnancy history data collected in the Bangladesh Fertility Survey, 1975-76. In the original application, P refers to the mean parity of women at the time of the survey and F to the equivalent value calculated from

the reported births in the past 12 months. When applying the method to the pregnancy history data, P relates to the mean parity for any cohort of women at a particular point of time calculated from the date of survey backwards and F relates to the equivalent value cumulated from fertility rates in a

given period for any synthetic cohort. Under the assumption of constant

fertility over a certain period of time, all the ratios P^/F^ (where i = 1, 2,

3, ..., 7 refers to the 5 year age groups of women 15-19, 20-24, 25-29 ..., 45-49

respectively) are unity. Deviations of these ratios from unity indicate the

effects of response errors (age misstatement, omission and misplacement of

events), changing fertility or both. Since the impact of fertility change

and errors are typically different in different periods before the survey,

successive application of the technique for different periods may help identify

the distortions caused by reporting of births, especially for earlier periods.

61

According to Brass and Rashad (1980 : 14), under the assumption of

constant fertility and the same misallocation of births for all age groups of

women in the recentyears, the P/F ratios for the recent period are constant

and show a gradual decline with increasing age of women. If, on the other

hand, the P/F ratios are greater than unity and show a decline at older ages,

it may be an indication of declining fertility in combination with omission of

births at older ages.

This method can be applied to lower and higher order births. If the

existence of a fertility decline rather than errors in data is suspected, a

comparison between the pattern of change in the P/F ratios obtained by applying the method to lower and higher order births may establish the presence

of errors. According to Brass and Rashad (1980 : 13), higher order births are more likely to be influenced than lower order births in the case of a fertility decline. If the ratios show the opposite situation, the existence of errors in the data is well substantiated.

For adjusting period fertility rates based on this method, generally P^/F^ is used as the correction factor. The reason is that cumulative cohort

measures for younger women are considered to be less affected by omissions of

births. A discussion of the important findings derived by applying this method to the pregnancy history data for Bangladesh by Brass and Rashad (1980) is

presented below.

P/F ratios were calculated for the years 1966-70, 1971-75 and 1975 under

the assumption of constant fertility and for the periods 1961-65 and 1966-70 under the assumption of changing fertility over the 5 year period preceding

the survey. The results are reproduced in Appendix Table C 1.

In successive application of the P/F ratio method for different periods

preceeding the survey under the assumption of constant fertility, cumulated

measures for cohorts up to the time of the survey were used. In application

62

of the method to an earlier period under the assumption of changing fertility,

cumulated measures up to the end of that period were used. Irrespective of

which assumption was valid, sudden drops in the ratios for the women aged

45-49 at the time of the survey have revealed a considerable amount of omission

of births by this cohort of women. For all other age groups of women, under­

reporting of births were suggested to be small.

Under the assumption of constant fertility, the magnitude of the P/F

ratios (higher than unity) for all age groups of women in the 5 year (1971-75)

period preceeding the survey indicated that either fertility declined sharply

or the rates were affected by errors due to misplacement or omission of events.

The P/F ratio in the period 1966-70 was much lower than unity for the age

group 15-19 and none of the ratios exceeded unity for the other age groups of women. This might also be due to misplacement of births in time in addition to the under-reporting of births. Thus, according to Brass and Rashad (1980 : 16),

if fertility in Bangladesh remained constant, the rates for the period 1971-75 were under-stated and the rates for the period 1966-70 were over-stated.

"Under the assumption of declining fertility over the last five years, the P/F ratios for 1966-70 and 1961-65 suggest that the fall was preceeded by a substantial increase in period fertility at ages above twenty-five years"(Brass and Rashad, 1980 : 16).

This situation is unlikely because of the high fertility rates implied for the

period 1966-70. It was argued that even if some decline in fertility occurred,

misplacement of events in time was the dominating factor (Brass and Rashad,

1980 : 16).

The P/F ratio method was also applied for first births and 4+ order births

for the most recent period (0-5 years preceding the survey). The results are

reproduced in Appendix Table C 2. In case of a real decline in fertility it is expected that the 4+ order births would be more affected than the first births

because the proportion of women who became mothers would be relatively stable.

63

But quite the opposite situation was revealed by the P/F ratios which strongly

suggests that the data were affected by errors. It was argued that with some

increase in age at marriage the P/F ratios for first births show that "....the

mean age of women at first birth would have to be increasing by more than

one year in five calendar years steadily over some twenty years to account for

the P/F ratios" (Brass and Rashad, 1980 : 16). Using P /Fn as the correction

factor, the adjusted total fertility rates for the years 1971-75 and 1975 were

found to be 7.50 and 7.36 respectively under the assumption of constant fertility. It is to be noted that Brass and Rashad (1980 : 15) found the unadjusted

total fertility rates to be 6.34 and 5.37 for the years 1971-75 and 1975

respectively.

Brass and Rashad (1980 : 18) by applying the P/F ratio method concluded

that the low levels of fertility for Bangladesh in the years 1971-75 and 1975

were mainly due to event misplacement errors in the location of births. Although

it was argued that fertility in the period 5-10 years before the survey was

overstated and that in the period 0-5 years before the survey was under-stated

(as discussed earlier),Brass and Rashad (1980 : 18) were of the opinion that

there was no significant fertility decline in the period 1971-75. This is

contradictory to what has been observed in the present study. As pointed out

earlier, evidence of a significant decline in fertility was found from the

reasonably complete vital registration data in Matlab Thana during the same

period. Amin and Faruqee (1980 : 6) using the data from the Bangladesh

Fertility Survey, 1975-76 also found a sharp decline in marital fertility for

currently married women in the period 1974-75. The adjusted total fertility

for the period 1971-75 by P/F ratio method is also found to be much higher than

the corresponding observed total fertility estimated in the present study.

3.8.3 Gompertz Relational Mode1

The P/F ratio method, though simple, has some limitations. It can provide

effective estimation of the total fertility only for a relatively short period

before the survey under unchanging or moderately changing fertility and when

the data are mainly affected by reference scale error in the location of births.

64

It is also not quite satisfactory for detecting fertility trends when

complicated response errors influence the data. Moreover, the method is

incapable of displaying the patterns of age-specific fertility rates. Attempts

were made by Brass (1979 : 40-53) and Brass and Rashad (1980 : 18-28) to over­come those shortcomings by imposing an underlying pattern of fertility by age

and time.

The variation of fertility with age of women can be represented by the(x-x )B oGompertz curve as F (x)/f = A where F(x) is the cumulative fertility to

age x, F is the total fertility, A and B are positive constants less than

unity and x^ is a convenient origin for age. Taking natural logarithm

twice and simplifying, the above equation becomes Y(x) = a + ßx where Y(x) =

-£n [-&n (F(x)/F)j. According to Brass and Rashad (1980 : 19), this model does not represent the pattern of fertility adequately at the tails of the distribution. Here Y(x) is related linearly to x. It was suggested that the model would be improved if Y(x) was related to a standard set of

fertility pattern Ys(x) where Ys(x) = -&n [-£n (Fs(x)/Fs) ]• Heather Booth (Cited in Brass 1979 : 47, and Brass and Rashad, 1980 : 12) derived a

standard set of Ys(x) from an extensive study of observed fertility patterns and the Coale-Trussell(1974) fertility model. The relational Gompertz model

Y(x) = a + B Y (x) with parameters a, B and F is found to be more accurate

than the original Gompertz.

The parameters ct and B of the relational Gompertz model can be determined when the age specific fertility rates for the whole reproductive period of

women are known. In this case, F(x)/F and taking the negative of the double

natural logarithm Y(x) can be calculated for a given x. A comparison of

Y(x) and Ys(x) graphically gives the characteristics of the observed

fertility rates. It is then possible to estimate a and B by fitting a straight line. However, F is not known for the cohort of women who are still in their reproductive ages. A method developed by Zaba (Cited in Brass and Rashad,

65

1980 : 20) employs the distribution of mean parity of a cohort up to the time

of survey for estimating a , 3 and F. This method involves the ratios of

cumulative fertility measures (mean parity) in two successive 5-year age

groups within a cohort of women. Let P , P^, ..... be the mean parity of

women aged 15-19, 20-24, 25-29 .... years respectively for a cohort over time.

Fh is defined as Pi-p/P-^ (R = PQ/Pp where P^ is the mean parity of womenaged 10-14, R = P-./P , R = P./P etc.). It has been shown approximately by

2 ± 2 3 z 3Zaba that G(i) = - £ (-£ R.) = e. + a + 3 <}>(i) where e. and (f)(i) are then n l l l

*values calculated by using the Booth relational standard and ot is approximated 2by a + 0.48 (3-1) . To apply the procedure, the values of [G (i) - e j are

obtained and compared with the cj)(i) values graphically or by some other means.*3 and a can be estimated by fitting a straight line. It has been noted that

*a is very close to a in most cases. After estimating a and 3 from thereported fertility measures, Py/F can be estimated by Z (i) = - £ (Pp/F)] =

a + 3 Z (i) where z (i) has been tabulated using the Booth relational

standard. Equating Pp/F with the reported cumulative fertility measures

provides a series of estimates of F. These F values are averaged in a

suitable way. These are denoted by F where c refers to age cohorts of womencat the time of survey.

The next step is the estimation of a and 3 for each cohort of women.c cAttempts were made to estimate these by varying them linearly over time, that

is, by fitting models ac=a+ca and 3c = 3 + cb. However, by applying these

procedures to a number of sets of pregnancy history data, it was noted that

".... the two trend measures a and b left too much freedom so that the fitting

was distorted by the effects of particular reference scale errors" (Brass and

Rashad, 1980 : 21). It was suggested that the a could be fixed as a and 3c ccould be varied linearly with c. The argument for this is that the spread of

fertility (as estimated by inverse of 3) reduces with level due to greater

influence of family planning at later ages in the childbearing period while

66

the behaviour of a cannot be well predicted because of its dependence on

both family planning and changes in age at marriage. Thus the model becomes

a relational Gompertz over age with parameters F^a and 3 + cb. The fitting

procedure is illustrated in Brass and Rashad (1980) for Bangladesh and for

more educated women in Sri Lanka. Important findings derived by the

application of the model in the context of Bangladesh are presented below.

Observed fertility rates for birth cohorts aged 25-29, 30-34, 35-39 and

40-44 years at the time of survey were compared with those obtained by fitting

the model. It was noted that the reported births per woman for the period

of 5 to 10 years before the survey were too high and those in the 0-5 years before the survey were too low for all the cohorts. Deviations of the observed

fertility rates from the fitted ones were found to be less systematic in the

5 year periods from 1946 to 1965. The causes of these large discrepancies were attributed mainly to reference scale errors in the location of births.

Using the model, the total fertility rates for the periods 1961-65, 1966-70 and 1971-75 were found to be 7.57, 7.58 and 7.38 respectively. According to Brass and Rashad (1980 : 29), the apparent decline in total fertility in the period 1971-75 has some reason to be treated with caution. Brass and Rashad

were doubtful whether all the event misplacement errors in the location of

births were eliminated by fitting the model. It was argued that fertility

decline might have occurred at earlier ages as the fertility difference of

about 0.2 was established by age 20-24 years of women between the periods

1966-70 and 1971-75.

Summarizing the findings in a paper, Brass. (1980 : 20) concludes that

the Relational Gompertz model has probably provided a satisfactory represent­

ation of trends but not of levels of fertility in Bangladesh where the response errors have affected the data considerably. However, it can be noted here

that the total fertility rate of 7.57 for the period 1971-75 appears to be

67

over-estimated by this procedure as compared to the corresponding observed

rate of 5.9 estimated in the present study. It is worth pointing out that

Rodriguez and Cleland (1980 : 23) using the data from the Bangladesh Fertility

Survey, 1975-76 estimated the total fertility rate at 6.2 (unadjusted) in the 5 year period preceding the survey. This estimate does not differ much from the corresponding estimate derived in the present study.

CHAPTER 4 68

INFANT AND CHILDHOOD MORTALITY

4.1 INTRODUCTIONMortality in the region now comprising Bangladesh has been declining more

or less steadily since the first decade of this century (Robinson, 1967 : 39).

The crude death rate has declined from 45.6 to 18.5 per thousand population

during the period 1901-65 (Table 4.1). The downward trend in mortality was

halted in the 1940's as a result of the Bengal famine in 1943 and also

possibly because of the disturbances associated with the partition of the

former British India in 1947 (Robinson, 1967 : 28; Khan, 1973a : 180). The decline in mortality has been comparatively faster since the 1950's. Available evidence suggests that the rate declined further during 1963-69 (Khan, 1973b :

191; Stoeckel, 1973 : 21). Although recent national estimates of mortality are not precisely known, except for the year 1974, estimates based on the longitudinal vital registration data from Matlab Thana indicated a sharp increase in the crude death rate during the year 1971-72 as a result of the war of

liberation (Chen and Chowdhury, 1977 : 411). Examining the mortality fluct­

uations in Matlab Thana for the years following the liberation war, Chen and

Chowdhury (1977 : 411) have observed that"In 1972-73 some recovery was noted but full recovery was not attained until 1973-74. The 1974 famine marked the onset of another sharp increase of the death rate and recovery, although detectable, remained incomplete in 1975-76".

It is worth noting that the national estimate of the crude death rate of 20 per

thousand population in 1974 has agreed with that .found for Matlab Thana in the

same year (see Chowdhury and Curlin, 1979 : 114).

Available evidence suggests that infant and childhood mortality rates also

follow an almost similar trend as the crude death rate (see for example, Robinson,

1967 : 24-40; Chowdhury and Curlin, 1979 : 114). This chapter presents infant

TABLE 4.1 69

ESTIMATED CRUDE DEATH RATES FOR BANGLADESH, 1901-1974

Period Crude Death Rate ('000) Source

1901-11 45.6

1911-21 47.3

1921-31 41.7 Khana (1973a : 179)

1931-41 37.8

1941-51 40.3

1951-61 29.7

1962-65 18.5

1962-65 20.0 Farooqui and Farooq (1971 : 91)

1962-65 16.0 cFarooqui and Farooq (1971 : 91)

1974 19.4 Rabbani et al.d (1979 : 27)

1974 19.6 0Census Commission (1977 : 88)

Notes : a Rates for the decades 1901-11 through 1951-61 were obtained using the data from various censuses and the rate for the period 1962-65 was obtained from the Population Growth Estimation Experiment, 1962-65.

b Chandrasckar-Deming estimate from the Population Growth Estimation Experiment, 1962-65.

c Longitudinal registration estimate from the Population Growth Estimation Experiment, 1962-65.

d Derived from the 1974 Bangladesh Population Censuse Derived from the 1974 Bangladesh Retrospective Survey of

Fertility and Mortality.

and childhood mortality estimates for Bangladesh based on the pregnancy history

data from the Bangladesh Fertility Survey, 1975-76. The procedure for estimat­

ing these rates has been outlined in Chapter 2. Rural-urban and sex different­

ials in infant and childhood mortality will also be studied. Finally, the

estimates will be compared with those obtained in other studies. This will

70

provide an opportunity to assess the reliability of the pregnancy history

analysis in the context of Bangladesh mortality data.

4.2 INFANT MORTALITY

The risks of death are not uniform throughout the life span. Generally,

infants undergo high risks of death and the expectation of life increases after one has attained the exact age 1 (Robinson, 1967 : 10). Infant mortality

in' a society is mainly dependent on genetic factors, infants' exposure to

infection from environment, the care they get from their parents and the health

services they are offered. These are probably the reasons for which infant

mortality is suggested to be a reliable indicator of the overall health status

of a community or a country (Chandrasekar, 1959 : 67).

Table 4.2 shows the national as well as rural and urban estimates of infant mortality by sex in single calendar years and in 5 year periods for Bangladesh between 1955 and 1973. The rates for the 5 year periods were

calculated in order to reduce the effects of sampling variation. The estimates based on 5 year periods might give a better indication of the levels and trends

of infant mortality as these involve a larger number of reported births and infant deaths. As the Bangladesh Fertility Survey started interviewing the

respondents from December, 1975 (PCFP Division, 1978 : 36), the most recent calendar year for which a direct estimate of infant mortality could be made

was 1973 since some births after 1973 would not have been exposed to the

risks of dying for a full calendar year by the date of interview.

National estimates of infant mortality, in general, have exhibited a

declining trend during the period 1955-69. An almost similar downward trend

was also found for both the rural and urban areas. It may be noted that the

single calendar year rates have shown some fluctuations, the fluctuations being wider for the urban rates. The downward trend in infant mortality was

found to have halted during the period 1970-73. The following might be the

INFANT MORTALITY RATES ('000 LIVE BIRTHS) BY AREA AND BY SEX FOR

BANGLADESH, 1955-73

ir of Rural Urban Total (Weighted)L r t h _________________________________________

Male Female Both Sexes Male Female Both Sexes Male Female Both Sexes

1955 169 (213) 214 (196) 191 (409) 88 (68) 195 (41) 128 (109) 162 (278) 214 (248) 186 (526)

L956 203 (236) 168 (208) 187 (444) 90 (67) 96 (52) 92 (119) 193 (305) 165 (266) 180 (571)

1957 195 (215) 161 (254) 177 (469) 212 (66) 172 (58) 194 (124) 196 (280) 161 (323) 177 (603)

L958 185 (270) 159 (246) 172 (516) 145 (76) 174 (69) 159 (145) 183 (249) 160 (318) 171 (667)

L959 211 (265) 108 (278) 158 (543) 205 (83) 92 (87) 147 (170) 212 (345) 108 (362) 157 (707)

L960 141 (304) 158 (298) 150 (602) 122 (82) 215 (79) 168 (161) 141 (391) 162 (383) 151 (775)

L961 215 (339) 149 (328) 183 (667) 102 (98) 153 (98) 128 (196) 207 (439) 150 (426) 178 (864)

L962 188 (340) 159 (333) 174 (673) 162 (130) 181 (105) 170 (235) 186 (451) 161 (434) 174 (885)

1963 153 (411) 133 (377) 143 (788) 149 (101) 69 (101) 109 (202) 152 (526) 128 (485) 141 (1011)

1964 158 (430) 129 (426) 144 (856) 163 (135) 88 (113) 129 (248) 159 (560) 126 (548) 143 (1108)

L965 133 (443) 154 (436) 143 (879) 110 (146) 120 (142) 115 (288) 131 (580) 151 (570) 141 (1149)

L966 160 (500) 117 (520) 138 (1020) 130 (146) 131 (130) 130 (276) 157 (648) 119 (666) 138 (1314)

(.967 143 (530) 125 (471) 135 (1011) 137 (146) 106 (132) 122 (278) 143 (683) 123 (608) 134 (1292)

t968 127 (558) 120 (517) 124 (1075) 161 (143) 83 (132) 124 (275) 130 (716) 118 (663) 124 (1379)

1969 149 (531) 112 (544) 130 (1075) 127 (165) 62 (145) 97 (310) 148 (691) 109 (700) 128 (1391)

1970 153 (563) 120 (567) 136 (1130) 151 (146) 111 (162) 130 (308) 152 (723) 119 (733) 136 (1456)

1971 132 (597) 148 (527) 140 (1124) 99 (172) 115 (174) 107 (346) 129 (773) 145 (690) 137 (1462)

L972 156 (467) 125 (489) 140 (956) 158 (152) 84 (143) 122 (295) 156 (610) 122 (633) 138 (1244)

L973 153 (510) 139 (498) 146 (1008) 133 (150) 127 (150) 130 (300) 151 (661) 138 (647) 144 (1308)

Grouped Data (5-Year Average)

55-59 193 (1199) 158 (1182) 176 (2381) 150 (360) 140 (307) 145 (667) 190 (1557) 158 (1517) 173 (3074)

50-64 171 (1824) 144 (1762) 158 (3586) 143 (546) 137 (496) 140 (1042) 169 (2367) 144 (2276) 156 (4643)

55-69 142 (2562) 125 (2488) 134 (5050) 133 (746) 100 (681) 117 (1427) 142 (3318) 123 (3207) 133 (6525)

; o - 7 3 a 148 (2137) 133 (2081) 140 (4218) 134 (620) 110 (629) 122 (1249) 146 (2767) 131 (2703) 139 (5470)

;es: Figures in brackets refer to the reported number of births which occurred in each calendar year to all respondents (ever married)in the survey.

a Based on 4-year average.

tree: Pregnancy history data, Bangladesh Fertility Survey, 1975-76.

72

plausible reasons for the increase in infant mortality during the period 1970-73.

Firstly, a devastating cyclone accompanied by massive tidal waves with water up to 20 feet high swept across the southern coastal regions of Bangladesh

in November, 1970 having a significant consequence on mortality. The cyclone

mortality was claimed to have been about half a million (Sommer and Mosley,

1973 : 122-123; Khan, 1973b : 193). Secondly, the 1971 war of liberation

which continued for about 9 months had a serious impact on mortality in

Bangladesh. Although it was a difficult task to estimate precisely the loss

of lives due to the liberation war, "Impartial foreign observers and the

Government of Bangladesh have estimated 1 to 3 million civilian deaths"

(Greenough et al., 1973 : 242; Khan, 1973b : 193).

Infant mortality in the rural areas were found to be generally higher than those in the urban areas. Comparatively lower infant mortality rates in

the urban areas might be partly explained by the higher standards of living of the majority of urban people with the privilege of better health and medical

facilities as the large public hospitals are generally located in cities and the district headquarters. Kabir (1977 : 163) and Arriaga (1980 : 11) also

found a similar urban-rural differential in infant mortality for Bangladesh.

It can be noted from Table 4.2 that infant mortality rates among males,

based on 5 year average periods, were consistently higher than those among

females in Bangladesh. Sex differentials in infant mortality might be attributed to the fact that ".. the female infants are biologically better

fitted than male infants for survival" (Chandrasekar, 1959 : 110). Higher

male than female infant mortality rates in Bangladesh were also observed by

Yusuf (1967 : 538), Curlin et al. (1976 : 97) and Kabir (1977 : 163). However,

some departures of this finding with regard to sex differential in infant

mortality were found in the case of single calendar year rates, particularly

in the urban areas.

73

Sex ratio at infant death is one of the means for checking the quality of the reporting of deaths. Male deaths are usually better reported than

female deaths (Potter, 1977a : 351; Brass, 1979 : 25). Table 4.3 shows that

the sex ratios at infant death in the distant past were comparatively higher except for the urban areas in the period 1960-64.

This indicated that female infant deaths were under-reported in the earlier

years. Sex ratio at infant death of 146 per 100 females in the period 1965-69

in urban areas appeared to be very high. This might be partly due to sampling

variations arising out of the small number of the reported infant deaths and

the misplacement of events in time. Thus, some of the infant mortality rates,

especially in urban areas might not be accepted as demonstrating the real situation in Bangladesh.

TABLE 4.3

SEX RATIO AT INFANT DEATH ('00 FEMALES) BY AREA FOR BANGLADESH, 1955-73

Period Rural Urban Total (Weighted)

1955-59 124 126 124

1960-64 122 115 122

1965-69 118 146 1191970-73 114 120 115

Source: Pregnancy history data, Bangladesh Fertility Survey, 1975-76

4.3 COMPARISON OF INFANT MORTALITY RATES WITH THOSE OBTAINED IN OTHER INVESTIGATIONSTable 4.4 presents the estimates of infant mortality for Bangladesh during

the period 1957-75 from various sources. The estimates are not strictly

comparable because of their different methods of estimation and different

reference populations. Moreover, errors in the data collection procedures in

74

different investigations make it more difficult for a direct comparison. However, a study of the estimates from different sources might help to assess

the reliability of the estimates derived in the present study.

It should be pointed out that none of the rural infant mortality estimates

with the exception of Kabir's (1977 : 163) represented entire rural areas of

Bangladesh. The rural estimates of infant mortality in the present study have generally fallen between the estimates derived from the other sources in

the corresponding period of time. Schultz's (1972 : 419) estimates of infant

mortality for the period 1957-61, based .on the pregnancy history data from

the Demographic Survey of East Pakistan, 1961-62 conducted in the central

region of Bangladesh, were lower while the estimates obtained by Stoeckel

and Chowdhury (1973 : 24) for the periods 1958-59, 1963-64 and 1964-65 were

higher than those obtained in this study in the comparable periods of time. Stoeckel and Chowdhury (1973) have applied the Bogue's Pregnancy history program to the data collected in Comilla-Kotwali Thana in early 1967. Infant mortality rates, derived by Chowdhury and Curlin (1979 : 114) using the

reasonably complete vital registration data from Matlab Thana, were generally

lower than those obtained in the present investigation for the period 1966-73. As Matlab Thana enjoys the benefit of emergency medical facilities offered by the Cholera Research Laboratory (Schultz, 1972 : 419), the rates for Matlab

Thana may be expected to be lower as compared to those for other rural areas

of Bangladesh. Nevertheless, it was claimed that the rates have truly

reflected the infant mortality situation in Matlab (Chowdhury and Curlin,

1979 : 110).

National estimates of infant mortality obtained by Sirageldin et al.

(1975 : 213) for the periods 1960-62, 1963-65 and 1966-68 were found to be

lower than those derived in this study for the corresponding periods of time. With regard to the lower infant mortality rates derived from the National

TABLE 4.4 75

ESTIMATES OF INFANT MORTALITY ('000 LIVE BIRTHS) FOR BANGLADESH FROM DIFFERENT SOURCES, 1957-75

Period Infant Mortality Rato*

Rural National

1957-61 150 -

1958-59 176 -

1960-62 - 139

1962-65 - 147

1963-64 148 -

1963-65 - 116

1964-65 156 -

1966 111 -

1966-67 139 -

1967 125 -

1966-68 - 113

1968 124 -

1969 128 -

1969-703 - 152

1970 _ 128

Source

Schultz (1972 : 419)

Stoeckel et al. (1973 : 24)

Sirageldin et al. (1975 : 213)

Farooqui et al. (1971 : 107)

Stoeckel et al. (1973 : 24)

Sirageldin et al. (1975 : 213)

Stoeckel et al. (1973 : 24)

Chowdhury et al. (1979 : 114)

Stoeckel et al. (1973 : 24)

Chowdhury et al. (1979 : 114)

Sirageldin et al. (1975 : 213)

Chowdhury et al. (1979 : 114)

Chowdhury et al. (1979 : 114)

Mitra (1979 : 29)

Arriaga (1980 : 5)

76

Impact Survey, 1968-69, Sirageldin et al. (1975 : 213) noted that the pregnancy

histories from which the rates were estimated might have suffered from under­

reporting of infant deaths, misreporting of age at death and inability of the

respondents to distinguish between a stillbirth and an early infant death.

It can be observed from Table 4.4 that the Brass estimate of infant

mortality rates for different time periods from different sources were almost

similar. All these estimates were found to be higher than the corresponding

estimates derived in the present study. It is worth pointing out that

Mitra (1979) and Arriaga (1980) used the data from the Bangladesh Fertility

Survey, 1975-76 as used in the present study. However, Arriaga's estimates

for different calendar years almost agree with the corresponding estimates

obtained in this study. Thus it may be argued that the estimation techniques

used have caused the difference in the infant mortality rates. Regarding the

reliability of the infant mortality rates based on the Brass technique, United

Nations (1967 : 36) remarked that "... these estimates exceed (often by a

major extent) the average level of infant mortality derived from registered

births and deaths in the four years before the survey ...". On the other

hand, Bogue and Bogue (1970 : 152-153) have noted that the infant mortality

rates derived from pregnancy history data give the lowest plausible estimates

where the actual level of infant mortality is considerably higher than the

observed level. Bogue and Bogue (1970) further noted that the infant

mortality rates from pregnancy history data for the period 5 years before the

survey are under-reported by 10-25 per cent depending on the quality of

interviewing and the recall ability of the respondents. Thus, the true

national estimate of infant mortality in the early 1970's would probably lie

between the estimates obtained by the Brass technique based on the survival

of children ever born and the method based on the survival of birth cohort

from pregnancy history records.

77

4.4 CHILDHOOD MORTALITYChildhood mortality based on pregnancy history data is usually confined

to studying the proportion of children dying before age 5. This is because

of the smaller number of children reported dead beyond age 5. However, there

are some studies in which the proportions of children dying before age 10 were calculated. For example, Hull (1975 : 229-232) has computed for various

birth cohorts of children, the probability of dying at selected ages from

birth to age 10.

Mortality of children between ages 1 and 4 (4 1) is generally free of the

impact of endogeneous risks (congenital malformation, premature birth and

injuries at birth) and reflects the extent of nutritional deficiency, in

particular, around and after the weaning period. Because of this fact, child­hood mortality in the age group 1-4 has been suggested to be a more sensitive

indicator of the general health conditions in a community (Willis and Waterlow,q q q1958). Further, Figures 4.1 and 4.2 show that the estimates 2 o, 3 o, 5 o and

q10 o, reflect almost similar trends in childhood mortality in Bangladesh.Hence, the use of any of these estimates would closely reflect the childhood

q qmortality situation. In this section, 4 1 and 5 o are taken as the indicators

of childhood mortality levels and trends in Bangladesh.

4.4.1 Mortality in the Age Group 1-4 (4^1)

Table 4.5 presents the probabilities of dying among children aged 1-4 by

single calendar years and 5-year periods for the national as well as rural and

urban areas of Bangladesh. Single calendar year values of 4C 1 fluctuated

from year to year making it difficult to establish any trends. The fluctuations

were more pronounced in the urban areas (see Figure 4.2), probably because of

the small number of reported deaths between ages 1 and 4. However, mortality

of children aged 1-4 based on 5 year periods has shown a declining trend for

both males and females in the urban areas. As found in the case of infant

PROB

ABIL

ITIE

S OF

DYI

NG B

ETWE

EN78

saiJiniavaoHd

Cale

ndar

Yea

rs

EXACT

AGE

x, xno

FOR

BANGLADESH,

1955-70

79

saiiimavaoHa

Calendar Y

ears

80

mortality rates, mortality of children in the age group 1-4 was also generally

lower in the urban areas. A similar rural-urban differential in childhood

mortality between ages 1 and 4 was found by Kabir (1977 : 164). Unlike the

findings of the sex differential in infant mortality, Table 4.5 indicates

that mortality among children aged 1-4 in Bangladesh was, in general, higher

for females than for males. This presumably reflected differential parental

concern and greater care of male children during the fragile years of early

childhood (Curlin et al., 1976 : 97). Ruzicka (1972 : 8) also observed a

similar sex differential in a number of South Asian countries.

Estimates of 4^1 are available only from a few sources in Bangladesh.

However, a comparison of the 4^1 values from different sources (Tables 4.5 and

4.6) reveals that the rural and national estimates derived in this study were, in general, somewhat closer to those obtained using the data from the Population Growth Estimation Experiment, 1962-63, and those derived from the Coale-Demeny West model life tables based on the data from the 1974 Retrospective

Survey of Fertility and Mortality. The national as well as rural and urban estimates of 4qi obtained directly from the 1974 Retrospective Survey data were markedly lower while those derived from the Coale-Demeny North model

(based on the 1974 Retrospective Survey data) were higher than those found in

the present study.

It is to be noted that the national estimate of 4qi for the year 1969-70

obtained by Mitra (1979), using the data from the Bangladesh Fertility Survey, 1975-76, was lower than the estimate for the year 1969 derived in this study.

(See Tables 4.5 and 4.6). Difference in the estimation procedure might

be the reason. According to Ruzicka (1972 : 7), in a number of developing

countries, 4 1 values have been found to exceed the corresponding

estimate based on the West Model life tables. If this were true for

Bangladesh, the national estimate of 4qi obtained in this study might

possibly have better indicated the childhood mortality in Bangladesh in the most recent year for which mortality of children between ages

TABLE 4.5 bl

ESTIMATES OF PROBABILITY OF DYING BETWEEN AGES 1 AND 4 (4qi)

BY AREA AND BY SEX FOR BANGLADESH, 1955-69

Year of Birth of Child

Rural Urban Total (Weighted)

Male Female Both Sexes Male Female Both Sexes Male Female Both Sexes

1955 .102 .130 .115 .048 .152 .084 .090 .128 .112

1956 .090 .069 .080 .115 .085 .102 .093 .072 .083

1957 .087 .099 .093 .115 .083 .100 .089 .096 .093

1958 .109 .121 .115 .123 .123 .123 .109 .124 .114

1959 .110 .117 .114 .030 .063 .048 .103 .111 .107

1960 .077 .084 .080 .083 .081 .082 .077 .081 .079

1961 .090 .104 .097 .080 .108 .094 .086 .105 .097

1962 .101 .054 .077 .073 .128 .097 .098 .060 .079

1963 .052 .095 .073 .058 .053 .056 .054 .092 .071

1964 .077 • 111 .094 .040 .078 .060 .076 .106 .092

1965 . 060 .095 .077 .023 .064 .043 .056 .091 .074

1966 .086 .102 .094 .110 .097 .104 .086 .102 .095

1967 .093 .068 .081 .079 .068 .074 .091 .068 .080

1968 .072 .097 .084 .033 .066 .050 .069 .096 .082

1969 .097 .106 .102 .063 .096 .079 .095 .106 .101

Grouped Data (5-Year Average)

1955-59 .100 .108 .104 .085 .095 .089 .098 .106 .102

1960-64 .078 .091 .084 .066 .089 .077 .077 .090 .084

1965-69 .082 .094 .088 .062 .078 .070 .080 .093 .087

Source: Pregnancy history data, Bangladesh Fertility Survey, 1975-76

82

TABLE 4.6

ESTIMATES OF PROBABILITY OF DYING BETWEEN AGES 1 AND 4 (4%)

FOR BANGLADESH FROM DIFFERENT SOURCES

Source Rural :-V- Urban National

Male Female Both Sexes Male Female Both Sexes Male Female Both Sexes

jpulation Growth Estimation tperiment, 1962-63*

- - - - - - .081 .101 .094

lolera Research»boratory, 1967-68* .093 .122 .109 - - - - -

rtrospective Survey of (a) .052 .051 .052 .042 .030 .036 .051 .050 .051irtility and Mortality (b) .092 .097 .094 .090 .087 .089 .091 .096 .093)74* (c) .125 .127 .125 .122 .115 .119 .123 .126 .126

ingladesh Fertility (d) - _ - - - - - - .075arvey, 1975-76 ** (e) - - - - - - - - .078

(f)'

.076

jtes: - Not available* Estimated by Kabir (1977 : 164)** Estimated by Mitra (1979 : 29) based on the Coale-Demeny West model life tables(a) Direct estimate(b) Based on Coale-Demeny West model life tables(c) Based on Coale-Demeny North model life tables(d) Refers to the period 1969-70(e) Refers to the period 1971-72(f) Refers to the period 1972-73

83

1 and 4 could be estimated.

With the exception of urban 4^1 values derived from the 1974 Retrospective

Survey of Fertility and Mortality, the estimates of 4* 1 obtained in various

studies were higher for males than for females (Table 4.6). The 5 year average

estimates of 4* 1 obtained in the present study were also higher for females

than for males. A comparison ofTables 4.5 and 4.6 reveals that mortality of

children between ages 1 and 4 were generally higher in rural areas than in

urban areas as found in the case of infant mortality.

4.4.2 Mortality Between Birth and Exact Age 5 (5^o)

National as well as rural and urban estimates of 5^o for children born

between the years 1955 and 1969 by single calendar years and 5 year periods

are shown in Table 4.7. As found in the cases of l^o and 4^1, singleqcalendar year estimates of 5 o have fluctuated and 5 year average rates have

clearly shown a declining trend in childhood mortality between birth and

exact age 5 during the period 1955-60 both in rural and urban Bangladesh.

Childhood mortality as measured by 5^o was generally higher for rural areas

than for urban areas. Except for the urban areas during the period 1960-64,

all the 5* 0 values based on 5-year average periods were found to be higher

for males than for females. This is in contradiction with the findings in

the case of mortality of children between ages 1 and 4. The 5^o estimates are

heavily influenced by the risk of dying under exact age 1. It was observed

earlier (Section 4.2) that mortality under exact age 1 was generally higher

for males than for females. Thus the sex differentials in childhood mortality

between birth and exact age 5 might have shown the real situation. Mortality

data compiled by the United Nations (1953 : 48) have revealed that in almost

all countries mortality for males was higher than for females with some

exceptions of females aged 1-4 years and in the childbearing ages.

Estimates of 5* 0 in the present study based on the survival of birth cohorts

34

TABLE 4.7

ESTIMATES OF PROBABILITY OF DYING BETWEEN BIRTH AND EXACT AGE 5 (5qo)

BY AREA AND BY SEX FOR BANGLADESH, 1955-69

Year of Birth of Child

Rural Urban Total (Weighted)

Male Female Both Sexes Male Female Both Sexes Male Female Both Sexes

1955 .254 .316 .284 .132 .317 .202 .237 .314 .277

1956 .275 .226 .252 .194 .173 .185 .269 .226 .249

1957 .265 .244 .254 .303 .241 .274 .268 .241 .254

1958 .274 .260 .267 .250 .275 .262 .272 .264 .265

1959 .298 .212 .254 .229 .149 .188 .293 .207 .247

1960 .207 .228 .218 .195 .278 .236 .207 .230 .218

1961 .286 .238 .262 .173 .245 .209 .276 .239 .258

1962 .271 .204 .238 .223 .286 .251 .266 .212 .239

1963 .197 .215 .206 .198 .119 .158 .198 .208 .203

1964 .223 .225 .224 .200 .159 .181 .223 .219 .221

1965 .190 .234 .209 .130 .176 .153 .179 .228 .204

1966 .232 .208 .220 .226 .215 .221 .230 .209 .220

1967 .223 .185 .205 .205 .167 .187 .221 .183 .204

1968 .190 .205 .197 .189 .144 .167 .190 .202 . 196

1969 .232 .206 .219 .182 .152 .168 .229 .203 .216

Grouped Data (5-Year Average)

1955-59 .274 .248 .262 .222 .221 .222 .268 .247 .258

1960-64 .235 .222 .229 .200 .214 .206 .233 .221 .227

1965-69 .213 .211 .210 .186 .170 .179 .210 .205 .208

Source: Pregnancy history data, Bangladesh Fertility Survey, 1975-76

FOR

B

AN

GL

AD

ESH

FR

OM

D

IFF

ER

EN

T

SOU

RC

ES

n30

Otr1

LD

wo<

PmOCOwE-*gHEHCOW

CO03X0)CO o m

i l l 04 1 04 COXJ 04 04 Ol4-1 • • .O

1—1 03rdC 030 rH co cn tn CO

•H rd r—l O l—1 1 H 1 14-3 e Ol 04 Ol Ol03 03 1 • • .2 P^

03 CN O ' rH or—1 m o j ro 1 H | 1rd CM 04 04 04s . . •

co03X03CO Ol

1 1 1 Cn 1 1 1x : H4-1 •o03

Co3

XI 03 OSH 1--1 1 1 1 1 O ' 1 1D o3 H

e •03

Pm

03 ini—1 1 1 1 1 O ' 1 105 HS •

co03X03CO CD

1 1 1 04 1 1 1X3 Ol4-3 •o03 .

i—i03 0)SH i—1 OlP o3 1 1 1 1 Ol 1 I04 e 04

0) •Pm

0) CDi- 1 1 1 1 1 H 1 103 04

2

4e * * *C i—1 C H03 H rd H> 0) > 03

CO -H CO CO CO -H COCO rH CO CO CO H CO( O H 3 rc3 rd H PSh P L Sh U P U03 CO Em 03 03 CO Eh

O h J h

0 4-J•H

pH H Ph N0) <t5 0) 4-3> 4J > •Hi n H Sh HP 0 P rOX3 ■H U

CO S CO C d - 4-> CDrd e '­ SH O '

T3 03 TJ 03 en 03 10 > C > tH H Pm cn

x ; •H cc3 •H 4-3 O '4-1 -P 4-1 •H ** x: co0) Ü >i U H PH CO Hs 03 4-1 03 •H 4-1 03\ O , H & 4-3 -H TS '0) CO H CO M i—1 rd > iC3 0 "H ad 0 03 cd H 03SH Sn 4-) Sh Pm 4-) Cn P>P 4J H h 4-1 Sh C U0 03 03 cn 03 Oh 0 rd PCO OO Pm H 04 0 £ 03 CO

incdrH1

I (d X3 Ü *

co03-P02

on

Wes

t M

odel

li

fe

ta

ble

s

86

are compared with those estimated from different available sources using the

methods of Brass, Sullivan and Trussell as shown in Table 4.8. It should be

noted that the Brass method is based on the assumption of constant fertility

and childhood mortality in the recent past and the methods of Sullivan and

Trussell are modifications of the Brass method (National Research Council, n.d.

: 6-2; Sembajwe, 1977 : 127). On the average, the Brass estimate of 5^o

refers to the mortality level 6.5 years before the survey (Sullivan, 1972 : 82).

The 5^o estimates shown in Table 4.8 might be taken to approximately refer to

the childhood mortality level in the late 1960's corresponding closely to the

estimates for the period 1965-69 in this study. It can be observed from

Table 4.8 that the national estimates of 5^o both for males and females based

on the Sullivan's method were lower than those based on the methods of Brass and Trussell. The national estimates of 5^o for the period 1965-69 derived in this study were even lower than the corresponding estimates based on Sullivan's method. However, it should be noted here that the 5^o values for

both males and females in urban areas estimated by Mitra (1979) have agreed closely with the corresponding estimates derived in the present study

(Tables 4.7 and 4.8).The 5^o estimates derived from pregnancy history data are likely to be less

influenced by the mis-statement of age at death (Potter 1977a : 352; Asare,

1980 : 125). However, these estimates were, in general, found to yield lower

values as compared to the corresponding estimates based on the methods of Brass,

Sullivan and Trussell. As pointed out earlier, the 5^o estimates are heavily influenced by the infant deaths and infant mortality based on pregnancy history

method and are under-estimated. This is probably the reason for the lower

estimates of 5^o obtained in this study.Consistent with the general findings of the sex and rural-urban different­

ials in the present study, all other studies have found that the mortality of

children between birth and exact age 5 was lower for females than for males and

the 5^o estimate was lower in urban areas than in rural areas.

CHAPTER 5 87

SUMMARY AND CONCLUSION

This study has been based on the pregnancy history records from the

Bangladesh Fertility Survey, 1975-76. The pregnancy history method has some

advantages over a number of indirect methods for estimating fertility and

mortality. However, this method has been found to suffer from a substantial

amount of response errors, such as, age misreporting, under-reporting and

misplacement of events in time (Chapter 2). Keeping these errors in mind,

the objectives of the study were (1) to estimate fertility and infant and

childhood mortality rates for Bangladesh; and (2) to explore whether the

pregnancy history data were well reported for achieving- the first objective.

In Chapters 3 and 4, some general and specific findings were discussed.

Attempts will be made in this chapter to summarize these findings and to draw some conclusions.

Fertility in Bangladesh was found to have declined substantially during the period 1971-75. The social disruptions caused by the 1971 war of liberation and the 1974 widespread famine might have been the reasons for this decline. However, the observed dip in the levels of fertility might

probably be only temporary and fertility might revert to its previous roughly

constant trend (Amin and Faruqee, 1980 : ii). Evidence from the former

Cholera Research Laboratory vital registration data in Matlab Thana has

indicated that the total fertility rate has been increasing since the year

1975"*”. It was observed that fertility in the period 5-10 years before the

survey (1966-70) was somewhat exaggerated, a feature mainly associated with

1 Total fertility rate for Matlab Thana were found to be 4.3, 6.2 and 6.7 in the years 1975, 1976 and 1977 respectively (Ruzicka and Chowdhury, 1978a : 9; Ruzicka and Chowdhury, 1978b : 9; Samad et al., n.d. : 20).

88

the misplacement of earlier births forward and the recent births backwards in time while reporting the events. Observed fertility rates beyond 15 years

prior to the date of survey have been very doubtful because the pregnancy

history data in the relatively distant past are more likely to be affected by

various response errors, especially omission and misplacement of events.

Interviewing the respondents from the most recent to the earliest event for

collecting pregnancy history data as suggested by Potter (1977a : 364),

".... would minimize the potential for distorting recent trends in fertility."

A consistent rural-urban fertility differential was found in Bangladesh.

Fertility was lower in urban areas than in rural areas. Apart from sampling

variations due to a relatively small number of the respondents in urban areas,

urban fertility appeared to be better reported. Fertility schedules from different studies in Bangladesh were generally characterized by a peak in

the age group 20-24. This was mainly due to an early age at marriage. On the whole, fertility rates derived in this study might have been affected by the

procedure of imputing the dates of birth of about 99 per cent of the respondents and 88 per cent of their children that were not reported in calendar years

and months.

The applicability of the suggested procedures for adjusting fertility

rates derived from pregnancy histories were discussed in the context of

Bangladesh (Chapter 3). Brass's (1971, 1975) first birth procedure, which

uses the distribution of first births in the 0-4 years before the survey as the

standard for adjusting all order births, might not be applicable in the case

of Bangladesh because of the possible effects of the 1971 war of liberation

and the 1974 famine on fertility in the 0-4 years preceding the survey. It

has also been demonstrated by Potter (1977a : 347) that Brass's first birth

procedure would not be effective for detecting and adjusting fertility rates

for event misplacement errors. Potter has argued that the timing errors in

89

the reporting of first births are not the same as those for all order births.

Brass and Rashad (1980) have applied the P/F ratio method and the

Gompertz relational model for adjusting the observed fertility rates derived

by using the pregnancy history data from the Bangladesh Fertility Survey, 1975-76.

They have found the adjusted total fertility rates for the period 1971-75 to be

about 7.5 and 7.4 using the P/F ratio method and the Gompertz relational model

respectively. The national estimate of total fertility of 5.9 derived in the

present study during the period 1971-75, which has closely agreed with the

corresponding estimate obtained by Rodriguez and Cleland (1980 : 23), has been

about 25-27 per cent lower than the adjusted estimate during the same period.

Brass (1980 : 20) was also doubtful about the levels of fertility during the

period 1971-75 as found by the application of the adjustment procedures.

Infant mortality rates have shown, in general, a downward trend during

the period 1955-69. A temporary halt in the declining trend in infant mortality

was observed during the period 1970-73. The devastating cyclone in 1970 and

the liberation war in 1971 might have been the reasons for the temporary rise

in infant mortality. Available evidence has demonstrated a further rise in

infant mortality in the year 1974 associated with the crop failure due to

monsoon floods (Chapter 4). A comparison of infant mortality rates from various

sources revealed that these rates based on the survival of birth cohorts from

pregnancy history data in the early 1970's were lower than the corresponding

estimates obtained by the Brass technique based on the survival of children

ever born. Infant mortality rates from pregnancy history data are suspected

to be under-reported by at least 10 per cent in the 5 year period preceding

the survey (Bogue and Bogue, 1980 : 154). The levels of infant mortality per

thousand live births in the early 1970's as found in different studies using

the Brass technique ranged from 152 to 158. The national estimate of infant

mortality of 139 during the period 1970-73 derived in this study was about

90

9 to 14 per cent lower as compared to the Brass estimates. Nevertheless, the

estimates found in the present study closely agreed with the corresponding

estimates derived by Arriaga (1980) from the pregnancy history data (Chapter 4).

Although childhood mortality rates between birth and exact age 5 have, in

general, shown a declining trend during the period 1955-69, the corresponding

rates for children aged 1-4 have not demonstrated such a crend except for the

urban areas. The national and rural estimates of 4^1 in the period 1960-64

appeared to be somewhat lower than what would be expected. The national

estimate of childhood mortality between birth and exact age 5 of 0.208 during

the period 1965-69 was also about 6 to 13 per cent lower than the corresponding

estimates obtained in other studies using the different methods of estimation

(see Chapter 4). The 5* 0 estimates are influenced by the risks of dying under

exact age 1 and, as mentioned above, infant mortality rates from pregnancy

history data are under-estimated. These might be the reasons for the

apparent lower 5^o estimates derived in the present study during the period

1965-69.

Infant and childhood mortality were generally found to be lower in urban

areas than in rural areas. This presumably has reflected the better health

and medical facilities in urban areas. Infant mortality was found to be

generally higher for males than for females. This has demonstrated the

universally higher biological risks of male death in infancy. 5^o estimates

also have shown such a sex differential. On the other hand, mortality of

children aged 1-4 was found to be generally higher for females than for males.

Higher female than male mortality of children between ages 1 and 4 was found

in a number of South Asian countries (Chapter 4).

Finally, it was realised that the pregnancy history data used in this

study were not well-reported and well-imputed. This has caused some distortions

in the levels and trends in fertility and childhood mortality. Nevertheless,

91

this study has provided a fair idea about the levels and trends in fertility

and childhood mortality in Bangladesh. It may be suggested that in a country

like Bangladesh, where a vast majority of women (respondents) are not calendar

year conscious and even ignorant of their dates of birth, the pregnancy history

data should be carefully edited for any discrepancies before analysis.

It may also be suggested that the analysis of fertility and childhood mortality

from pregnancy history data, for a developing country like Bangladesh should

not extend beyond the 15 years before the date of survey. Further research

should be carried out to study the possible effects and interactions of various

response errors on the estimates of fertility and childhood mortality.

APPENDIX TABLE A 192

PREGNANCY HISTORY TABLE

308. INTERVIEWER: USE ONE ROW OF THE PREGNANCY HISTORY TABLE FOR EACH PREGNANCY,STARTING WITH THE FIRST AND PROCEEDING CHRONOLOGICALLY, (THAT IS, THE FIRSTPREGNANCY IN THE FIRST LINE, THE SECOND PREGNANCY IN THE SECOND LINE......UP TO THE LAST PREGNANCY). IF TWINS, USE ONE LINE FOR EACH AND CONNECT WITH A BRACKET AT THE LEFT.

THE FOLLOWING QUESTION SEQUENCES ARE TO BE ASKED TO RECORD PREGNANCY HISTORIES FOR THE FIRST SEGMENT, THAT IS, THE INTERVAL BETWEEN MARRIAGE AND THE FIRST LIVE BIRTH.

IF RESPONDENT HAS HAD ANY LIVE BIRTH:

309. Now I want to ask you some questions about each of your pregnancies, that is each live birth, each still birth, each miscarriage and each abortion.If you have had any of these or any children who have died or who live away from home I would like to know about them.

What is the name of your first baby born alive _____________________________(Name)

310. After you (first) were married and before __________________________________was born, did you have any other pregnancies? (Name)

YESNOW ASK 319-323 AND RECORD FOR THE FIRST PREGNANCY, REPEAT 319-323 FOR EACH PREGNANCY UNTIL THE FIRST LIVE BIRTH IN THIS INTERVAL. THEN REPEAT NAME OF FIRST BABY BORN ALIVE UNDER 313 IN THE NEXT ROW,ASK 315-318 AND RECORD

NOW ASK 313 AND 314 AND ACCORDINGLY PROCEED TO EITHER 315-318 OR 319-323. REPEAT SEQUENCE FOR ALL REMAINING PREGNANCIES.

NOENTER THE NAME OF FIRST LIVE BIRTH BIRTH UNDER 313, ASK 315-318 AND RECORD

NCW ASK 313 and 314 AND ACCORDINGLY PROCEED IN THE NEXT ROW TO EITHER 315-318 OR 319-323. REPEAT SEQUENCE FOR ALL REMAINING PREGNANCIES.

IF PREGNANCY OCCURRED TO THE RESPONDENT IN THE PAST BUT NOT RESULTED IN LIVE BIRTH:

311. Think back to your first (second, ......... etc.) pregnancy and tell meabout it.

INTERVIEWER: NOW ASK 319-323, REPEAT SEQUENCE FOR EACH PREGNANCY.

IF NO PAST PREGNANCY OCCURRED TO THE RESPONDENT:

(Skip to 333)

t t t ' i ' t W U i A X>it5JjU il ± . /

P R E G N A N C Y H I S T O R Y T A R L E

312 313 314 I F A L I V E BIRTH

P r e g - W h a t was t h e A f t e r 315 31Ö 3 1 7 3 1 8n a n c yO r d e r

name o f y o u r f i r s t / n e x t b a b y b o r n

( n a m e ) a nd b e f o r e

wasI n w h a t m o n th and y e a r was

Was t h e b a b y a

I s t h e c h i I d s t i l l l i v i n g ?

( I f n o t a l i v e ) How l o n g d i d

a l i v e ? ( n a m e ) b o y o r a t h i s b o y / g i r lb o r n d i d y o u h a v e a ny o t h e r p r e g ­n a n c i e s ?

( n a m e ) b o r n ?

g i r l ? l i v e ?

YES Q ](SKIP TO 319)

B o n g / E n g

MONTH BOY Q J

YES □ (PROCEED WITH NEXT PREGNANCY

MONTHS

01 NO j j J YEAR YEARS

(ASK 3 1 5 - 3 1 8 ) YEARSAGO _ _ _ _

GIRL [7 ] no on(ASK 3 1 8 )

ASK THE NEXT PREGNANCY

YES □

(SKIP 10 319)

B e n g / E n g

MONTH BOY Q

YES 171(PROCEED WITH NEXT PREGNANCY) NO 1 1

(ASK 3 1 8 )

MONIHS02 NO G O YEAR YEARS

(ASK 3 1 5 - 3 1 8 ) YEARSAGO

GIRL [_2_l ASK THE NEXT PREGNANCY

YES QJ (S K 1 P TO 319)

B e n g / E n g

MONTH BOY f 1 1

YES [7) (PROCEED WITH NEXT PREGNANCY) NO ID(ASK 3 1 8 )

MONTHS

03 MO i l l YEAR YEARS

(ASK 3 1 5 - 3 1 8 ) YEARSAGO

GIRL [ I ASK THE NEXT PREGNANCY

■-

YES l j j

(SKIP 10 319)

B e r i g / E n g

MONTH BOY 171YES m (PROCEED WITH NEXT PREGNANCY) NO [ 2 j

(ASK 3 1 8 )

MONTHS

04 NO I 2 ] YEAR YEARS

(ASK 3 1 5 - 3 1 8 ) YEARSAGO

GIRL [71 ASK THE NEXT PREGNANCY

YES [7] (SKIP TO 319)

B e n g / E n g

MONTHr ——f

BOY ! _ j j

yes m(PROCEED WITH NEXT PREGNANCY) NO j _2 j

(ASK 3 1 8 )

<

MONTHS

05 NO | 7 ] YEAR YEARS.

(ASK 3 1 5 - 3 1 8 ) YEARSAGO

GIRL [ 7 | ASK THE NEXT PREGNANCY

YES [ i ]

(SKIP 10 319)

B e n g / E n g

MOUTH BOY l j j

YES IT| (PROCEED WITH NEXT PREGNANCY)NO [7 (ASK 318 )

1MONTHS

06 NO i 2 i YEAR YEARS1

(ASK 3 1 5 - 3 1 3 ) YEARS AGO - -

GIRL [7) ASK THE NEXT PREGNANCY

yes QJ(SKIP TO 319)

B e n g / E n g

MONTH BOY [7]yes r n(PROCEED WITH NEXT PREGNANCY

MONTHS

O7 NO i l l YEAR YEARS

(ASK 3 1 5 - 3 1 0 ) YEARSAGO

GIRL | 7 ] NO \T \

(ASK 3 1 8 )ASK THE NEXT PREGNANCY

APPENDIX TABLE A 1 (CONT.)94

P R E G N A N C Y H I STORY TABLEIF NOT A LIVE BIRTH

“3l97Tn vvhat Mth.and Yr. was that preg termi­nated? If 1D.K.' ask how many Yrs. ago?Beng/Eng MONTHS...YEARS. . . .YEARSAGO....BEnT/Eng" MONTHS...YEARS. . . .YEARSAGO.....Beng/Eng MONTHS...YEARS. . . .YEARSAGO.....Beng/Eng MONTHS...YEARS. . . .YEARSAGO.....Beng/Eng MONTHS...YEARS. . . .YEARSAGO.....Beng/Eng MONTHS...YEARS. . . .YEARSAGO.....Benq/Eng MONTHS...YEARS. . . .YEARSAGO.....

7 2 0 .How many months did that pregnancy las t?

LESS THAN 70(SKIP TO 323) 7 0R_ MORE [2(7 K 321) 1L0S T1TALT 70(SKIP TO 323) 7 OR MORE [YJ(ASK 321) LESS THAN 7

1(SKIP TO 323) 7 OR MORE M(ASK 32T)LESS THAN 70(SKIP TO 323) 7 OR MORE(ASK 321LESS THAN 7

(SKIP T 323) 7 OR MORE f~2(ASK 3 2 f T LESS THAN 70(SKIP TO 323) 7 OR MORE |?-j(ASK 321) LESS THAI! 70(SKIP TO 323) 7 OR MORE |7j^ASiL li_

371. Did tTTe~ baby show any sign of life (cried etc.) after its birth?

yes jrj(ASK 322) NO [2](PROCEED WITH NEXT PREG)_ _ _ _YES [l] (ASK 372) NO [2]iSMlPREGy e s rr](ASK 322) NO [2(PROCEEDMEETYES F ] (ASK 172) NO [2]imih.P-RLG)___YES [l ](ASK 372) NO [2]m W m TP_REG)_YES [7](ASK 372)NO [2 J(PROCEED WITH NEXT

___YES [T] (ASK 322) NO 1 7 PROCEEDi ' NEXTi)

377. Was. the baby a boy or a girl?

BOY 0GIRL [2 PROCEEi WITH NEXT PREG.BOY [7] YES [7]GIRL [7] NO [2]PROCEED PROCEEDWITH NEXT PREG.BOY [T] GIRL 7]PROCEED WITH NEXT PREG.BOY

WITH NEXT PREG.BOY [f]GIRL [7] PROCEED WITH NEXT PREG.BOY □GIRL PROCEEI WITH NEXT _RRLGL____

0:ed

BOY [T YES 'GIRL 17 NOPROCEED PROCEWITH NEXT WITHPREG.

323.Did you or a doctor or someone else do any­thing to end the preg­nancy?

YES 0NO [2] PROCEED WITH NEXT PREG.

WITH NEXT ISESu___0YES

NO 12 PROCEED WITH NEXT P R F . G .

0 YES 1

0 NO0NEXT

PREG.YES 0NO [2]PROCEED WITH NEXT PREG.

03 LD1 3

5 9ML FL TMLB TFLBu m m m11 13 15 17

19 21 22

32 34 35

45 47 48

58 60 61

□ *119 21 22

VE5 [T]NO |2 I PROCEED WITH NEXT EREIL____

NEXTPREG.

n

45 47 48

□ □

0

□ □

26 27 28 30

39 40 41 43

52 53 54 56

65 66 67 69

26 27 28 30

o m m32 34 35 39 40 41 43

□052 53 54 56

APPENDIX A 2SINGLE YEAR AGE DISTRIBUTION OF THE RESPONDENTS BY

AREA IN BANGLADESH, 1975'-76

Age of Women Rural Urban Total (Weighted)

9 1 - 110 4 1 511 13 2 1612 32 4 4013 65 13 8214 100 21 12715 154 36 19616 167 45 21517 200 52 25718 228 66 29519 183 68 24220 233 82 30721 181 72 24122 205 59 26523 187 68 24724 223 61 28725 187 60 24426 186 66 24527 184 45 23528 165 51 21529 132 34 16930 135 52 17931 108 34 14132 120 51 16133 98 30 12734 142 42 18435 123 37 16036 106 32 13837 91 23 11738 107 33 13939 92 26 11940 100 42 13441 89 21 11442 86 22 11043 97 22 12344 112 31 14445 107 26 13746 75 17 9547 73 19 9448 89 18 11349 44 13 57

Total 5024 1497 6517

Source: Bangladesh Fertility Survey, 1975-76.

1941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974

APPENDIX A396

LIVE BIRTHS FOR 5 YEAR AGE GROUPS OF THE RESPONDENTS BYSINGLE CALENDAR YEAR IN RURAL BANGLADESH, 1941-76

Number of Reported Live Births

)—14 15-19

Age Group of Women

20-24 25-29 30-34 35-39 40-44 45-49

2215 8 - - - - - -22 34 - - - - - -23 39 - - - - - -27 73 - - - - - -14 94 - - - - - -24 101 24 - - - - -24 108 40 - - - - -16 119 63 - - - - -19 143 80 - - - - -26 143 138 - - - - -31 109 138 20 - - - -23 138 124 40 - - - -24 138 145 74 - - - -39 138 159 73 - - - -34 154 140 114 - - - -31 147 142 133 16 - - -37 153 164 127 35 - - -32 175 139 154 43 - - -42 170 176 147 67 - - -45 184 161 161 115 - - -43 161 196 156 106 11 - -47 229 188 158 139 27 - -51 239 219 169 128 50 - -58 255 201 175 144 46 - -60 310 233 190 137 89 - -56 296 226 203 131 81 8 -43 310 256 201 169 83 13 -49 310 287 192 123 93 21 -60 305 309 209 151 73 23 -47 302 322 198 134 96 25 -40 248 303 155 124 62 23 130 267 323 191 113 61 23 034 256 304 178 127 53 17 320 259 280 202 109 47 17 11 32 34 20 16 5 1 0

- Not available

Pregnancy history data, Bangladesh Fertility Survey, 1976-76.

194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973

APPENDIX A 4

LIVE BIRTHS FOR 5 YEAR AGE GROUPS OF THE RESPONDENTS BY

SINGLE CALENDAR YEAR IN URBAN BANGLADESH, 1941-76

Number of Reported Live Births

10-14 15-19 20-24

Age Group

25-29

of Women

30-34 35-39 40-44 45-49

23 54

45

10-

-- -

- -

7 16 - - - - - -13 18 - - - - - -11 21 4 - - - - -7 28 13 - - - - -3 28 16 - - - - -

13 32 21 - - - - -12 36 47 - - - - -18 42 35 5 - - - -8 47 41 11 - - - -15 33 32 18 - - - -8 38 35 28 - - - -5 46 37 30 - - - -9 42 36 32 5 - - -10 41 48 36 10 - - -12 57 49 39 13 - - -13 49 42 34 23 - - -12 64 55 32 32 - - -19 75 59 54 25 3 - -15 53 58 45 26 5 - -15 70 75 48 30 10 - -17 82 78 70 27 14 - -19 73 72 51 45 16 - -20 86 74 54 28 14 2 -21 58 77 57 42 17 3 -19 91 76 67 40 13 4 -17 91 77 60 37 21 5 -16 93 105 58 44 24 6 -9 90 87 63 28 14 4 014 93 81 62 32 13 5 09 90 87 63 28 14 4 09 84 118 64 30 7 6 10 12 10 6 3 1 0 0

- Not available

Pregnancy history data, Bangladesh Fertility Survey, 1975-76.

1941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974

oq

APPENDIX A 5

LIVE BIRTHS FOR 5 YEAR AGE GROUPS OF THE RESPONDENTS BY

SINGLE CALENDAR YEAR IN BANGLADESH (TOTAL WEIGHTED SAMPLE), 1941-76

Number of Reported Live Births

10-14 15-19 20-24

Age Group

25-29

of Women

30-34 35-39 40-44 45-49

2719 11

- - - - - -

28 42 - - - - - -29 50 - - - - - -33 96 - - - - - -21 118 - - - - - -32 128 30 - - - - -31 139 52 - - - - -20 152 81 - - - - -27 182 103 - - - - -33 183 181 - - - - -43 145 177 26 - - - -30 181 162 52 - - - -34 176 184 95 - - - -49 178 202 97 - - - -42 200 180 147 - - - -40 190 182 170 21 - - -48 197 212 164 45 - - -42 229 183 197 56 - - -55 220 225 187 88 - - -58 242 211 203 148 - - -38 218 254 205 135 14 - -61 292 245 204 175 34 - -66 310 288 218 163 63 - -75 333 267 233 181 60 - -78 395 303 245 179 112 - -74 383 296 261 166 102 10 -59 390 332 260 216 105 17 -65 402 369 252 161 116 26 -78 396 396 270 193 94 29 -62 393 421 257 175 123 32 -51 327 392 207 158 79 29 141 351 414 250 146 77 29 044 337 401 230 164 67 22 427 338 375 263 141 59 22 21 42 44 26 20 6 1 0

- Not available

Pregnancy history data, Bangladesh Fertility Survey, 1975-76.

PERCENTAGE A

GE D

ISTRIBUTION

OF E

VER-MARRIED

FEMALES

IN B

ANGLADESH

100

APPENDIX C 1

P/F RATIOS FOR DIFFERENT PERIODS PRECEDING THE SURVEY

UNDER THE ASSUMPTION OF CONSTANT FERTILITY : BANGLADESH

Age P F P/F

1975 1971-75 1966-70 1975 1971-75 1966-70

15 - .585 .480 .553 .884 1.218 1.057 .662

20 - 2.344 1.709 1.996 2.685 1.371 1.174 .873

25 - 4.179 3.071 3.451 4.485 1.361 1.211 .932

30 - 5.686 4.155 • 4.702 6.144 1.368 1.209 .925

35 - 6.635 4.873 5.626 7.348 1.361 1.179 .903

40 - 7.148 5.270 6.163 8.079 1. 356 1.159 .885

45 - 6.466 5.370 6.336 - 1.204 1.020

P/F RATIOS FOR DIFFERENT PERIODS PRECEDING THE SURVEY

UNDER THE ASSUMPTION OF CHANGING FERTILITY : BANGLADESH

Age 1966-70 1961-65

15 -

20 -

25 -

30 -

35 -

40 -

1.021.01.99

.93

.90

.78

.99

1.01.94

.92

.80

Source: Brass and Rashad (1980 : 15)

P/F

RATIOS B

Y BIRTH

ORDER FOR

RECENT P

ERIODS : B

ANGLADESH

101

m<u

i>H i—i o CO i—1 m rH oo fOO O' o o i—( i— i i— 1 o OO

oo • • • . i • •+ r—1 1— 1 1— 1 rH i— i i—1 1—1

uoO'

Pm i\ i—i o~ 1—1 00 O'PM >H O' o CM 04 04 04

(1) OO • • • • •Ttf t—i r—1 1—1 rH rH rH>hO4-1COU•HPm in ID oo O' O CO

O' 1—1 04 04 CO 04oo • • t 1 •i—i 1—1 i—1 rH 1—1 i—1

LOO'i r- i— 1 m CO O ID rH

Pm i— i o rH CO ID UO O'r» o CO 04 CO CO o

CO co • • • • • • •X i— i rH 04 CO CO4->P4•HCQ

+r-~ »— 1 04 O' ID rH OOo 04 CO CO CO

CM o CO CO O' 04 O'

rH 04 CO C0

inO'i oc 04 ID uo 00 o .—r— 1 o OO r— 1 m in IDr- o CO O' O' O' O' rHoo • • t • • •r— 1

oto Pm 00Ä OO4-) rHL —•H o OO ID CO CO COOQ in 1— 1 m r- uo uo uo TO

e'­ o co ID O' O' O' rtf4J er» • • • • • •CO i— 1 10L rtf•H OrJPH

artfCO o CO 04 O' O' CO 04o 04 O' ID 0' ID CO CO

CM o CO 00 Co 00 OO CO to. . • • • • • • rtf

UCQ

CDU

1 i 1 1 1 1 1 i P4(1) 3tP o m o in o UO o uo 0< 1— 1 i— i 04 04 CO CO CO

APPENDIX D 1 102

ESTIMATES CF PROBABILITY OF DYING BETWEEN BIRTH AND EXACT AGE x,

0 ii 2,3,10) BY AREA FOR BANGLADESH , 1955-72

Year of Birth of Child

Rural Urban Total (Weighted)

2qo 3qo 10qo 2qo 3qo 10qo 2qo 3qo 10qo

1955 .225 .240 .313 .138 .165 .229 .219 .234 .306

1956 .209 .221 .277 .134 .143 .212 .205 .215 .273

1957 .200 .217 .288 .210 .250 .306 .201 .219 .289

1958 .211 .242 .293 .207 .221 .297 .210 .240 .294

1959 .186 .217 .291 .165 .176 .191 .184 .214 .283

1960 .176 .198 .257 .186 .217 .248 .177 .199 .257

1961 .210 .238 .283 .153 .179 .235 .205 .233 .279

1962 .189 .202 .275 .200 .221 .272 .190 .203 .277

1963 .168 .184 .230 .124 .129 .178 .165 .181 .226

1964 .165 .190 .257 .141 .153 .206 .163 .188 .254

1965 .160 .182 - .128 .139 - .156 .178 -1966 .172 .192 - .156 .181 - .170 .192 -1967 .156 .180 - .147 .155 - .156 .178 -1968 .143 .161 - .142 .153 - .144 .161 -1969 .153 .189 - .113 .135 - .151 .185 -1970 .167 .188 - .156 .172 - .167 .188 -1971 .171 .200 - .133 .171 - .168 .198 -1972 .174 - - .142 - - .170 - -

Note; - Not available

Source: Pregnancy history data, Bangladesh Fertility Survey, 1975-76.

REFERENCES103

AHMED, Bashir U.

AHMED, Feroz

AMIN, R. and FARUQEE, R.

1978. "Differential Fertility in Bangladesh". Unpublished M.A. in Demography Thesis. Canberra : Australian National University.

1970. "An Evaluation of the Vital Rate Estimates of Pakistan". Unpublished paper presented at the annual meeting of the Population Association of America, April, 1970.

1980. Fertility and its Regulation in Bangladesh. Washington D.C. : Population and Human Resources Division, The World Bank.

ARRIAGA, Eduardo E. 1980. "Direct Estimates of Infant Mortality Differentialsfrom Birth Histories". A paper presented at the World Fertility Survey Conference, London, July 7-11.

ASARE, Kofi A. 1980. "Estimating Fertility and Childhood Mortality forthe Western Region of Ghana from Pregnancy History Data". Unpublished M.A. in Demography Thesis. Canberra : Australian National University.

BANGLADESH BUREAU OF 1977. National Volume, 1974 : Bangladesh Population STATISTICS Census Report. Dacca : Statistics Division, Government

of the People's Republic of Bangladesh.

BOGUE, D.J. and 1970. Techniques of Pregnancy History Analysis. ChicagoBOGUE, E.J. Community and Family Study Center, University of

Chicago.

BONGAARTS, John 1980. "The Fertility Inhibiting Effects of theIntermediate Fertility Variables". A mimeographed paper presented at the Seminar on the Analysis of Maternity Histories. London, April 9-11.

BRASS, William 1971. The Analysis of Maternity Histories to DetectChanges in Fertility, United Nations Economic and Social Council (E/CN.9/AC.12/R.11), New York : United Nations.

---------------------- 1975. Methods of Estimating Fertility and Mortality fromLimited and Defective Data. Chapel Hill, N.C. Laboratories for Population Statistics, The University of North Carolina.

---------------------- 1977. "The Assessment of the Validity of Fertility TrendEstimates from Maternity Histories". International Population Conference. Mexico : International Union for the Scientific Study of Population. Vol. I, pp. 89-100.

---------------------- 1979. "Screening Procedures for Detecting Errors inMaternity History Data". Regional Workshop on Techniques of Analysis of World Fertility Data : Report and Selected Papers. United Nations, Economic and Social Commission for Asia and the Pacific, Bangkok, pp. 15-36.

1980. "Birth History Analysis". A paper presented at the World Fertility Conference. London, July 7-11.

JLU 4

BRASS, William and RASHAD, Hoda

1980. "Exploratory Demographic Analysis of Imperfect Maternity Histories to Determine Levels and Trends of Fertility". A mimeographed paper presented at the Seminar on the Analysis of Maternity Histories. London, April 9-11.

CAIN, Mead T., NORRIS, Douglas, and SIRAGELDIN, Ismail

1979. "Pre-independence Levels of Fertility : Evidence from the Pakistan National Family Planning Impact Survey, 1968-1969". In M. Hossain et al. (ed.) Fertility in Bangladesh : Which Way Is It Going? Dacca : National Institute of Population Research and Training, pp. 1-9.

CENSUS COMMISSION 1977. Report on the 1974 Bangladesh Retrospective Surveyof Fertility and Mortality. Dacca: Statistics Division,Government of the People's Republic of Bangladesh.

CHANDRASEKAR, S. 1959. Infant Mortality in India, 1901-55. London :George Allen & Unwin Ltd.

CHEN, Lincoln C. and 1977. "The Dynamics of Contemporary Famine." CHOWDHURY, A.K.M.A. International Population Conference. Mexico :

International Union for the Scientific Study of Population. Vol. I, pp. 409-426.

CHEN, Lincoln C. and 1975. "Demographic Change and Food Production in CHAUDHURY, Rafiqul Huda Bangladesh". Population and Development Review, Vol. I,

No. 2, pp. 201-227.

CHIDAMBARAM, V.C., CLELAND, J.G., GOLDMAN, N. and RUTSTEIN, S.

1980. "An Assessment of the Quality of WFS Demographic Data". A mimeographed paper presented at the Seminar on the Analysis of Maternity Histories. London,April, 9-11.

CHOWDHURY, A.K.M.A. and 1979. "Recent Trends in Fertility and Mortality CURLIN, George T. in a Rural Bangladesh". In M. Hossain et al. (ed.)

Fertility in Bangladesh : Which Way Is It Going?Dacca : National Institute of Population Research and Training. pp. 110-119.

COALE, Ansley J. and TRUSSELL, T.James

1974. "Model Fertility Schedules : Variations in the Age Structure of Childbearing in Human Populations"Population Index. Vol. 40, No. 2, pp. 185-258.

CURLIN, G.T., CHEN. L.C. and HUSSAIN, S.B.

1976. "Demographic Crisis : The Impact of the Bangladesh Civil War (1971) on Births and Deaths in a Rural Area of Bangladesh". Population Studies, Vol. 30, No. 1 pp. 87-105

DAVIS, K. and BLAKE, J.

1956. "Social Structure and Fertility : An Analytical Framework". Economic Development and Cultural Change. Vol. 4, pp. 211-235.

FAROOQUI, M.N.I. and 1971. Final Report of the Population Growth Estimation FAROOQ, G.M. Experiment, 1962-65. Dacca : Pakistan Institute of

Development Economics.

1 U wl

FREEDMAN, Ronald 1961. "The Sociology of Human Fertility, a Trend Report and Bibliography". Current Sociology. Vol. 10/11,No. 2, pp. 35-121.

FREJKA, Tomas 1974. Bangladesh Country Prospects. New York : Population Council.

GLASS, D.V. and GREBENIK, E.

1954. The Trend and Pattern of Fertility in Great Britain : A Report on the Family Census of 1946. London : Her Majesty's Stationary Office. Papers of the Royal Commission on Population, Vol. 6.

GREENOUGH III, William B. 1973. "Post-Civil War in Bangladesh : Health ProblemsandCASH, Richard A.

and Programs". In Lincoln C. Chen (ed.) Disaster in Bangladesh. London : Oxford University Press, pp. 214-254.

HULL, T.H. 1975. "Each Child Brings Its Own Fortune". Unpublished Ph.D. Thesis. Canberra : Department of Demography. Australian National University.

KABIR, M. 1977. "Levels and Patterns of Infant and Child Mortality in Bangladesh". Social Biology. Vol. 24, No. 4, pp.158-165

KHAN, A.R. 1972. The Economy of Bangladesh. London : Macmillan.

KHAN, M.R. 1973a. "Demographic Profile of Bangladesh". Proceedings of the Seminar on Family Planning. November 21-25, 1972. Dacca : Bangladesh Government Press.

■ 1973b. "Bangladesh Population During the First Five Year Plan Period (1973-1978) : An Estimate". The BangladeshEconomic Review. Vol I, No. 2, pp. 186-198.

KPEDEKPO, G.M.K. 1976. "Age Patterns of Fertility in Selected African Countries". Jimlar Mutane. Vol. I, No. 1, pp. 9-26.

MCDONALD, Peter F. YASIN, Mohammad, and JONES, Gavin W.

1976. Levels and Trends in Fertility and Childhood Mortality in Indonesia. Jakarta : Lembaga Demografi, Fakultas Ekonomi, Universitas Indonesia.

MITRA, S.N. 1979. "Infant and Childhood Mortality in Bangladesh - Levels and Differentials". Unpublished M.A. in Demography Thesis. Canberra : Australian National University.

MOSLEY, W. Henry and HOSSAIN, Monowar

1973. "Population : Background and Prospects". In Lincoln C. Chen (ed.) Disaster in Bangladesh. London : Oxford University Press, pp. 8-17.

MUKHERJEE, B.N. 1975. "Reliability Estimates Estimates of Some Survey Data on Family Planning". Population Studies. Vol. 29, Part I, pp. 127-142.

NASSEF, A. and GAAFAR, M.O.

1971. "Age Patterns of Fertility". Fertility Trends and Differentials in Arab Countries. Cairo : Cairo DemographicCentre.

NATIONAL RESEARCH COUNCIL

NYROP, R.F., BENDERLY, B.L., CONN, C.C., COVER, W.W. and EGLIN, D.R.

n.d. "A Simple Manual on Methods for Estimating Fertility and Mortality Levels and Trends with Incomplete and/or Inadequate Data". Unpublished Draft. Washington : National Research Council, Assembly of Behavioural and Social Sciences, Committee on Population and Demography.

1975. Area Handbook for Bangladesh. Washington D.C. : U.S. Government Printing Office.

0'MUIRCHEARTAIGH, C.A.1980. "An Assessment of the Reliability of WFS Data". and A paper presented at the World Fertility SurveyMARCKWARDT, A.M. Conference, London, July 7-11.

PCFP DIVISION 1978. World Fertility Survey : Bangladesh FertilitySurvey, 1975-76, First Report. Dacca : Ministry of Health and Population Control. Population Control and Family Planning Division. Government of the People's Republic of Bangladesh.

PLANNING COMMISSION 1973. The First Five Year Plan, 1973-78. Dacca :Bangladesh Government Press.

POTTER, Joseph E. 1975. The Validity of Measuring Change in Fertility byAnalysing Birth Histories Obtained in Surveys. Ann Arbor, Michigan : University Microfilms International.

--------------------- 1977a. "Problems in Using Birth History Analysis toEstimate Trends in Fertility". Population Studies.Vol. 31, No. 2, pp. 335-364.

--------------------- 1977b. "Methods of Detecting Errors in World FertilitySurvey Data : An Application to the Fiji Fertility Survey". International Population Conference. Mexico : International Union for the Scientific Study of Population. Vol. I, pp. 101-120.

RABBANI, A.K.M.G., 1979. "1974 Census Estimates of Fertility Levels inD'SOUZA, S., and Bangladesh". In M. Hossain et al. (ed.) Fertility inRAHMAN, S. Bangladesh ; Which Way Is It Going? Dacca : National

Institute of Population Research and Training, pp. 10-47.

REVELLE, Roger et al. 1972. "Population Projections for Bangladesh, 1973-2003"Cambridge, Mass. : Harvard Center for Population Studies. Mimeographed.

ROBINSON, Warren C. 1967. "Recent Mortality Trends in Pakistan". In WarrenC. Robinson (ed.) Studies in the Demography of Pakistan. Karachi : Pakistan Institute of Development Economics pp. 1-49.

1980. "Socio-Economic Determinants of Marital Fertility in Twenty Countries : A Multivariate Analysis". A paper presented at the World Fertility Survey Conference. London. July 7-11.

RODRIGUEZ, German andCLELAND, John

107

ROSENBERG, Irwin H.

RUKANUDDIN, A.R.

RUZICKA, L.T.

RUZICKA, L.T. and CHOWDHURY, A.K.M.A.

SAMAD, A., SHEIKH, K., SARDAR, A.M., BECKER, S.,and CHEN, L.C.

SCHULTZ, T.P.

SCHULTZ, T.P. and DAVANZO, Julie.

SEMBAJWE, Israil S.L.

SHRYOCK, Henry S., SIEGEL, Jacob S. and ASSOCIATES

SIRAGELDIN, Ismail, NORRIS, Douglas, and AHMAD, Mahbubuddin

SIRAGELDIN, I., HOSSAIN, M. and CAIN, M.

1973. "Nutrition : Food Production, Dietary Patterns, and Nutritional Deficiencies". In Lincoln C. Chen (ed.) Disaster in Bangladesh. London : Oxford University Press, pp. 31-51.

1967. "A Study of the Sex Ratio in Pakistan". In Warren C. Robinson (ed.) Studies in the Demography of Pakistan. Karachi : Pakistan Institute of Development Economics, pp. 139-225.

1972. "Infant and Childhood Mortality in the Countries of South Asia : A Review". A paper presented at the Second Asian Population Conference. Tokyo, November 1-13.

1980. Information Obtained Through Personal Communication. Canberra : Department of Demography. Australian National University.

197&a- Demographic Surveillance System - Matlab : Vital Events and Migration - 1975. Vol. 4, Dacca : Cholera Research Laboratory.

1978b- Demographic Surveillance System - Matlab : Vital Events, Migration and Marri.ages — 1976. Vol. 5, Dacca : Cholera Research Laboratory.n.d. "Demographic Surveillance System - Matlab : Vital Events and Migration - 1977". Volume 6, Dacca : International Centre for Diarrhoeal Disease Research.Draft Report.

1972. "Retrospective Evidence of a Decline of Fertility and Child Mortality in Bangladesh". Demography. Vol. 9,No. 3, pp. 415-430.

1970. Analysis of Demographic Change in East Pakistan :A Study of Retrospective Survey Data. Rand, Santa Monica : Agency for International Development. R-564-AID.

1977. "Fertility and Child Mortality, Levels and Differentials Among Yoruba of Western Nigeria". Unpublished Ph.D. Thesis. Canberra : Department of Demography, Australian National University.

1973. The Methods and Materials of Demography, Volume 2, Washington : U.S. Bureau of Census.

1975. "Fertility in Bangladesh - Facts and Fancies" Population Studies. Vol. 29, No. 2, pp. 207-215.

1975. "Family Planning in Bangladesh : An Empirical Investigation". The Bangladesh Development Studies. Vol. 3, No. 1, pp. 1-26.

108

SOMMER, Alfred and 1973. "The Cyclone : Medical Assessment and DeterminationMOSLEY, W. Henry of Relief and Rehabilitation Requirements". In Lincoln

W. Chen (ed.) Disaster in Bangladesh. London : Oxford University Press. pp. 119-132.

STOECKEL, John 1973. "Population : Demographie Trends" In Lincoln C.Chen (ed.) Disaster in Bangladesh. London : Oxford University Press.

STOECKEL, John and 1973. Fertility, Infant Mortality and Family Planning in CHOUDHURY, Moqbul A. Rural Bangladesh. Dacca : Oxford University Press.

SULLIVAN, J.M. 1972. "Models for the Estimation of the Probability ofDying Between Birth and Exact Ages of Early Childhood". Population Studies. Vol. 26, No. 1, pp. 79-97.

UNITED NATIONS 1953. The Determinants and Consequences of PopulationTrends. New York : Population Division, Department of Social Affairs.

---------------------- 1963. Population Bulletin of the United Nations.Number 7. New York : Department of Economic and Social Affairs.

WFS/TECH.

WILLIS, V.G. and WATERLOW, J.C.

YUSUF, F.

1967. Manual 4 : Methods of Estimating Basic Demographic Measures from Incomplete Data. New York : Department of Economic and Social Affairs.

1979. "Bangladesh Standard Recode File". Data File Documentation Series. World Fertility Survey.International Statistical Institute. Unpublished.

1958. "The Death Rate in the Age Group 1-4 as an Index of Malnutrition". Journal of Tropical Pediatrics.Vol. 3, pp. 167-170.

1967. "Abridged Life Tables for Pakistan and Its Provinces, 1962-1964". Contributed Papers, Sydney Conference, Sydney. August 21-25.