- .. v ‘ - .- an r - , , Mg’iz; t I ~ . ' ' 243:2?“ . .aw 32‘. 951;“ ‘ ‘ng-S 1 «i , x ' x , ' - “‘1 ‘1 '5 ‘ ‘ J. _. ~ ‘ M} ". “”- v; {An-“L“? , ~ - , - g . . ‘ .._,., 5: ' maxi? . . 791.1%; » flaxfim - . ., '~~- " ‘ V -- . , n n 4-,. - . - 1 "fit (me , x; ‘q. was?" «a.» A» ‘ ’ ’ ' y' ‘ ' ‘ ‘ - 5 3.; ""33 f"; , H - ”a - ié‘fixz‘xa. W ‘ > .—. 4‘ - ~ ‘3; , .". ' r " ‘ - ram-v .r. a. '. , fag - ‘ “ wpn "9:13:9er : Mr. ' " fiaafii 5:??ka ‘ l '1‘ I ‘ - _ fl . . , . - . , nu , ' g. r- ‘ _~ > 3; . U _“ . ’ ‘ ‘ -,_ . V . C fl: ' . ‘ .fi )‘3 es. . . g: 3533'? ' ' ' 4.; - ,— s- . ‘ .x n. . a t L ' ‘ t 3.7; Yv?‘ .n , A“ > s “u ‘ ‘fi ' '5 i '> ' C " . . n ”a. .h . : ii ' '"‘ ' ‘ = ‘ J. ‘ ‘ C “Lea-w 31,—..uqu . h t‘ ., a n — , ; q "fi&\5{ -p‘q- . - . 4 e u 5 - , ‘ v5 » “ . l. , 3. 3 5, . '35 1533?; “ . were“ ‘91! . ‘ .4 w J ~zv a Ely—*- 3 2 ~* 3‘. ’{fi-igfi'. z .1 Rwai‘Ia. -< . .. a%fi;~ r. h A “I . - 2f“ - - ‘x “E. ‘:2 .eJ “We; - fia‘g 5-“. {h ;, 5m ~ C. m ‘ 7“ an - x," {gum ‘u _ . mart .. - » ~ “aka-n ...--s.g—aa.~~...“i§ ~~ « Hg: ’1} ' ‘u ‘ r w. ‘ - 1" fit ’; v , fi$fluhfil ._ an, A V A _ v w": 7, ' % war. ~" 1,4313 {gwfié ‘* afifé ‘7“: r a}? a. w . ‘ 1‘ m. 1‘. 35kg! m ‘Vflfré‘ 1£§§i33£ 31‘735' a“ ’3’:- ' r. ' W «55: . ? .. v: ~‘k ‘ 't'v‘e!_-q ‘ “J." a i‘iig. .— um» £éuua-I/ . u“ an... .u wt «maker 4 1:1, ‘4 xv. c w . . ‘ wis. . vni‘b-d-h. f H” ‘F J ‘ it 0...,“ ... 1"», “Jana" ‘ . .<-,.( u. ..,..w- ’r ' .. .r v- . U ‘ v. - v w-» v. n .X-.:. t w I‘"» yum: ran-«'54 --., "as: f1 “2" .,,. . - 'N' “M ,,. v » ,‘¢--' '7‘. p ..... .- H u w. ‘. - ’ -,,.,...,-,—1.u—< u w ‘LF - n. :53}. 'v w - a N v ‘w-rtu-l-‘yn-l '- ~~.~-¢--.~« L ...r m. ""343”: “fix-‘2’": ”2"..." m , mm M v . « N .v. - ~-' ' . 1 3.5.1:; '1' -. 4 - (’2'...- A.,.r.. 1.: “w; ~.«"=:::t= - - :r emxguxn. .L ¢‘~:.Lv~ < - u: ‘ .3“ I‘Vo’v <. I" m t ‘I w” w m hwy-~0<« . urn-.- .- v ‘ ' ...,. Amazing 1”7"' ,3. . rm... . 31/3?” 5/312 O . , "M”, f]! GSAN STATE USNIVER ITY LIBRAR IIIIIIIIIIIIIILIIIII III IIIIIIII IIIIIII 300916 6269 This is to certify that the dissertation entitled Rural-Urban Difference in Contraceptive Use in Pakistan: The Effects of Women's Literacy and Desired Fertility presented by Khalida Parveen Zaki has been accepted towards fulfillment of the requirements for Ph.D. Sociology degree in A‘ I ' 7 n 9 Daté’t/QILLL 33/ / /&7 / V or professor MS U it an Affirmative Action/Equal Opportunity Institution 0- 12771 I LIBRARY Michigan State I University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity Institution cmmut IA: ‘~.‘_ h"FL_ THE RURAL-URBAN DIFFERENCE IN CONTRACEPTIVE USE IN PAKISTAN: THE EFFECTS OF WOMEN'S LITERACY AND DESIRED FERTILITY BY Khalida Parveen Zaki A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Sociology 1991 ABSTRACT THE RURAL-URBAN DIFFERENCE IN CONTRACEPTIVE USE IN PAKISTAN: THE EFFECTS OF WOMEN'S LITERACY AND DESIRED FERTILITY BY Khalida Parveen Zaki Literate people are more assertive in seeking personal health and control over their own bodies. Foeromen, this may mean greater autonomy from a social and economic dependence on husbands and. children, and. greater' reproductive freedom. Thus, a rise in literacy should be associated with lower rates of mortality and fertility. Following Easterlin (1975), we reasoned that literate women would view a smaller number of children as appropriate for families like theirs and, due to better health, would need fewer live births to arrive at this number. Hence, we hypothesized that literate women would be more likely to exceed their fertility goals and to contracept. Because rural wives in Pakistan are much less likely than their urban counterparts to be literate, we theorized that this lower literacy would fully interpret the lower'prevalence of contraception in rural than urban Pakistan. The 1984-85 Pakistan Contraceptive Prevalence Survey was used to test this possibility for currently married females aged 15-49 years. Urban women were twice as likely to be literate as were rural women. However, the literate women were not significantly more likely than the illiterate to have reached or exceeded the number of children they deemed to be appropriate. WOmen of all literacy levels whose number of living children had reached or surpassed what they said was appropriate were more apt to contracept currently. However, the likelihood of contraception intensified as the family- building life-cycle advanced, more so for literate than for illiterate women. Although, literacy rates were higher among urbanites, even illiterate urbanites were more likely to contracept than their rural counterparts, literate or not. A likely reason is the much higher psychological and market costs of fertility regulation for rural wives, regardless of their ability to read and write. Rural wives reported greater difficulty in reaching the source of family planning equipment, aids, materials and information and were more likely to choose birth control methods requiring few visits to clinics or hospitals. The findings suggested guidelines for family planning and educational programs in order to boost the prevalence of marital contraception and to reduce the rural-urban difference in contraceptive use in Pakistan. Dedicated to my late father who wanted to see his daughter placed among the highly educated people of the world, and to my mother who always spoke of the sweet fruits of patience. iv ACKNOWLEDGEMENTS I am very grateful to many people in and out of the Department of Sociology whose support made my dream come true. My deepest appreciation goes to my academic advisor, Dr. Nan Edith Johnson, whose persistent encouragement and efficient advice throughout my studies strengthened my efforts which made the completion of this dissertation possible. I learned a great deal from her scholarly, professional and diplomatic expertise. The helpful cooperation of my dissertation committee members provided an extremely pleasant working environment for which I owe them my very special thanks. Dr. Allan Beegle, Dr. George Axinn, Dr. Melissa Barker and Dr. Rita Gallin gave constructive comments which improved my work to a great extent. Dr. Anne Meyering served as the Dean's representative, and her insight was very enlightening. Financial support for my studies, provided by the Department of Sociology, Agricultural Experiment Station, the Office of the women in International Development (Michigan State University), Population Reference Bureau (Washington D.C.) , and Thoman Foundation (Michigan) , is duly acknowledged. The National Institute of Population Studies (Pakistan) very generously provided me the data for this research. Also, I thank Professor Syed Nawab Haider Naqvi, Director, Pakistan Institute of Development Economics (Islamabad) for his sincere understanding and support of my candidacy for this Ph.D. degree. I fully recognize the devotion of my parents, who taught me to kindle my life with the light of knowledge, and.progress by my own hard work. Continued affection of my sisters and brothers was a source of great satisfaction. My husband, M. Javed Akbar Zaki, who is my academic partner also, provided an open-minded, loving and an intellectually stimulating context which empowered me in the pursuit of my academic goals“ Our son, Amir Zaki, deserves my heartiest gratitude for his love, respectful understanding, and smiling face which filled me with joy, pride and determination. I, also, appreciate the caring and cheering support of my friends who shared the joy of my scholarship. vi TABLE OF CONTENTS Page LIST OF TABI‘Eso.0.000000000000000000000000000000COOOOOOOOOix LIST OF FIGURESOOOO0.0.0..0.0.00...0.0000IOOOOOOOOOOOOOOXij-i CHAPTER INTRODUCTIONOOOOOOOOOOOOOOOOOOOOOOOO0.0.0.000000001 Women's Literacy and Desired Fertility......1 Pakistan's Population Problem...............4 Family Planning Program in Pakistan.........6 An Overview of Fertility Levels in Pakistan.................................8 Purpose of this Research...................10 Research Implications......................11 Plan of the Dissertation...................ll RESIDENCE, LITERACY AND CONTRACEPTIVE USE........13 The Meaning of Rurality....................13 The Meaning of Literacy....................23 Easterlin's Theoretical Framework of Fertility and Contraceptive Use.........33 Major Hypotheses...........................42 DATAANDMETHODOmGYeeeooeoeeoeoooeoeoeoo00.00.0044 The Data...................................44 Data Collection and Quality of Data........49 Measurement and Operationalization of Variables............................75 Place of Residence................76 Literacy Status...................76 Fertility Status..................76 Family-Building Life-Cycle Status.77 Current Use of Contraception......78 Statistical Techniques.....................78 Pearson's Chi-Square Test...............78 Logit Regression Model..................8O ANALYSIS OF DATA AND FINDINGS....................82 Some Selected Characteristics of Pakistani Women........................82 Literacy by Place of Residence: Hypothesis 1.........................84 Fertility by Literacy: Hypothesis 2.........................84 vii Contraceptive Use by Fertility Status: Hypothesis 3.................... ..... 86 Contraceptive Use by Residence by Family-Building Life-Cycle: Hypothesis 4.........................89 Contraceptive Use by Literacy by Fertility: Hypothesis 5.........................90 Selection of the Best-Fitted Logit- Regression Model................ ....... .90 Logit Regression Analysis of Hypothesis 5........................103 5. COST OF FERTILITY REGULATION: A RURAL-URBAN COMPARISON. O O O O O O O O O O O O O O O O O O O O O O O O O O O O 110 Current Contraceptors: A Prefile. O O O O O O O O O O O O O O O O O O O 0 O O O 112 Current Non-Contraceptors: A PrOfileO O O O O O I O O O O O O O 0 00000000 O 127 6. SUWY AND CONCLUSION. O O O O O O O O O O O O O O O O O O O O O O O O O 134 LIST OF REFERENCESOOOOOOOOOOOOOOOOOOOOOOO...O...00.00.00.153 APPENDIX O...0.0000000000000000000000000000000......0.0.0162 viii LIST OF TABLES Page The Percent Distribution of Women by Marital Status in Rural and Urban Pakistan, PCPS 1984-8500000...0.00.0.0...-OOOOOOOOOOOOOOO....48 Percentage Distribution of Sample Women in PCPS 1984-85 and 1981 Population Census in Pakistan by Place of Residence................53 Percent of Women Wanting More/No More Children By How the Current Number of Living Children Compares to the Number deemed Appropriate: Rural Pakistani Women, PCPS 1984-85..................................66 Percent of Women Wanting More/No More Children By How the Current Number of Living Children Compares to the Number deemed Appropriate: Urban Pakistani Women, PCPS 1984-85..................................68 Percent of Women Who Wanted/Did Not Want The Last Child By How the Current Number of Living Children Compares With the Number Deemed Appropriate: Rural Pakistani Women, PCPS 1984-85...........................70 Percent of Women Who Wanted/Did Not Want The Last Child By How the Current Number of Living Children Compares With the Number Deemed Appropriate: Urban Pakistani Women, PCPS 1984-85...........................7l Percent of All Couples by Future Desire of Additional Children for Rural Residents, PCPS1984-85000000000OOOOOOOOOOOOOOOOOOOOO0.0073 Percent of All Couples by Future Desire of Additional Children for Urban Residents, PCPS1984-85.0...0.00000000000000000000000000.74 Some Selected Characteristics of Pakistani Women Based on The Pakistan Contraceptive Prevalence Survey 1984-85.....................83 ix Literacy Status of Currently Non-Pregnant Wives by Urban-Rural Place of Residence in Pakistan, 1984-85.0000000000000000000000000.0085 Deficit Fertility Among Currently Non- Pregnant Wives by Literacy Status in Pakistan, 1984-8500000...0.0.0.0000...0....0.087 Current Use of Contraceptives Among Currently Non-Pregnant Wives by Deficit Fertility in Pakistan, PCPS 1984-85.......................................88 Percent Contraceptive Use Among Currently Non-Pregnant Wives by Family-Building Stages and by Place of Residence in Pakistan, PCPS 1984-85........................9l Goodness-of Fit Statistics (G2) for Various Log-Linear Models Applied to Test HypothBSis.5...C......................... ..... 96 Comparison of Difference in Goodness-of-Fit Statistics (62) for Various Reduced Log- Linear Models with the Large Model, Applied to Test Hypothesis 5..........................99 Logit Regression of Current Contraceptive Use (1=Yes: 2=No) upon Wife's Residence, Literacy and Control Variables...............104 Percent Contraceptive Use Among Wives by Deficit Fertility Status and by Place of Residence in Pakistan, PCPS 1984.............106 Contraceptive use Among Currently Non- Pregnant Women by Family-Building Stage and Literacy Status in Pakistan, l984-85......................................108 Percent of Wives Currently Using Specific Methods of Family Planning by Residence, PCPS1984-85.00.00000000......0.00.00.00.0000113 Source of Contraception by Residence of Currently Contracepting, Non-Pregnant Wives: PCPS1984-85.00.00.00.0000000000000000114 Percent of Wives Currently Using, Second Preferred Methods by Residence, PCPS1984-850000000000000OOOOOOOOOOOOOOOOO0.0116 Percent of Non-Pregnant Wives Currently Using, Have Stopped Family Planning for More Than a Month by Residence, PCPS 1984-85......................117 Percent of Non-Pregnant Wives Currently Using, Reasons for Stopping Family Planning for More Than a Month by Residence, PCPS 1984-85......................118 Percent of Non-Pregnant Wives Currently Using Family Planning, Had Stopped Use for More Than a Month by Out-of-Supply-Method by Residence, PCPS 1984-85.................................120 Percent of Non-Pregnant Wives, Currently Using a Method by Who Gets the Method and Residence, PCPS 1984-85.................................121 Percent of Non-Pregnant Current Users by Difficulty in Reaching the Source of Family Planning by Residence, in PCPS l984-85..............................123 Percent of Non-Pregnant Current Users by Mode of Transport for Reaching the Source Family Planning by Residence, in PCPS 1984-85..............................125 Percent of Non-Pregnant Current Users by Reasons for Difficulty in Reaching the Source of Family Planning by Residence, in PCPS 1984-85..............................126 xi 5.11. 5.12. 5.13. Percent of Women Currently Non-Contracepting Intending to Use in Future by Preferred Method by Residence, lJIPCPS 1984-85..............................129 Percent of Women Currently Non-Contracepting Indicating to Use in Future by Preferred Source by Residence, PCPS l984-85............130 Percent of Non-Pregnant Wives Currently Not Using by Reasons for Non-Use of Family Planning by Residence, PCPS 1984-85..........132 APPENDIX Husband's Occupational Status by Place of Residence in Pakistan, PCPS 1984- 1985.00.00.000000000000000000000000000000000.162 Percent of Wives Literate by Husband's Occupational Status in Pakistan, PCPS 1984-85..0.0.00.00000000000000000COOOOO0.0.0.163 Mean Number of Children Deemed Appropriate, Ever Born and Surviving to Pakistani Women by Literacy Status, PCPS 1984-85.............165 Percent of Wives Literate by Age at Marriage in Rural and Urban Pakistan, PCPS1984-85.00.00...OOOOOOOOOOOOO0.0.0.0....166 LIST OF FIGURES Figure Page 2.1. A Hypothetical Model of Education- Modernization, Desire for Children and contraceptive UBeOOOOOOOOOOOOOOOOOOOOOOOOOO0.0.39 3.1. An Approximation of Underreporting- of Contraceptive Use in Rural Areas, PCPS1984-85.00.00...OOOOOOOOOOOOOOOO000.......58 3.2. An Approximation of Underreporting- of Contraceptive Use in Urban Areas, PCPS1984-85000000000000OOOOOOOOOOOOO0.0.0.000059 xiii CHAPTERI INTRODUCTION Women's Literacy and Desired Fertility Social modernization is a process of the adoption or creation of new skills that enable one to manipulate the social or physical environment to one's advantage. It induces innovative behavior, changes the norms of a society, and affects the demographic processes. In the context of the contemporary developing world, social modernization implies any or all of the following: the socio-economic changes ensuing from technological and industrial developments, urbanization and urban influences, improvements in literacy levels and Westernization. Westernization includes an acceptance of reason as a tool that should be used to achieve personal autonomy over the conditions of one's existence, including the number of children begotten. Westernization occurs because the development of high technology enables communication on a global level. However, literacy is necessary for one to interact, send and receive messages through. global channels: such. as, books, magazines, and letters. This dissertation uses Easterlin's economic framework of fertility analysis to link social modernization with contraception by Pakistani wives. It is hypothesized that the medium is change in wives' perceptions about the number of 1 2 children appropriate for them. Among the changes implied by modernization, literacy, particularly among women, is deemed here to represent the most consistent determinant of fertility and contraceptive use (Caldwell, 1980; 1981: Cochrane, 1979). The studies for developed countries have shown quite a uniform relationship between literacy and fertility but the picture has remained ambiguous for several developing countries (Cochrane, 1979: Mason, 1985). In the latter, it is important to know the levels of literacy among women and the conditions that shape their perceptions of the number of children appropriate for them. Among these conditions, rural or urban residence may have a differential impact. Rural residence is assumed to make women desire a higher number of children and thus to keep contraceptive prevalence relatively low. Literacy is assumed to affect the fertility goals of women which in turn affect their demand for contraception. In this regard, the case of Pakistan is well placed for comparative analyses of rural-urban differences. Seventy percent of Pakistan's population reside in the countryside. With the improved communication and transportation network urban influences have crept into rural society over time. Still it may'be the low literacy levels among rural women that explains a more traditional behavior than what is feund among women living in more urbanized areas. 3 It is generally believed that national fertility rates will decline if and only if the couples are motivated enough to use fertility control methods and the cost of fertility regulation is relatively low (Easterlin, 1975). So long as couples have fewer surviving children than they want, they have "deficit fertility" and no motivation to contracept. The motivation to use fertility control comes when individual couples decide that their number of surviving children exceeds the number they want (I call this excess "undesired fertility") . However, motivation is a necessary but not a sufficient condition for the adoption of birth control. The sufficient condition, according to Easterlin (1975), is that the cost of exceeding one's fertility target must weigh more heavily than the cost of fertility regulation. I believe that literacy prepares Pakistani women to manage the extrafamilial environment more profitably and that excessive fertility taxes this gain, In other'words, female literacy in Pakistan should create the necessary condition for contraception by reducing the desired number of children below the number biologically possible and the sufficient condition by creating penalties for exceeding the desired number of children. Also, literacy ought to reduce the search costs of acquiring information about how to obtain and use contraception and the psychological costs of adopting it. Social modernization provides the macro-level context in which women attain literacy and are transformed by it. Pakistan's Population Problem Pakistan has experienced a rapid rate of population growth since independence in 1947. The growth rate has remained almost stable at 3 percent annually for the past few decades, and this implies a doubling of the population size every 23 years. By 1990, the size of the Pakistani population had grown to 114.6 million people, a fact which made Pakistan the tenth most populous nation in the world (Population Reference Bureau, 1990). Such a high population growth rate is detrimental to Pakistan's socio-economiC~ development since it exerts too much pressure on the country's narrow resource base. The rapidly growing population has also alerted the economic development planners to the serious implications it has for the country's cultivable land area. One estimate of the cultivable land area indicated the person-land ratio to be 0.3 hectare per capita (Pakistan Planning Commission, 1986). Pakistan has an agriculturally based rural economy. About 70 percent of the total population lives in rural areas. The large percentage of youth under age 15 (43%: Population Reference Bureau, 1990) bears a great potential for further growth. The growth is mainly due to an excess of births over deaths, for net immigration to Pakistan is very small. Thus, this dissertation will address ways to lower the birth rate. 5 Literacy levels are low in Pakistan for men and women. According to the 1981 census, 16 percent of females were literate as compared to 35 percent of males. Similarly, the primary school enrollment of females is 13 percent compared.to 81 percent for males. The socio-cultural milieu prevalent in the society values the education of boys over girls. Education is considered less relevant for females because they are not expected to take paid employment outside the home. All children are regarded as very special but sons in particular are considered as potential producers of income and as sources of support in old age and in difficult social and political times. Usually a large number of children is desired, and sons are preferred to daughters. Society and religion favor high fertility values, and marriage is universal. The World Fertility Survey in Pakistan showed that half‘of all women aged.25—29 had been married by age 16.5: and half of all women aged 40-49 had been married by age 15 (MacDonald, 1985: 98). This increase over a 15-year’period.of 1.5 years in the median age at female marriage might foretell a slight future fertility decline by shortening the lifetime exposure to marriage and by increasing the proportion single at the younger ages. Nevertheless, the extremely youthful median age at female marriage probably means a continuation of very high fertility for Pakistan. 6 Like most other developing and Islamic countries, the status of women is determined by a gender stratification system which places males in privileged positions. The patriarchal social organization of the society emphasizes the family and defines the domestic tasks of child bearing and rearing as the responsibilities of females. This division of domestic labor reduces women's chances of involvement in extrafamilial activities and keeps them in a low profile. In spite of this, a small fraction of women works in factories, universities, schools and colleges (most of them exclusively for females), government offices, scientific institutions, airlines, and hospitals. Thus, the social barriers are not insurmountable if there are attitudinal and structural changes in the country. Family Planning Program in Pakistan In the 19603, the. government. maintained. that rapid population growth hampered economic growth and adopted a vigorous family planning policy through the ‘Economic Development Plans. Over the past twenty years, this policy has remained in effect: but no remarkable success has been achieved. Among the currently married women, the percent reporting the current use of contraception had been fluctuating between 3-6 percent until 1979. However, higher percentages of contraceptive use of any method (9.1%) and of efficient methods (7.6%) were reported in the 1984-85 Pakistan 7 Contraceptive Prevalence Survey (Population Welfare Division, 1986: Table VIII.7, p.84). These rates are still far below that reported.in other South Asian countries“ For example, in Indonesia according to the 1987 Demographic and Health Survey, the contraceptive use of any methods was 48% and of efficient methods was 44% (Institute for Resource Development, 1990). Contraceptive use of any method included methods, such as, oral pill, condom, vaginal methods, injection, IUD, male and female sterilization, rhythm, withdrawal and all other methods. Contraceptive use of efficient methods excluded methods, such as, rhythm, withdrawal and all other methods. The success of the family planning program of Pakistan was expected to depend on dissemination of family planning information. Massive propaganda campaigns, along with an inundation of contraceptives, were carried out. Social resistance to the acceptance and practice of birth control methods was anticipated. To create acceptance and demand among the eligible couples (20 million during 1965-70), an administrative and supply oriented strategies were adopted (Robinson, 1978). These strategies emphasized the inundation of contraceptive methods, mass media campaigns and a large- scale, highly' decentralized administrative organizational structure. These strategies failed to bring the birth rate down even after a number of reshufflings of the program. By the end of the 19703, not much demand could be created. It is 8 difficult to induce such behavioral changes in Pakistan. The factors related to demand for contraception in Pakistan were completely ignored. Generally it is believed that the success of any birth control program depends upon the level of demand for, as well as supply of, contraception among the eligible couples of that society. The revised population policy of the early 1980s set the "new beginning". The program was dedicated to manipulating factors related to the demand for children and contraception. It adopted a general population welfare approach and pointed out the need for’ raising literacy levels and. providing employment opportunities for women, as well as for maternal and child health care (Robinson, 1987). The success of the "new beginning" has yet to be seen. Nevertheless, while discussing options for development particularly in rural areas, Lieberman (1982) was quite pessimistic and called the birth control policy "timid and defensive" and "conservative”. An Overview of Fertility Levels in Pakistan The total fertility rate (TFR), the number of births a woman will bear between menarche and menopause if she goes through the reproductive span bearing children at the age-specific fertility rates recorded for a given year, has remained nearly unchanged over time in Pakistan (Alam.gt_al., 1984: Alam and Dinesen, 1984: Shah et al., 1986). The 9 evidence relates back to the findings from the surveys done in the 12605 ( (Population Growth Estimation (PGE) project of 1962-65, and National Impact Survey (NIS) of 1968)) and.12193 ((Pakistan Fertility Survey (PFS) 1975, Population Growth Surveys (PGS) of 1976 through 1978, and Population, Labor Force and.Migration (PLM) survey of 1979)). The value of the TFR obtained through these sources has always been constant around 7.0 children per woman, except for a TFR of 6.3 during 1970-1975. However, the analysis of Age-Specific-Fertility- Rates (ASFRs) indicated no change except for a decline in the age group 15-19 which was attributed to the increasing age at marriage in the country for this period. The estimates of Total Marital Fertility Rates (TMFRs) have remained quite stable around 8.0 over this period. The most recent figures of TMFRS, available from the 1986 and 1987 Pakistan Demographic Surveys (PDS) and Pakistan Contraceptive Prevalence Survey (PCPS) of 1984-85 are 8.0 and 8.09, respectively. These rates mean that Pakistan has the second highest level of human reproduction in South Asia (after Afghanistan). A common finding from the above studies has been that educational attainment for women beyond the primary level of schooling was negatively associated with fertility (Casterline, 1984: Sathar, 1984) and positively associated with intentions for future contraceptive use (Shah and Shah, 10 1984). Moreover, cumulative fertility at early durations of marriage differed from that at later durations (Casterline, 1984). Also, in Pakistan, marital fertility is higher in urban areas than in rural areas (Alam et_al., 1984, and Shah g;_g1., 1986), for the possible reason that when women become modernized, they abandon breastfeeding. The quicker resumption of ovulation which prepares a woman for another conception results in more frequent births among urban than rural women. Rural women breastfeed their children longer than urban women. Purpose of This Research The purpose of this research was to investigate the difference in current contraceptive usage by married couples in rural and urban Pakistan. To evaluate the family planning program performance, it is essential to know the interlinkages of current nonusage of contraception with other social, economic and demographic factors that differ in rural and urban Pakistan. For example, is literacy related to one's fertility goals? Does falling short of that goal mean an avoidance of contraception? Does attaining that goal lead to contraception use? .And if so, how does this relationship vary by rural-urban place of residence? Since public education is a service which policy makers can affect, my study will offer guidelines to policy makers which can boost the demand for services provided by the national family planning program. 11 The socio-economic and demographic background variables determining the motivation to use and the actual use of fertility control methods are the core of this research. Research Implications This research will analyze the rural-urban difference in current use of contraception by couples in Pakistan. The major thesis of this study is that the rural-urban difference in women's literacy is the key factor to explain the higher fertility and the lower use of contraceptives in rural areas of Pakistan. The study is important for both practical and theoretical reasons. On the practical side, rural people make up 70% of the population of Pakistan. Social factors that inhibit contraceptive use by rural couples perforce doom the country to a continued high rate of population growth. By examining one such social factor---wife's 1iteracy--- I can suggest whether literacy campaigns for girls in the rural areas could alter the motivations of rural couples to contracept. On the theoretical side, my research investigates whether wife's literacy is closely related to the couple's deficit fertility (as reported by the wife). Plan of the Dissertation Chapter 2 reviews the relevant literature and then uses Easterlin's (1975) economic framework to show how much higher illiteracy rates might explain the much lower contraceptive 12 prevalence in rural Pakistan vis-a-vis urban Pakistan. The methodology, the data and the statistical procedures for testing our research hypotheses are described in Chapter 3. The findings are presented in Chapter 4. Chapter 5 explores the cost of fertility regulation among users and nonusers of family planning methods by place of residence. The results of research and suggestions for further research and policy are given in Chapter 6. The ways to improve future research as well as government policies to enhance female literacy and family planning program services are also discussed in this final chapter. CHAPTER 2 RESIDENCE, LITERACY, AND CONTRACEPTIVE USE The Meaning of Rurality It is important to distinguish between rurality and urbanity in a geographical and a social sense. Traditionally, Pakistani population surveys have followed the census definitions of a rural and an urban area. These definitions have changed over time (1951-1981) in the census-and survey- conducting history of the country. In the censuses prior to 1981, a settlement ‘with a size Ibelow 5,000 people ‘was classified as "rural": the residual category was "urban”. The 1981 census definition classified an area as rural or urban not by number of inhabitants but by the presence of some form of government administrative organization: such as, a municipal committee, town committee or a cantonment. The regional census directors had the discretion of classifying an area as rural or ‘urban based on their observations of commercial activities, agricultural activities, public utility systems, road systems, water supply systems and street lights in an area (Abbasi, 1987). Accordingly, the 1984-85 Pakistan Contraceptive Prevalence Survey adopted the 1981 census definition. Hence, whether the size of the population was below or above 5,000 persons did not matter. Although the numerical size is 13 14 important for administrative purposes, it is notable that a greater diversity of demographic and socio-economic characteristics exists among people in urban areas than that in rural areas. However, the social-structural, political and economic differences of the places where people reside are more important in an analysis of the behavioral aspects of a human population than merely their size. Traditional behaviors are more ubiquitous among rural people than urban people in Pakistan. In rural areas, the majority of people's livelihood depends upon agriculture or other primary occupations, (e.g., mining and fishing): but in urban areas it depends upon nonagricultural (secondary or tertiary) occupations. In order to justify such a claim, in Pakistan, I presented some empirical evidence from 1984-85 Pakistan Contraceptive Prevalence Survey in Appendix Table 1.1. The farming system of Pakistan is labor-intensive rather than capital-intensive, and is not dependent on large amounts of formal training. It is unsurprising that there are very few agricultural training institutions. The knowledge of farming is usually passed on informally by word and deed from parent to child. The skills for making tools are the crafts of a few. The previous political administration developed agricultural extension programs targeted to men in the 15 community. Male extension workers contacted male farmers for informal courses , discussions , and demonstrations on occasional visits. To the extent that any nonformal educational program was launched, it addressed itself to men and ignored women. Despite the shortcomings of some agricultural programs, female labor force participation in the agriculturally based rural economy is of immense significance. To prove this empirically, however, remains a difficult task. The inconsistency of definitions for measuring female labor force participation in agriculture by censuses and surveys has underestimated women's overall work participation through underreports of their productive endeavors (Shah, 1986) . Women husk, thresh, and parboil rice: but if it is only for family consumption and.not for trade, it usually does not.meet the census or survey definition of "work." Women work for cash, share, or kind in small-scale industries at home, such as, in sewing, tailoring, embroidering and carpet weaving, which require special skills usually imparted by the elderly women on an informal basis. According to the 1975 Pakistan Fertility Survey (Shah, 1986) , most women who worked outside the home were employed as housekeepers or domestic servants (94% urban and 84% rural). Women who worked inside their own homes considered themselves 16 to be self-employed as spinners, weavers, tailors, and other craftspeople. The percentages of women in such activities (except tailoring) were generally higher in rural areas than in urban areas. However, in rural areas among women who were involved in farm-related jobs outside the home, 75% were employees as compared to 25% who were self-employed. It may be surprising to note that of urban women who were involved in farm related jobs, 83% worked outside the home as employees (Shah and Shah, 1980). However, among those who were classified as unskilled workers and laborers and worked outside the home, 48% (rural) were self employed as compared to 14% (urban): and 36% of rural and 43% of urban women reported to be working for employers. It is possible that most of the rural women who worked outside the home, other than as teachers, nurses, housekeepers, and servants, were involved in their family-farm activities as unpaid family helpers and reported themselves as "self-employed". These categories of activity emphasize the segregation of men's "work! fromnwomen's "work" and the unimportance of literacy to women's labor in such a gender-segregated economic environment. Furthermore, it highlights the greater independence of women's work from expectations of literacy in rural than in urban locations for Pakistan. 17 The segregation of women's and men's work is consistent with the religiously mandated purdah, the protection of women from contact with unrelated men. Purdah also extends to women's clothing. The usual custom among Islamic women is to wear a costume or a covering on their bodies that completely veils. Among rural women a modest dress of shalwar (pants) and kameez (top) and a chadder (sheet of cloth to cover head and shoulders and spread over the breast) or dopatta (fancier than chadder) is a common view of purdah as an appropriate dress, in Pakistan. Additionally, the roots of such a practice lie in the patriarchal social structure of many Asian societies and not only in religion, since purdah is not the exclusive practice of Islamic nations although it is most strictly observed there (Caplan and Bujra, 1979). Unsurprisingly, the directors of agricultural activities are men: and.women's role is considered subsidiary. If a male member of the family is absent and there has to be a contact or communication with unrelated men, it is preferable for an elderly woman, in a village (where a communal relationship exists in its highest form) to address and treat younger men as sons and others as brothers. Such a practice pervades urban areas but to a lesser extent. Whether the women workers in the informal sector of the rural economy are paid in cash and/or kind, is not known, How 18 much freedom they have to disburse their own earned income is not known, either. How much of the work participation by women empowers them to make decisions with regard to marriage, fertility and towards contraceptive adoption, is not known. I think that the majority of women depend upon their husbands and elders for these decisions. This is because there exists a gender-stratification and age-stratification system in the society which envisions husbands and elders as authority figures in the family. Husband's role is not only to be economically responsible for the family but also to be the protector and director of the behavior of wife. This is apparently because of the prevalent norm of male-dominance in the society. Lee and Bulatao (1983), in "A Framework for the Study of Fertility Determinants," emphasized the social and cultural normative factors (macro-level variables) in a society along with couple's characteristics (micro-level variables) that affect fertility decisions. Women and men may view children to be more- or less- valuable depending upon their particular social and cultural environment. In their demand-for-children framework, Lee and Bulatao (1983), maintain that the value of children, interpreted as the perceived costs and benefits of having children (of a particular sex) determine parent's desire (demand) for a certain number of children (pp.l-13). Moreover, the authors argued that parent's desired number of 19 children is affected by their nuptiality history and childbearing experience. The time spent during a marital union, temporary separation between spouses, marriage dissolution or divorce, remarriage, and reproductive experience can alter parent's desire for a specified number of children over time and their motivation to control their fertility. Individual couples may want to have a specified number of children at one point of time which is subject to change depending upon that individual couple's childbearing experience: thus, that individual couple may desire a different specified number of children at a later point of time, depending upon their experience in life under their circumstances. Nevertheless, Arnold g; 31. (1975), in their cross-national investigation of value-of—children in selected developing countries found that even if men and women differed in their values for having children, their ideals--eventual fertility goals--were similar. Mason and Taj (1987) suggested four major conditions in cultural perspective which appeared to be associated with differing fertility goals of women and.men. These conditions are: (1) strong patriarchal system: (2) premodern economic conditions: (3) lineage-oriented kinship system: and (4) certain demographic conditions, for example, age differences between spouses, and a low remarriage rate, etc. These conditions produce inegalitarian gender relationships which 20 are mostly male-dominated. However, their study found no major differences in fertility goals of husbands and wives. Thus, it is possible that women in a particular social setting may realize the higher costs of producing children and may desire to have fewer of them, but this desire may not necessarily result in contraceptive adoption. I would assume some or all of the above conditions would apply to Pakistani society, however, further research is needed to understand the extent and degree of prevalence of such conditions. However, as stated.earliery I assumed that the fertility goals of*wives reflected husbands fertility goals in Pakistan. Because women's work is less valued than men's, sons are desired more than daughters in many developing countries (Mason, 1985). Attempts to have at least two sons contribute to larger family size. A large number of children is valued for parents' old age security, for securing women's status in the family, for insurance against divorce, and for economic support (Mason, 1985: Cain, 1980). Increasingly, poor families in Pakistan are finding the rising cost.of'daughters' marriages unaffordable. Dowries make girls "expensive", and the families lose their labor after their marriages. But boys contribute to the family's income and status all their lives (Miller, 1983: and Shah, 1986). For these reasons, sons have a higher net economic value to parents than daughters do: and, unsurprisingly, parents want more sons than daughters. 21 Ahmad (1981) analyzed the relationship of family size and sex preference by controlling socio-economic background variables of females and found a preference for large families and many sons all age groups of women in rural Bangladesh. The greater the son preference, the higher was the observed fertility. Since few Bangladeshi women preferred a small or moderate family size, there was a low demand for contraception. However, Khan and Sirageldin (1977: 1979) concluded on the basis of their analysis of 1968 National Impact Survey that couples with fewer living sons desired additional children. Miller (1983), while studying socio- economic patterns in son preference on the Indo-Pakistani subcontinent, found that juvenile sex ratios in Pakistan were similar to those prevailing in North India (the part adjacent to Pakistan's.boundary). Son preference in India is also very strong: Das (1984) reported that Indian couples wanted at least one son to support the family and continue its lineage. Farooqui (1984) discussed similar factors affecting fertility behavior in rural Pakistan and emphasized the importance of son preference in rural demographic behavior. While Sathar and Akhtar (1988) found a positive relationship between family size and contraceptive use in Karachi, they observed a higher percentage of contraceptive use after women have had two living sons or two living daughters (almost 50%) than when they had one son or one daughter (almost 40%). 22 The above studies suggest that an unfulfilled preference for giving birth to a certain number of sons can create indifference to marital contraception. However, I can not analyze how son preference affects current contraception because my data set (described in the next chapter) just asked the number of children the respondent considered appropriate for a family like hers, not the number of sons and the daughters. Retherford's "Theory of Marital Fertility Transition" (1985) , within the Easterlin's (1975) economic framework (which I discuss in a following section) of demand for- and supply- of children, incorporated the effects of child mortality, socio-economic development and contraceptive diffusion processes, in a society. The author presents a model, integrating the perspective of costs and benefits (the economic notion of cost-utility function) of having a certain number of children to the perspective of costs and benefits of contraception (cost-utility function of contraceptive innovation). The underlying assumption here is that some of those costs must be common to both perspectives. Retherford's (1985) theory assumes that speed of contraceptive diffusion determines the threshold level of marital fertility transition (from high to low after the desired number of children have been achieved by a couple) within a given level of social and economic development of a society. In turn, the level and 23 speed of contraceptive diffusion is viewed to emerge from endogenous factors, depending upon the society's developmental level. In the context of pretransitional societal conditions Retherford cited that in the 19708 in Pakistan the family planning program offered its services at very low costs. This did not encourage couples to limit family size. In spite of the indications that desired family size was four children, the fertility remained around 4-6 children. Thus, the marital fertility transition was not precipitated by the enormous family planning diffusion (p.265). Therefore, the author noted that the measurement of the variables related to couple's utility-function are indirect and difficult. The Meaning of Literacy Researchers often surmise that in patriarchal settings with rural-agrarian economies, women desire more children (more sons) than men do, to attain social power and status in the family and the society. However, Mason and Taj (1987) found similarity in reproductive goals of husbands and wives, although reasons for that consensus may have varied between them. Hence, wives' fertility desires ought to represent couples' desires. Thus, the use of wives' literacy as a key variable interpreting the prevalence of contraception is considered appropriate for this research. 24 An individual's literacy is generally related to the level of formal schooling, knowledge and information.about.the environment, variety of skills attained through nonformal educational institutions, and changes in perceptions of what is true or desirable. It is often assumed that literacy depresses fertility by altering norms (expectations) of appropriate feminine behavior and/or by enabling women to engage in. new* ways of controlling’ their environment. to advantage themselves. However, these effects are quite varied across societies (Caldwell, 1980: Cochrane, 1979: and Benavot, 1989). Therefore, it is important to make inferences about the relationship of literacy to fertility in the context of a particular social structure and to be cautious in generalizing to other social structures. Cochrane's (1979) survey of the literature suggested that the attainment of literacy by women may increase the biological capacity to bear children and thus the rates at which children are borne in countries (at early stages of development) where female literacy is the lowest, because literacy ameliorates fetal and infant mortality and maternal health. Thus, female literacy will probably have a greater impact on fecundity than male literacy does. Nevertheless, improvement in literacy may depress actual fertility among urban women more than among rural women, because urban women have more chances to gain awareness, knowledge, availability, 25 and accessibility to family planning counselling centers. Moreover, rural-urban fertility differentials were observed to be narrowing due to the spread of modernization effects of educationflandmurbanization.(Smith, 1983). .Additionally, urban women are more exposed to the media messages about family planning and thus less inhibited than rural women about discussing such matters. However, literate urban women also tend to wean their children sooner (Sathar, 1979) . Early weaning hastens the return of postpartum ovulation and the chance of a new pregnancy. Hence, literacy can have negative and positive relationships with actual fertility, and can speed or slow its decline. Graff (1979) presented an account of the historical demographic transition (Bogue, 1969), modernization (Karsada, 1973: Easterlin, 1974) and economic perspectives (Schultz, 1973: Easterlin, 1974) on fertility decline in the context of the role of education and literacy. In the perspectives reviewed by Graff, literacy was one of the central variables, which according to him affected many social processes, such as, marriage patterns, preference for children, labor force participation, increases in income, and many more phenomena. According to Graff, few studies paid attention to the conceptualization of literacy: he presented a broader generalized conceptualization of literacy and its contribution to fertility decline, irrespective of the levels of formal 26 schooling obtained. However, he argued that among other modern institutions, school is only one of many through which social modernization can occur. Diverging from Graff's (1979) viewpoint of literacy, Caldwell (1980) emphasized the importance of attending primary school. He argued that the ideas and value systems in Third- World primers are those of the Western nuclear family. The central idea in Caldwell's approach is that the model of the Western nuclear family is one in which intra-family flows of wealth are from parents to children. Families organized along this model find children increasingly costly as modernization advances such that parents desire a few of them of "better quality" (Caldwell, 1982) . Educated parents like to send their children to school which entails some costs, such as, cost of books and school uniforms, tuition fees and transportation costs. The costs may also include the replacement cost of child's forgone labor around the house or on the farm. Because of the high costs involved in children's schooling, parents desire to have fewer of them. Literate parents may be somewhat less dependent economically on their offspring, especially in old age. Therefore, educated parents in nuclear households find themselves in a position of making decisions in favor of their own children and protecting their children's interests and by 27 spending on their education and health. Hence, parents' tastes for better quality children are reflected in their children's education and its related costs. Caldwell's observation on the rising levels of education and its relation to changing marriage patterns, fertility desires and contraceptive use in Sri Lanka exemplify his perspective on fertility decline in a Third World setting (1988). However, the practical significance of Caldwell's theory is debatable in the context of many other developing countries. For example in urban Pakistan, the World Fertility Survey showed that the woman had to attain at least seven years of schooling before fertility began to fall (Singh and Casterline, 1985: 205). Since, the seven-year threshold is beyond the primary school level, we challenge Caldwell's thesis that the content of primary school education is enough to change ideas about desired family size in Pakistan. In any case, Caldwell's emphasis on the directionality of the intergenerational wealth flow attempts to link the macro-level changes to the individual couple's desired number of children. A major assumption of this dissertation is that the key mechanism through which schooling leads to fertility decline is the attainment of functional literacy, defined as the ability to read a newspaper and to write a simple letter. Notice that this is a strict definition of literacy because it requires not only the consumption but also the production of 28 written language. Writing is more useful in the everyday lives of Pakistani people than reading stories about Western- style nuclear families as portrayed in primers. Mastery of the written word requires more than a few years of primary schooling, especially given the underendowment of rural primary schools, and may explain why a secondary level of schooling must be reached before a drop in fertility is noticed. The low levels of female literacy in Pakistan can be understood in the context of its socio-cultural conditions, which empower rural women less than urban women (Shah, 1986: Khan and Bilquees, 1976: Farooqui, 1984). Formal schools are the common places where reading and writing (our definition of literacy) are learned. The quality of existing formal schools for boys is not great, but girls are even more restricted from attending any schools, since in some rural areas with limited resources only one school is available at the primary level. Additionally, the education of girls is not deemed as important by parents as is that of boys: and.this prejudice is stronger in rural areas than in urban areas. The investment in daughters' education is not expected to benefit parents in the long run, because daughters will be married and.be gone to their husbands' homes. For all these reasons, female enrollment rates are low'and.their'dropout rates are high: and very few adequately learn to read and write and then have 29 chances to practice these skills. Moreover, the material infrastructural difference between rural and.urban areas has kept the conditions of rural schools persistently poor. The rural development and education plans have not been very successfully implemented in the past. Whatever improvements have taken place have been geared more to fulfill the needs of boys than girls, and women have been left behind men. The education at the religious institutions, such as, mosques and 'madrassas' is quite common but insufficient to produce functional literacy. The language of the religious books and the Holy Quraan is Arabic, which most of the pupils would memorize how to read and recite in their prayers: but most would not understand it without any translation into the local language. At a few formal religious institutions some (men) may learn how to write Arabic: but this practice is not of much practical use in Pakistan. Reading and writing in everyday life occurs in one of the four regional (Sindhi, Punjabi, Baluchi and Pushto) and/or national (Urdu and English) languages. The failure of the madrassas to teach literacy in the vernacular language keeps its male students illiterate in a practical sense. However, even if the boys learning to read the Holy Quraan at the madrassas were taught a locally used language the result would probably not alter 3O gender roles. This is my assumption that not much new knowledge (contextually different than the traditional value system) can possibly be imparted through such a way to bring Westernization, modernization and attitudinal changes. Moreover, the Quraan quite clearly defines the gender roles. A woman is supposed to work as homemaker and a caretaker of children and husband. A man is supposed to be the bread winner and bearer of the economic responsibility of the family. Nevertheless, the Quraan quite explicitly addresses both.men and.women for acquiring education, dealing in business, earning and spending of income by their choice. According to Quraan, a woman is allowed to acquire education, engage in.business, earn and spend her own income in a way she desires to do so. However, in addition to religious precepts recorded in the Quraan, ethnic culture further assigns appropriate roles for males and females on the basis of their gender. These prescriptions which are rooted in ethnicity are often given the aura of sanctity because the clergy are overwhelmingly male. If this were not so then Muslim women of Turkey, the Philippines and Indonesia would not be less restricted than those of Saudi Arabia, Iran and Pakistan, since the Quraan is the same for all of them” Likewise, there exists a conflict between the progressive camps and fundamentalists in Pakistani society and other Islamic nations (Weiss, 1985: and, Mernissi, 1987). 31 Another route to female literacy besides formal schooling in Pakistan is mm, but this is an infrequent practice. One can hire an outside tutor or use the guidance of elders: the first course is chancy and the second is occasional and informal. This would teach a very few how to read and write. Whether to educate their daughters or not really depends upon parents' attitudes which vary across sections of the population. (For attitudes of women towards their daughter's formal education in rural and urban areas, see Khan and Bilquees, 1976: and Sathar and Kazi, 1988 respectively: and Shah, 1986). Any type of education for women must be completed before marriage. After marriage she must bear and rear children and serve her husband and (frequently) parents-in-law, and studying for a diploma is seen as constraining these responsibilities. Those exceptional wives who go back to school for a diploma are overburdened as they must perform two tasks (one of being a student and one assigned as a woman at home), thus the chances for quitting school for them.are high. Females who do climb the social barricades against female literacy gain not only the skills of using the written word but also the advantages of higher social and economic status. Literacy can be treated by a few as an explicit way to improve the chances of an unmarried woman's family of origin to 32 betroth her to an occupationally well-placed man (because people are very explicit about material gains, household goods and money at the time of wedding). (Appendix Table 1.2 explored this possibility). I have argued above that literate women are likely to be given in marriage to higher-income households, and the effect of this income on the number of children the women desire is hypothesized to be positive. Nevertheless, I reason that the price effect should outweigh the income effect such that the net effect of female literacy on the desired number of children is negative. Thus, I adopted Easterlin's (1975) economic framework of analysis for the current research which is discussed below. It is worthwhile to point out here that Lee and Bulatao's (1983) framework for fertility analysis, unlike Easterlin's (1975) framework, is more elaborate to accommodate many social, economic, cultural, institutional, environmental and demographic factors. It is more flexible to allow couples to sequentially revise their desired number of children according to their nuptiality history and childbearing experience. The 1984-85 Pakistan Contraceptive Prevalence Survey data does not provide us with entire reproductive histories for women to examine the changing effects on the demand for children. Neither does it provide information on the value of children 33 by their sex in determining parent's desire for a certain number of children: nor on the number of children considered appropriate by a respondent for a family like hers by the number of sons or daughters desired. These limitations of our data to testing the Bulatao-Lee model led us to prefer the Easterlin model. The development process of Easterlin's framework, its conflicts and similarities with earlier and later concepts as forwarded by Becker in the 19603 and the early 19703 have been discussed by Sanderson (1976). However, the greatest convergence over time in the Becker-Easterlin.model have left little room for conflict in the model by the end of 1976 (for a simple chronology of the arguments between the Becker's and Easterlin's thoughts, see Sanderson, 1976). The modified version (limited to women who are assumed to have completed their family size) of the model has also been tested by Easterlin and Crimmins (1982). However, Easterlin's (1975) framework appeared to be an appropriate model for analysis of 1984-85 Pakistan Contraceptive Prevalence Survey data. Easterlin's Theoretical Framework Of Fertility and Contraceptive Use The theoretical framework used in this study borrows four important concepts from Easterlin's (1975) economic framework for fertility analysis. They are as follows: 34 (1) Demand_fgr_ghilgren (Cd): the number of children desired by parents if fertility regulation were costless: (2) Supply_gf_ghilgren (On): the number of surviving children parents would have without any deliberate fertility control: (3) WW (Cr): the outlay which includes all material, physical and psychological costs incurred by parents in order to use family planning methods: (4) ugtiygtign_tg_ggntraggpt (Mt): a situation when for a couple the supply of children (Cn) exceeds or equals the demand for children (Cd). In Easterlin's framework the economic determinants of the demand, for children. are couples's income, the price of children, and.theicoup1e's tastes for children vis a‘vis other consumer' durables. A. way in ‘which literacy should be associated ‘with fertility is be reducing’ the number' of children a woman and her husband desire. It is unlikely that literate parents would want to rear illiterate children. Thus, the children of literate parents cost more, especially when the children attain literacy through formal schooling and require tuition fee, uniforms and the like (see Appendix Table 1.3, based on the 1984-85 Pakistan Contraceptive Prevalence Survey). As such Easterlin's (1975) model does not allow for 35 accounting differences between husbands' and.wives' fertility goals. The supply’ of’ children (the ibiological capacity‘ to produce surviving offspring) is determined by the woman's age at entry into a sexual union (her age at marriage), any factor which regulates the frequency of sexual intercourse (for example, a postpartum taboo against sexual intercourse during breastfeeding in many African countries, spousal separation due to labor migration, and restriction on sexual intercourse during the day while fasting in the month of Ramadan among Muslims), and infecundity from involuntary causes (such as, absolute sterility, uterine cancer, infectious disease, and fetal mortality from involuntary causes). Fertility experience among women may vary due to differences in factors related to "natural fertility" and "fecundity." ‘Literate *women. should. have lower "natural fertility": i.e., a lower capacity to produce living children in the absence of any attempt to regulate reproduction. The factors which affect natural fertility are: (1) a delayed marital age when fecundity is supposedly on the decline: (2) infant and child mortality experience by women due to which they' may revise their initial fertility’ goals: and (3) involuntary abstinence due to temporary separation between spouses, such as, occurring as a result of male migration from 36 rural to urban areas for employment purposes. Fecundity is the biological capacity of a woman to reproduce and unlike natural fertility, it is not.governed.by the death of children outside the womb or by the separation of spouses“ The lack.of fecundity (infecundity) affects, anovulatory cycle, conception and fertilization process of ovum, implantation of the fertilized ovum in uterus and fetal growth. However, the effects of spousal separation for labor’ migration, the incidence of untreated infections and other such factors do interfere with natural birth intervals but the lower age at marriage provides longer exposure to reproduction process and has a positive effect on natural fertility in the absence of the above factors- Marriage is universal in Pakistan and all childbearing and sexual intercourse occur within ii» The entry into a sexual union as marked by the age at marriage is the most important. determinant. of the length. of exposure tot the possibility of pregnancy for a woman in Pakistan. Also, since fecundity declines gradually with age starting in the late twenties, women who wed late may have a harder time conceiving. We expected that the relationship of female literacy to female age at marriage for PCPS 1984-85 must conform to findings from previous studies (see.Appendix Table 1.4). 37 There are many costs associated with the use of family planning: such as, monetary, physical and psychological. In Pakistan, the monetary cost of fertility control methods is minimal if obtained from a government family welfare center or a maternal and child health center. The methods or services are provided free through the nationally subsidized family planning program. Family planning services and methods are located in government hospitals, drug stores, family welfare centers, maternal and child health care centers, grocery stores and with traditional birth attendants. Despite this variety of sources, the service points are not dense, especially in rural areas. The efforts to reach the sources of family planning involve some travelling costs in terms of money, time and hazards of transportation, on the part of the consumers. The physical costs include the fatigue from travelling to the service centers, especially when the service centers are several miles away and public or private transportation to them are inadequate or unavailable. Other physical costs are the side effects from actual use of contraception: such as, morning sickness from taking the pill. Among the psychological costs, a major factor is the element of shyness among those who want to use contraception. This element usually creeps inmwith.the fear of usurping God's power, being labelled as "immoral" by the community, or 38 sometimes interfering with the pleasure of husband or his will or the will of parents-in-law. Most women face difficulty in the public character of obtaining the method of their choice or in reaching the specific sources of supply. Because literacy ought to increase the knowledge of how contraception works and how to obtain contraception, literacy should reduce some of the market and psychological costs of acquiring contraception. Because literacy makes social innovation possible and is one of the very few paths to it for Pakistani women, we hold that female literacy reduces the several psychological costs of adopting contraception. Thus, literate women should be more likely to contracept, if their'market and psychological costs for doing so are lower. Figure-1 depicts the process of social modernization (the process by which most females in a society historically achieve literacy) as a function time (see the horizontal axis t). As long as natural fertility remains below the demand for children, there can be no motivation to contracept, for there is an excess of demand. During the early course of modernization, even when the child demand curve (Cd) remains fixed, the child supply (Cn) curve usually rises (possibly due to improvement in health conditions and reduction in infant mortality over time). Eventually a point x is reached beyond which the ability to produce children outstrips the demand for them. Only then can a motivation to regulate fertility by 39 Figure 2.1: A Hypothetical Model of Literacy- Modernization, Desire For Children and Contraceptive Use. N 11 CD m / b w Cd 3 * r /* * * .1 Cn - Cd * \ .z’rif/’,r V”/;u«fi§;; excess demand supply SOHD-I-‘P-I'J‘O HIO 'Cn = supply of children Cd = demand for children 40 means of contraception exist. However, this motivation will not be followed up with actual use so long as the costs of exceeding one's demand for children fall below the costs of regulating fertility. We hold that for literate women, the former costs will far exceed the latter costs such that literate women will be more likely to contracept. Easterlin (1975) assumes the married couple as the unit of analysis: it is the sole decision maker with regard to the couple's fertility goals. This model can not account for those populations where couples do not make such decisions to maximize their economic utilities. Also, it does not provide the precise link between couple's fertility decision and the societal structure (McHenry, 1985: Meyering, 1990). The model further assumes that a couple may decide at an earlier period of their married life the number of children they would want to have eventually under their given "preferences" and "budget constraints." Thus, it presents a ”single—period planning function" dealing with "static" situation and leaves little room for couples to revise their desired number of children in view of their reproductive history. Also, the dynamics involving the timings and reasons for entering marital unions are ignored. Nevertheless, the model does incorporate infant and child mortality by assuming that natural fertility will be lower in populations (such as 41 Pakistan) where these rates of youthful mortality are high.and will thus be more likely to fall short of the number of children desired. However, the most important feature of Easterlin's model is the inclusion of cost of fertility regulation which affects the contraceptive adoption in conjunction with motivation to regulate fertility. Hence, the actual contraceptive adoption will follow if the contraceptive methods are affordable (cost is not too high). This does not imply that in societies where cost of contraception is low its prevalence be high too. Therefore, the necessary condition for contraceptive adoption is the motivation to contracept. Thus, Meyering (1990) is concerned with the general applicability of the model in pretransitional populations: where ”desired fertility exceeds natural fertility, the application of Easterlin's model becomes limited" and its validity becomes questionable in explaining fertility transitions (p.57). For the purposes of current study, within Easterlin's framework, the selected variables can be precisely specified and also the social modernization variable--literacy-- can be interpreted in its broader perspective. However, Meyering (1990) pointed out that no satisfactory theory of fertility transition has yet evolved which should fully explain, "the shift from natural fertility to family limitation" in the macro-micro structure of a society. 42 The process of human reproduction can be altered by interfering with any of the following three dimensions: the capacity to produce surviving children, the demand for children, and the cost of fertility regulation through.modern means. In our literature review, we examined the antinatalist role of female literacy in changing all three dimensions. Below, we offer five formal hypotheses derived from this conclusion. Major Hypotheses HmtheeiLl: Rural wives have greater illiteracy than.urban wives. Weiss: Wives with greater illiteracy have higher deficit fertility. W3: Wives with higher deficit fertility have lower current use of contraception. 3299321131134: The lower current use of contraception by rural wives will be sustained when the wife's life cycle status is controlled. 43 Mess—5: The lower current use of contraception by rural wives will vanish when wife's life cycle status, literacy status and deficit fertility have been controlled. CHAPTER 3 DATA AND METHODOLOGY The Data This research adopted a multivariate approach to investigate the rural-urban difference in current contraceptive usage by married couples in Pakistan. The data for this task were from the Pakistan Contraceptive Prevalence Survey (PCPS) of 1984-85. The PCPS was planned by the National Institute of Population Studies (NIPS), under the sponsorship of the United States Agency for International Development (USAID) with technical assistance from Westinghouse Inc. (now Macrosystems). Fertility surveys had been conducted in Pakistan periodically. However, the PCPS was the first survey of the series established. to assess ‘the level of’ contraceptive prevalence in Pakistan on a regular basis. The need to establish the PCPS as a regular source of data was felt for the reason that the Sixth-Five-Year-Plan (1983-88) of Pakistan had increased the emphasis on, and financial allocation to, the family planning program (FPP) in the country. Additionally, with continually changing FPP strategies, it was necessary to evaluate the FPP on a regular basis. The data were collected during the months of October, 1984, to March, 1985. The major questions in the survey were 44 45 the number of children considered appropriate by couples, contraceptive adoption, and availability of FPP services. Since in-depth multivariate analyses were considered highly desirable by Pakistani Officials (Population Welfare Division, 1986: p.53 and p.82), information was collected on wife's current age, age at marriage, literacy status, and children ever born. These variables are necessary to test our five hypotheses: therefore, the 1984-85 Pakistan Contraceptive Prevalence Survey was judged appropriate to the goals of this research. The tasks of survey design and.planning, data collection and data processing were jointly carried out by the PWD and the Federal Bureau of Statistics (FBS), Statistics Division, Karachi. The national-level sample survey covered 96 percent of the country's population, representing both rural and urban areas: Federally administered tribal areas in Northern Pakistan and cantonment areas were excluded from the survey. About 30 percent of the total Pakistani population reside in urban Pakistan, and 70 percent in rural areas. However, the urban population was over-sampled by 10 percent to allow the coverage of a more heterogeneous population. Thus, the sample consisted of 40% of an urban and 60% of a rural population. A two-stage stratified sample was selected. In the first stage, the (primary) sampling units were enumeration blocks 46 (EBs) in urban areas and W in rural areas. Each enumeration block consisted of 200—300 households, whereas parallel to this was a village in the rural area, varying by numbers of households, but inhabited by about 2000 persons. At the first stage, in urban areas, enumeration blocks were drawn with a probability proportionate to the household numbers (obtained from the household listing prepared by the Population Census Organization at the time of the 1980 housing census prior to the 1981 population census or quick count survey). At the first stage, in rural areas, villages were selected with a probability proportionate to population size (estimated in the 1981 census). At the second stage, a systematic sampling technique led to the selection of 7398 households (2944 in urban and 4454 in rural areas), of which 154 households could not be reached for various reason: such as, the dwellings were locked or vacant or the residents were absent. Out of those 154 households which could not be reached, 64 were located in rural and 90 were located in urban areas. Out of 7244 (7398 - 154 = 7244) enumerated households 7459 "eligible" women (2908 in urban and 4497 in rural areas, respectively) were identified. Women aged 15-49 and living in marital unions with husband present were considered eligible 47 for interview. Women whose husbands were away for more than six months were not interviewed. Information on remarriage was not obtained: all women were treated as living in the first marital union. This assumption does not appear to be affecting our data for the reasons that even though.divorce is legal and allowed in Pakistan, the female divorce rate is very low because of the social unacceptability attached to it. Remarriages, though legally' and. religiously' allowed, are socially disapproved and thus often underreported in survey data. However, the divorce rates are lower in rural than urban areas (Table 3.1): 0.5% in rural, 0.6% in other urban and 0.8% in major areas (major urban areas included all large cities, such as, Karachi, Lahore, Faisalabad, Multan, Rawalpindi, Hyderabad, Gujranwala, Islamabad, Peshawar and Quetta: the other cities were included in "other urban" areas). The widowhood rate is higher at ages 45 and above, by the time most women had completed their family size. Table 3.1 also provides the overall marital status situation for rural, major urban as well as other urban areas. We noted from this table that likelihood of remarriage for widows is very low. Because female age at marriage is very low and a great proportion of females are married at younger ages, by the time (late ages) they become widows, due to higher male mortality, they might have already completed their reproduction span. 48 Table 3.1: The Percent Distribution of Women by Marital Status in Rural and Urban Pakistan, PCPS 1984-85. Marital Status Rural Major Urban Other Urban % % % Never married 18.2 26.8 23.4 Married 73.9 64.9 68.4 Widowed 7.4 7.7 7.5 Divorced 0.5 0.6 0.8 Total 100% 100% 100% Source: Population Welfare Division of Pakistan: Pakistan Contraceptive Prevalence Survey Report, 1986. 49 Additionally, the possibility of inclusion, in the survey, of temporarily visiting eligible women (guests) could not be ruled out. No attempt was made to separate those guests from the usual residents. However, this problem is not expected to pose any’ major analytical constraints The interviews were completed for 7405 out of 7459 eligible women. Fifty-four women who were temporarily away from home could not be interviewed. Out of these 54 women who could not be interviewed, 23 belonged to rural and 31 belonged to urban areas, respectively, for the reason they were not present at home at the time of interviewu INonetheless, the response rate was very high: 98.9% among urban and 99.5% among rural women. Hence, the completeness of the survey was excellent. Thus, the final sample size ‘became about .065% of the total currently married women 15-49 years of age (Population Welfare Division of Pakistan, 1986: pp. 14-15). Data Collection and Quality of Data The data were collected by interviewers using a Household Schedule and an Individual Questionnaire. The latter was a modified version of the standard questionnaire developed by the Westinghouse Contraceptive Prevalence Survey project to collect information on reproductive behavior from the women respondents. The forty-five female interviewers held at least a bachelor's degree: and some of them had previous experience of data collection. They underwent a one-week instructional 50 program organized by master trainers, who had been previously coached.by the senior PWD and F88 officials. ‘The interviewing and other field operations were closely scrutinized by field supervisors. These steps were undertaken to ensure a better quality data. Usually three types of errors are assumed to affect the quality of survey data: coverage errors creating differential completeness: respondents' biases in reporting of certain events or in providing an erroneous response: and interviewers' biases by misassessing the , occurrence of a certain event or a characteristic of a respondent (Shryock and Seigel, 1976). The ggyerage__bia§e§ induce errors in data due to differential coverage of an area or population. Generally, in rural areas, lack of roads and bridges may restrict the interviewers from reaching certain areas or respondents and can cause the rural population to be omitted to a greater extent than the urban population. However, as noted earlier the coverage rates were very good for both rural as well as urban areas: actually the number of household missed was lower for rural (1.4% = 64/4390) than urban (3.2% = 90/2854). Besponse_bia§e§ occur in age data because women do not know their birthdates or because they round off their ages to 51 end in certain preferred terminal digits. Likewise, women may understate their number of children ever born alive because they forget to mention those who died in childhood or who now live somewhere else. These two types of response errors have been detected before in Pakistani surveys and censuses (Booth and Shah, 1984: Zaki and Zaki, 1984). Therefore, we do not expect the PCPS to be errorless. Literacy is considered to play an important role in determining the quality of data. Idterate respondents may better understand the importance of reporting dates or numbers of vital events accurately than may illiterate respondents. As a matter of fact, such data suffer from biases in many societies but to a greater extent in developing countries with higher illiteracy. Errors of coverage (the inability to locate and interview sampled women) interact with errors of age reporting to bias downwardly‘ the number of ‘women observed in certain age brackets and to make these errors more pronounced possibly in rural areas. Those errors are usually less severe in survey than census data, since surveys interview smaller numbers of people and so can devote more resources to reducing errors. It is also possible that rural-to-urban migration in Pakistan may boost the percentages of urbanite women if they accompany their husbands to urban areas, especially in the young age 52 groups, independently of any errors in coverage or’ age reporting. Nevertheless, we compared data from the 1984-85 PCPS with data from the 1981 census to gauge the severity of these errors in the former data set. Findings from this comparison are described below. The comparison of rural women in the two data sources suggested an "overcount" of rural women at ages 20-24 by 3.2% (819.9% - 16.7%) and at ages 25 - 29 by 4.9%(=22.9% - 18%) in the 1984-85 Pakistan Contraceptive Prevalence Survey (see Table 3.2). This overcount was unexpected since rural—to- urban migration between 1981-85 should have reduced the percentages of young rural women observed at those ages. These apparent overcounts probably resulted from the greater ability of the survey to locate‘wives aged 20-29, from an upward bias in the reporting of age by interviewed women who were truly aged 15-19, and from a downward bias in reporting of age by interviewed women who were truly aged 30- 34. The mean age of wives at the time of the survey was 30.3 years for rural and 31.0 years for urban women. It is impossible to quantify the degree of each of these three sources of error, but we believe that age misreporting must have been more serious than the undercounts of rural women, particularly at ages 25-29. This evaluation seems to be true for urban women aged 25-29 whose numbers were 2.8% (=22.0% - 53 Table 3.2: Percentage Distribution of Sample Women in PCPS 1984-85 and 1981 Population Census in Pakistan by Place of Residence. PCPS 1981 Census Age Groups Rural Urban Rural Urban % % % % 15 - 19 8.1 6.7 8.6 7.3 20 - 24 19.9 17.3 16.7 17.6 25 - 29 22.9 22.0 18.0 19.2 30 - 34 15.2 16.5 16.4 16.6 35 - 39 15.1 17.2 15.3 16.2 40 - 44 11.3 12.0 14.3 13.5 45 - 49 7.3 8.3 10.7 9.7 Total 100% 100% 100% 100% (N=4512) (N=2880) 54 19.2%) higher in the 1984-85 PCPS than the 1981 census. If this line of reasoning be true, then the more accurate coverage of women aged 20-29 in the 1984-85 PCPS should indicate a lower rate of literacy than what was reported for women aged 20-29 in the 1981 census, possibly because literate women are less likely to be missed by censuses. Unfortunately, the data to assess this possibility are unavailable. Unlike the younger age groups (less than 30 years), the comparison of rural women at ages 30-34 suggested an undercount by 1.2% (=15.2% - 16.4%), and at ages 40-44 by 3% (-11.3% - 14.3%) and by 3.4% (87.3% - 10.7%) at ages 45-49. The relative difference in age distributions from two sources at ages 35-39 is negligible (.2%=15.1% - 15.3%). Thus it appears that.overcount.at ages below 30 is somewhat.consistent with undercount at ages above 30 years. For urban women at ages 30-34 the difference is negligible (.1%=16.5% -16.6%). For women at ages 35-39, an overcount of 1% (=17.2% - 16.2%) has been noted: at ages 40-44 and 45-49, an undercount of 1.5% (=12.0% - 13.5%) and 1.4% (=8.3% - 9.7%) seems to exist. This observation agrees with our argument for rural women as well. However, whether ages below 30 indicate a better coverage or misstatement of age is very hard to verify. 55 Possible conclusions from the above evaluations are that the sample coverage appeared to be upwardly biased for younger women: thus, the greater misstatement of their ages possibly may correspond to the lower level of contraceptive use for them because the reproductive potential is much greater among younger women than among older women. For most of them, the number of currently surviving children may fall short of the number they considered appropriate for their families. Hence, the likelihood of contraceptive use might be lower for them and particularly for rural women. The reporting of contraceptive use (revealing information on a private sexually-related matter), in particular can be very embarrassing for Pakistani women. In a culture which considers talking about such matters, a shame, and which supports the Islamic emphasis on high fertility, marriage and family, many women may believe that contraceptive use is against religion. They may not report the use of any contraceptive method at all, even when they are using one. This type of underreporting has been observed in other South Asian countries, as well. For example, in the Bangladesh Condom User Survey, Ahmed e_t__a_l. , (1987) noted that among condom user couples, 11-12 percent of husbands and 15-44 percent of wives underreported the condom use. Bangladeshis residing in semi-rural places underreported to a greater extent than urban Bangladeshis. Again, this is possibly due 56 to 1more conservative rural cultural ‘values coupled. with greater prevalence of illiteracy than what it is in urban areas. Therefore, we expected underreporting of contraceptive use in PCPS, also. The 1984-85 Pakistan Contraceptive Prevalence Survey suspected that actual contraceptive use had not been reported correctly at the national level, for the reason that in order to be consistent with the Total Fertility Rate (TFR = 5.95 children per woman) , contraceptive use should have been around 15% rather than what was estimated from the actual reporting (9.1%). There is a discrepancy in the reporting of current contraceptive use as cited in Pakistan Contraceptive Prevalence Survey Report (Population Welfare Division, 1986): the current users at another place in the report are cited as 7.63% instead of 9.1% (see Table VIII.15 on page 97). Pullum M” (1988) provided a rate of 7.63% and suggested abandoning the use of the rate of 9.1% because he thought it was erroneous. My calculations showed a national rate of current contraceptive use of 7.9% (based on N=7394): 4.1% in rural areas and 13.7% in urban areas. However, I disagree with Pullum on the basis that he included the pregnant along with nonpregnant women in the denominator, although in principle, only nonpregnant women are candidates for current contraception. Hence, I excluded 1212 pregnant 57 women and 46 missing cases (1212 + 46 81258) from a total of 7394 readable cases from my data files and thus used 6136 (= 7394-1258) women as the denominator and 580 current users of all methods (580) as the numerator in my computations. This produced a rate of 9.45% (= 580/6136), which is very close to the rate 9.1% given in PCPS report for all methods. Exclusion of only the pregnant women from the denominator does provide a rate of 9.1% for all methods. Thus, the whole discrepancy is derived from the usage of different denominators. However, an estimate of "shy" contraceptive users (underreporters) was made by experts of the National Institute of Population Studies at the national (not by place of residence) level only» ‘With the inclusion of "shy users" they came up with a contraceptive prevalence rate of around 15%. We estimated the number of "shy users" for rural and urban areas who underreported contraceptive use. This exercise attempted to approximate the number of women who did not report use of any method and had not had any birth in last 3 years (PCPS estimate was based on last 5 years): they did not report themselves as sterile or menopausal yet they did not want to have any more children. Hence, it was suspected that they must be protected by some form of contraception. .58 Figure 3.1: An Approximation of Underreporting of Contraceptive Use in Bu:g1_A;ea§, PCPS 1984-85. Total Respondents 4512 Current User *Non Users 186 4324 Pregnant Not Pregnant 806 3518 Want More Children? I I 1 Yes No Don't know 1605 1435 478 I l I I Yes birth in last No birth in last 3 years 3 years 796 639 l - Perceived sterile Probably silent and menopausal users 82 557=12.9% * 2 cases are missing in rural areas. 5'? Figure 3.2: An Approximation of Underreporting of Contraceptive Use in U;ban_A;ga§, PCPS 1984-85. Total Respondents 2880 Current User Non Users 395 2485 Pregnant Not Pregnant 448 2038 Want More Children? 1 l 1 Yes No Don't know 786 935 317 i F 1 Yes birth in last No birth in last 3 years 3 years 505 430 l l Perceived sterile Probably and menopausal silent users 49 38l=15.3% 60 According to this approximation, 12.9% of rural nonusers (Figure 3.1) and 15.3% of urban nonusers (Figure 3.2) might.be actually ("silent users") using some kind of contraception no matter whatever reasons they gave for nonuse. The higher percent (implying a greater extent of underreporting) of urban "shy users" than that of rural, appears to be implausible and inconsistent with our expectation that greater literacy might lead to better reporting in urban areas. Because urban women have relatively easier access to family planning, it is possible that they could adopt contraception secretly and did not report it. Another reason for nonreporting of contraceptive use by her might be due to the lack of privacy (most urban housing structures have a smaller space or fewer rooms which house a large number of people, hence with the fear of being overheard by some one, the respondent did not report use even if she was interviewed at a private space). As such, this greater underreporting of contraceptive use by the urbanites would somewhat bias the data towards rejecting Hypothesis 5 when it is true. Nonetheless, no attempt was made to correct or smooth the data before analysis. Regarding the 111W. the subjective assessment of an event or a characteristic by the data collectors may result due to the faulty survey' design, phrasing of the question and probing, all of which are hard to check. However, the individual questionnaire was pretested in 61 rural and urban areas both for its linguistic flaws as well as its design problems in eliciting the correct responses from the respondent as well as the interviewer. Yet, the much higher education level of the female interviewers who were probably unmarried women of younger ages as compared to that of the female interviewees may have led to the social distance between the two groups of women that inhibited candor. The discussion of contraceptive use which is considered a personal and private-sexual matter may not even be an appropriate topic of discussion with an urbanite stranger (interviewer) in addition to the social distance posed by the higher level of education and urbanite appearance of the young interviewer as compared to the interviewee. Pullum (1988) remarked that ”sexuality and contraception are not approved topics of conversation, particularly between different generations--not even between daughters and mothers" (p.36). This remark is true for such an interaction among women in Pakistani culture. Ever since the Indianapolis Study of 1941 (Westoff, Mishler, and Kelly, 1957) , demographers have tried to measure orientations towards numbers of children as ways of predicting actual fertility. The W, such as the appropriate or ideal number of children, has always been under criticism in demographic literature. The concepts of appropriate or desired and undesired fertility are complex to measure but are important in the study of fertility desires. 62 The desired number of children has been considered meaningful in several developing countries. As a response to the question on desired family size, couples have given an actual number and some "verbal justification" for that selection (Lee and Bulatao, 1983: p.274). There are advantages and disadvantages associated with any method of measurement of the desired and undesired fertility. Ryder and Westoff (1971) , in order to determine the "degree of redundancy" compared the responses of wives to questions on. the number' of children intended, expected, desired and ideal for an American family, as revealed in data from the 1965 National Fertility Study of United States. The greatest overall variation‘was found in the number of children already born (current parity) . A statistically nonsignificant small difference in the additional number'of'children intended or expected was found. The absolute differences by race, religion and education were also small. Eighty-seven percent of the women in the sample gave a consistent response to ”intended" and "expected" number of children. Therefore, the total (lifetime) numbers of children expected.was judged.to be synonymous with. the total numbers intended. These two measures had the strongest correlation with actual numbers of children ever born but that may have happened only because the lifetime number of births intended or expected included the number already born. 63 The reliability ratio measuring the consistency of individual responses was also compared for the number of children considered "ideal" and "desired". It was argued that the ideal number of children may have been measured unreliably because it had the weakest correlation with actual number of children ever born. That is, some women may have stated the number of children they thought was ideal for an average American family: other women may have replied with reference 'bo a privileged American family, and still others may have referred to themselves. Such a reSponse could be further biased because respondents with different characteristics could have responded differently to the ambiguity. Ryder and Westoff (1971) concluded that the "ideal" number of children was relatively unreliable variable which lacked the "face validity" as well. However, there was no difference in the mean number of ideal and the mean number of desired children for the total population but for the subgroups of education by race, religion and education. In ‘the context of Pakistani culture, Pullum (1988) investigated the norms regarding the 'ideal' family size on the basis of the question on "the appropriate number of children for a family like yours" from PCPS data. He argued that a woman who receives more social rewards for having a larger number of children and who is fatalistic about reproduction would likely report a relatively higher "random" 64 number of children considered ideal. Pullum further argued that if the ideal number reflects the couples' attitudes because of social pressure to have a large number of children (and fertility is considered in the realm of human control) then it should be regarded as a reliable "family size norm," not just a "random" response. Fatalism would imply that fertility is out of human control (it is God's will) and that the contraception is useless. When the norm regarding family size changes in favor of smaller families, reproduction can be considered a behavior which can be consciously controlled by couples. Pullum (1988) considered the :mean ideal number' of children (4.90) in Pakistan to be quite high which roughly corresponded with the observed mean number of children born among women of ages, 30- 34 (4.99) .and 35-39 (5.10): the age by which probably the family size is completed by most women in that society. Thus Pullum's (1988) findings suggested a high correspondence between ideal and actual number of children ever born such that the former measure could predict the latter. The findings from the World Fertility Survey also supported the notion that couples in.many developing countries of the world attempt to attain a desired goal-- "an ideal number of children", which determine their motivation for adoption of family planning (Kent and Larson, 1982). 65 Nevertheless, the insights from the World Fertility Survey (Lightbourne, 1985) indicated that the contrast between the preferred (desired) and actual family size provided a better estimate of the desired number of children while comparing the responses at two time periods than the question regarding the desired number of additional children or whether the last birth was wanted or unwanted. Lightbourne (1985), also performed inter-item consistency tests to check the validity of responses to the question on ideal number of children in the World Fertility Survey, based on the single set of data. In order to test the validity of the responses to the question on the appropriate number of children, inter-item consistency checks were performed by comparison of responses to the question.on future desire for additional children. 'The ideal number of children for a respondent in our survey was assessed by the question: "In your opinion what is the appropriate number of children for a family like yours?" The desire for future children was assessed by the question: "Do you want to have children in future (in addition to the one (if) you are expecting)?" The analysis of this comparison is provided in Table 3.3 for rural residents and in Table 3.4 for urban residents. We noted for rural residents (Table 3.3), 96.3% of those who said 66 Table 3.3: Percent of Women Wanting More/No More Children By How the Current Number of Living Children Compares to the Number Deemed Appropriate: Rural Pakistani Women, PCPS 1984-85. Want Approp. # Approp. # Approp. # more/ less than equal to more than Total Approp. Living # Living # Living # % % % Want more 1.4 2.3 96.3 100% (1845) Want no more 24.3 57.1 18.6 100% (1729) Unknown or uncertain 0.3 3.4 96.3 100% (595) Total sample = 4514 Missing cases = 345 Study sample = 4169 (=4514-345) 67 they wanted more births had fewer living children than they considered appropriate to have for a family like theirs. A large majority (81.4% = 24.3% + 57.1%) of women who said they did not want more births had stated an appropriate number of births that was less than or equal to their surviving number of children. These strong correlations suggested that consistent answers were being given to the questions about appropriate number of children and the desirability of more births in view of the number of living children. 'The greatest inconsistency was that 18.6% of women who said they wanted no more births stated an appropriate number of children that was above their currently surviving number. It.is possible that.most.of these women just for the sake of courtesy to interviewer'might.have stated they did.not.want any more or just gave a random response. Such inconsistency in responses to preference data are quite common: Lightbourne (1985) suggested a similar pattern of inconsistency for the countries in the World Fertility Survey, where 36% of women who wanted no more births gave a preferred number of children that was above their currently living number. On the other hand, 93.7% of urban women (Table 3.4) who said they wanted more children stated an appropriate number that was higher than their currently surviving children, and this indicated that the response was not necessarily a "random" or a "careless" answer. 68 Table 3.4: Percent of Women Wanting More/No More Children By How the Current Number of Living Children Compares to the Number Deemed Appropriate: Urban Pakistani Women, PCPS 1984-85. Want Approp. # Approp. # Approp. # more/ less than equal to more than Total Approp. Living # Living # Living # % % % Want more 2.9 3.4 93.7 100% (969) Want no more 34.6 48.9 16.4 100% (1369) Unknown or uncertain 2.3 4.9 92.9 100% (350) Total sample = 2880 Missing cases = 192 Study sample = 2688 (2880 - 192). 69 Consistently, 83.5% (=34.6% + 48.9%) of the urban women who said they wanted to cease childbearing stated an appropriate number of children which was less than or equal to their actual number of living children. These relationships suggest two conclusions: ( 1) the comparison of the stated appropriate number and actual number of living children gives a reliable measure of which women do and do not want further pregnancies: and (2) the differences in reliability are not great between rural and urban women. A similar level of response-consistency, 85.3% of rural and 85.5% of urban women indicated no desire for future birth because their "ideal family size" was greater than their number of living children, was found in 1979-80 Population, Labour Force and Migration Survey of Pakistan (Ali, 1989). We further examined the consistency between responses to questions on appropriate number of children with the past desire for last child. Tables 3.5 and 3.6 present this analysis for rural and urban respondents, respectively. Whether the last born child was wanted or unwanted was measured by the question: "Before your last child, did you want to have more children?" A low'percent of rural women who said their last child was wanted (22.3%) reported that their living number of children was less than their appropriate number. However, a high percent of both rural (64.8%) and urban (59.8%) women who said their last child was wanted also 70 Table 3.5: Percent of Women Who Wanted/Did Not Want The Last Child By How the Current Number of Living Children Compares The Number Deemed Appropriate: Rural Pakistani Women, PCPS 1984-85. Last child Approp. # Approp. # Approp. # wantedness / less than equal to more than Total Approp. # Living # Living # Living # % % % Last wanted 12.9 64.8 22.3 100% (937) Last unwanted 39.2 45.6 15.2 100% (712) Uncertain 16.5 54.0 29.5 100% (139) Total sample=4514 Missing cases=1788 Study sample = 2726 (=4514-1788). 71 Table 3.6: Percent of Women Who Wanted/Did Not Want The Last Child By How the Current Number of Living Children Compares The Number Deemed Appropriate: grrag Pakistani Women, PCPS 1984-85.* Last child Approp. # Approp. # Approp. # wantedness/ less than equal to more than Total Approp. # Living # Living # Living # % % % Last wanted 18.3 59.8 21.8 100% (655) Last unwanted 49.1 38.2 12.8 100% (705) Uncertain 29.6 48.1 22.2 100% (54) 'IAll women who have had no prior birth have been excluded in this table and are reported along the missing cases as given below. ‘ Total sample = 2880 Missing cases = 1466 Study sample = 1414 (=2880 - 1414). 72 said their appropriate number of children equalled their actual living number. Consistently, a near-majority (49.1%) of urban women who said their last child was unwanted also said that their living number of children exceeded the number of children deemed appropriate for a family like theirs. Inconsistently, only 39.2% of rural women who said their last child was unwanted also said their number of living children exceeded the appropriate number. Either, the difference between the current number of living children and the number deemed appropriate was less reliably measured for the rural women, or else the rural women are more likely to revise upwardly the number of children deemed appropriate to incorporate births that were prospectively unwanted. We performed an additional inter-item consistency check by comparing wife's desire for future children to her husband's desire for future children (which was also reported by her), and by current use of contraception for rural and urban residents in Tables 3.7 and 3.8, respectively. We observed that the percentage of couples using contraception when neither husband nor wife wanted more births was higher than when one or both spouses wanted more, was higher for urban couples (27.2%) than rural couples (11.3%) in this situation, and higher than the national rate (9.45%). This concordance suggested that the desire for future births was a reliable predictor of current contraceptive use. Similarly, 73 Table 3.7: Percent of All Couples by Future Desire of Additional Children for Rural Residents, PCPS 1984-85. Future Desire Contraceptive Use Yes No No Response Total % % % Both want more 1.3 52.0 46.7 100% (N=1912) Husband wants more 4.6 50.3 45.1 100% (N=306) Wife wants more 0.0 48.4 51.6 100% (N=31) Neither wants more 11.3 53.8 34.9 100% (N=1265) Total sample = 4512 Missing cases = 392 Uncertain cases = 606 Study sample = 3514 [(= 4512 - (392=606)]. 74 Table 3.8: Percent of All Couples by Future Desire of Additional Children for Urugu Residents, PCPS 1984-85. Future Desire Contraceptive Use Yes No No Response Total % % % Both want . more 5.5 57.1 37.4 100% (N=961) Husband wants more 10.8 63.2 26.0 100% (N=204) Wife wants more 7.1 65.7 27.1 100% (N=70) Neither wants more 27.2 51.2 21.6 100% (N=1106) Total sample = 2880 Missing cases = 157 Uncertain cases = 382 Study sample = 2341 [(=2880 - (l57=382)] 75 the percent of current users was quite low when both spouses wanted more children: and the percent was lower for rural residents (1.3%) than urban residents (5.5%). This finding appears to be consistent with the desire for future children. When neither of the spouses wanted more children, the contraceptive use was the highest within the rural (11.3%) , as well as, within the urban (27.2%) areas (which are higher than the national average of 9-10%). Among those couples where both of the spouses wanted more children, and that possibly were currently contracepting for spacing purposes, the percent of users was lower (1.3%) for rural than urban (5.5%) couples. On the other hand, in a situation when one of the spouses did not want additional children the level of contraceptive use remained low for rural (4.6%) than urban (7.1%) residents. Nevertheless, the relatively lower level of contraceptive use for rural than urban residents did persist. It is possible that a considerably higher proportion of urban residents may be using contraception for spacing purposes than rural residents. Measurement and Operationalization of variables The bread accepts. xeriahles and Wetter; of the variables, used in this research are described below: 76 (1) W: Our goal was to explain the extremely low level of contraception in rural Pakistan. The current place of residence (independent variable) was measured as a dichotomy: urban = 1 and rural = 2, in this study. (2) s a s: Wife's literacy (an intervening variable) was measured by the question: "Can you read a newspaper and write a simple letter?" When respondent reported that she could (yes = 1), she was considered literate: and if the respondent said that she could not (No = 2), she was considered illiterate. In this way, the women who had attended school for some time but had not learned how to read and write were recorded as illiterate. Hence, this measure was considered a conservative indicator of literacy. (3) EEEEJIIIY_§IQEQ§: The deficit or excess fertility status (intervening variable) was computed as the current number of surviving children subtracted from the number of children considered appropriate by' the respondent for' her family. If the difference was positive, the respondent was found to have a "deficit fertility" and should have no demand for fertility regulation (contraception). If the difference was negative, the respondent had excess fertility and should be currently 77 jpracticing'contraception. The appropriate number of children was measured by asking the question: "In your opinion what is the appropriate number of children for a family like yours?" The response was recorded as a total number of children (irrespective of gender specification). This limited us to the study by comparing the total number of children desired with the number of children currently surviving. The number of currently surviving children was measured by the response to the following question: "How many of your children are living (alive) now?" (4) W: This composite control variable was computed by combining wife's (a) current age and (b) age at marriage. (a) £9:I§HL.§Q§ was measured by asking two questions: (1) "It is important in this study to know your exact age. In what month and year were you born?" The month and year of respondent's birth was recorded according to the English or any other calendar. The probing was done by asking a second question: (2) "How old are you?" The response was recorded in completed years. The consistency in responses was verified and then completed years of age were recorded as current age. If the respondent's age could not be determined by the first question and. by probing, the age was estimated. by the interviewer. If the respondent.happened to be below age 15 or above age 49, the interview was terminated. 78 (b) Age_er_merrlege was assessed as a response the following question: "What was your age when you got married?" If the probing was needed, it was done. 'The response was recorded in completed years. Since the majority (57%) of wives in our sample had married by age 17, we used this as a cut-off point, plus their current age, to operationalize three stages of family-building life-cycle status: (1) Early: wives with current ages below 30 years and age at marriage above 17 years: (2) Middle: wives with current ages below 30 years and age at marriage 17 years or lower: (3) Late: wives with current ages 30 years and higher whatever there age at marriage was. (5) W: The current use (dependent variable) of any contraceptive method, either by wife or by husband, as reported by the respondent was measured.on the basis of the question: "Are you or your husband currently using some family planning method or doing something to avoid a pregnancy?" A "Yes" response was coded as "1" and "No" response was coded as "2". Obviously, women who reportedly were currently pregnant were not asked this question. i Statistical Techniques I. Eeersen'e Gui-Square Test: Hypotheses l, 2, 3 end 5: In order to test.Hypotheses 1, 2, 3 and 4, Pearson's Chi- 79 square statistic was applied. The frequency distributions analyzed for these hypotheses consisted of two attribute classes, such as, in fiynerneele_l, place of residence (rural vs. urban): and literacy status (illiterate vs. literate). Both variables were presented in a 2 x 2 table. For W, deficit fertility (yes =- 0: no =- 1) and literacy (illiterate vs. literate) also were presented by 2 x 2 table. Similarly, in.nyuerneele_g, current use of contraception (yes = 1: no = 2) by deficit fertility (yes = 0: no = 1) was also presented by 2 x 2 classification. For the test of mneeje 1, we prepared a 2 x 3 [(contraceptive use (two classes) by life-cycle status (three classes)] table controlling for the place of residenoe. In order to establish the statistical significance of the hypothesized relationship for categorical data given in two-dimensional or three-dimensional tables, application of Pearson's Chi-square statistic (Pearson, 1899) was considered appropriate. The formula for computing Chi- square (X2) is: (Observed - Expected)2 X2 = S. , Expected Where W are estimated. by multiplying the row and column totals (n1i x nfi) for specific categories of the attributes in the table, and dividing that number by the whole sample total (N) and so on for all the cells of the table. Then the expected cell frequencies are subtracted from the observed cell frequencies and the deviation is squared: 80 the squared deviation is divided by the expected frequencies. Finally, 2 is the summation of all the cells of the table and represents the value of x2, which is referred to with the degrees of freedom (N - 1). The calculated value of Chi- square is then compared with the theoretical Chi-square distributions, for which tables are widely available in any good textbook of Statistics (Snedecor and Cochran, 1977). Usually the large values of Chi-square are significant. However, this significance is determined by comparing the calculated value of Chi-square with the theoretical value, at certain degrees of freedom and at a chosen level of significance (conventionally, 5% or 1%). If the calculated value of Chi-square is greater than the theoretical value of Chi-square, it is interpreted to be statistically significant: in other words, we can reject the stated hypothesis. If the calculated value is smaller than the theoretical value, we accept the hypothesis, and declare that the difference between the observed and the expected cell frequencies is statistically not significant. If we have proportions, ratios or percentages in the categories of the attributes (characteristics), Chi-square can not be computed. II. Wei: BYEQJL sis : Logit regression technique was adopted from the rubric of multivariate regression models. The logit-regression 81 subroutine is available in the Statistical Package for Social Sciences (SPSSx), 1986. Logit-regression is preferred to Ordinary Least Square (OLS) regression in analyzing current contraceptive use in Pakistan because the percentage of users represents a very small minority of all married couples. When the distribution across categories of a dichotomous dependent variable fall far from.a fifty-fifty' split, the assumption of homoscedasticity which is fundamental to OLS regression may not.be met. For this reason, logit regression must be used to study current use of contraception in Pakistan. The statistical theory on which logit regression is based is discussed in the following chapter. CHAPTER 4 ANALYSIS OF DATA AND FINDINGS Some Selected Characteristics of Pakistani WOmen Somewhat surprisingly, the 1984 Contraceptive Prevalence Survey of Pakistan showed that the average number of children ever'born.and.currently'surviving'per“woman was lower in.rural areas (3.36) than in urban areas (3.83) (Table 4.1). The lower rate of cumulative fertility for' the rural ‘women probably results from their greater propensity to breastfeed children for a long duration, which delays the return of post partum ovulation. This delay makes for lower cumulative fertility'by spacing births farther apart (Tsui, 1985: Cleland and Sathar, 1984: Rao and Beaujot, 1986). In addition, lack of adequate medical care (pre- and post- natal) may result in lower rates of survival following birth. A "desire to replace the lost child", might account for the much higher number of children deemed appropriate by rural than urban women (5.03 and 4.79, respectively). As such, there was a gap between the actual number of surviving children and the number believed appropriate by the typical respondent, but this gap was wider for the rural than for the urbanwwoman (-1.67 and -.96 surviving children, respectively). We named the gap "deficit fertility" and reasoned that disincentives to contracept. would. prevail until the Igap switched from a deficit to an excess. 82 83 Table 4.1: Some Selected Characteristics of Pakistani Women Based on The Pakistan Contraceptive Prevalence Survey 1984-85. Characteristic Rural Urban (N=4512) (N=2880) Mean number of children ever born 4.20 4.40 Mean number of surviving children 3.36 3.83 Mean number of appropriate children 5.03 4.79 Deficit fertility -l.67 -0.96 Mean age of wife 30.30 yrs. 31.03 yrs. Mean age of wife at marriage 17.21 " 17.65 " Percent literate wives 4.45 9.79 Percent of couples currentlx using any contraceptive method 4.20 13.70 Source: Pakistan Contraceptive Prevalence Survey 1984-85. The Total study sample size - 7392 cases after excluding 11 missing cases from my data file and 2 missing cases in contraceptive use. 84 For reasons cited in the literature review above, literacy should reduce a wife's "demand for surviving children” (the number of surviving offspring she deems appropriate for a family like hers) and raise her ability to keep the living ones alive. If so, the movement from deficit fertility to excess fertility should be faster for literate than for illiterate women. Thus, if rural wives are less likely than urban ones to achieve literacy, this fact would interpret the wider gap in deficit fertility just reported for the rural wives. This possibility motivated Hypothesis 1. Literacy by Place of Residence: Hypothesis 1 The percent of respondents who reported they could read a newspaper and write a simple letter was more than twice as high in urban as in rural areas (9.79 and 4.45 percent respectively: see Table 4.2). At 1 degree of freedom, the chi-square statistic of 67.23 was highly significant with an alpha less than five percent. Hence, we could not reject Hypothesis-1, that the literacy rate of rural wives would be lower than that of urban wives. Fertility by Literacy: Hypothesis 2 If the attainment of literacy means that a Pakistani woman will report a smaller number of surviving children as appropriate for a family like hers and will more successfully safeguard the survival of those already born to her, then she 85 Table 4.2: Literacy Status of Currently Non-Pregnant Wives by Urban-Rural Place of Residence in Pakistan, 1984-85. Residence Literacy Status Urban Rural % % Literate 9.79 4.45 Illiterate 90.21 95.55 Total 100% 100% (N=2432) (N=3706) Note: Chi-square = 67.23, D.F. = 1, P value < 0.05. Total sample size = 7394 Excluded: pregnant women a 1212 and missing 44 cases on pregnancy status (1212 + 44 = 1256) Study sample size = 6138 (= 7394 - 1256) 86 should more rapidly depart from a state of deficient fertility (Hypothesis 2). IHowever, our data (Table 4.3) showed that the percent of women for whom the number of surviving children fell short of the number deemed appropriate was almost the same for literate (82.01 percent) as for illiterates (84.80 percent). At one degree of freedom, the chi-square statistic of 1.908 was not statistically significant at the 5 percent level, a result rejecting Hypothesis 2. Contraceptive Use by Fertility Status: Hypothesis 3 Earlier, we argued that a Pakistani woman whose number of surviving children did not reach or exceed the number she deemed appropriate for a family like hers would be far less likely to avail herself of current contraception (Hypothesis 3). It was necessary to limit the test of this hypothesis to currently married women who were not pregnant at the interviewu .Among such women, 8.15 percent of those with fewer surviving children than they deemed appropriate for a family like theirs were currently protected by contraception as compared to 20.14 percent of those whose number of surviving children equalled or exceeded what they considered appropriate (Table 4.4). This statistically significant difference (Chi- square - 117.449 with one degree of freedom) meant we could not reject Hypothesis 3 but portended a rejection of Hypothesis 5. 87 Table 4.3: Deficit Fertility Among Currently Non- Pregnant Wives by.Literacy Status in Pakistan, 1984—85 . Literacy Status Deficit Fertility Literate Illiterate % % [fig 82.01 84.80 (number of currently surviving children less than the number desired). HQ 17.99 15.20 (Number of currently surviving children equal to or more than the number desired). Total 100% 100% (N=378) (N=5337) Note: Chi-square = 1.908, D. F. = 1, P > 0.05. * All additional cases with no responses and non-numeric responses (n=423) regarding the appropriate number of children along with currently pregnant women (1212) with missing cases (44) have been excluded from the analyses. The study sample size = 5715 (=7394 [- 423 + 1212 + 44]). 88 Table 4.4: Current Use of Contraceptives Among Currently Non-Pregnant Wives by Deficit Fertility in Pakistan, PCPS 1984-85‘. Deficit Fertility Contraceptive Use Yes No % % Yes 8.15 20.14 No 91.85 79.86 Total 100% 100% (N=4834) (N=879) Note: Chi-square = 117.449, D.F. = 1, P < 0.05. *See footnote to Table 4.3: additional 2 cases with missing information on contraceptive use have also been excluded from this analysis. Study sample size = 5713 (= 4834 + 879). 89 Contraceptive Use by Residence by Family—Building Life-Cycle: Hypothesis 4 The percent of wives who themselves or whose husbands were currently contracepting was 4.20% in rural areas but 13.70% in urban areas. Rural wives had a younger average age at marriage than did urban wives (17.21 and 17.65 years, respectively) but also a younger average age at interview (30.30 and 31.03 years, respectively). These findings meant the average duration of marriage was much shorter for rural than urban wives (13.09 years and 13.38 years respectively). We reasoned, therefore, that rural wives were in the earlier stages of the reproductive life cycle (commenced by them at marriage) and were less likely to want to contracept since they would have had less time to bear the number of surviving children they think is appropriate. The majority (57%) of all wives in our sample had married by age 17. We used this cut- off, plus their current age, to operationalize three stages of the reproductive life cycle: (1) Early} wives with current ages below'30 years and age at marriage above 17 years: (2) Middle: wives with current ages below 30 years and age at marriage 17 years or lower: (3) Late: wives with current ages 30 years and higher. However, we reasoned that even when differences in the stage of the reproductive life cycle had been controlled, the 90 prevalence of contraception would remain lower for rural than for urban couples (Hypothesis 4). We examined the contraceptive prevalence by family- building life cycle and by place of residence through simple cross-tabulations and found that the percent of current users among rural wives at all family building stages was much lower than among urban wives (Table 4.5). This result could not contradict Hypothesis 4. Contraceptive Use by Literacy by Fertility: Hypothesis 5 So far, a lower contraceptive prevalence has been reported for women with fewer surviving children than what they deemed appropriate for families like theirs. Contrary to what was hypothesized, illiterate women were not more likely to report such a deficit in fertility. Therefore, while illiteracy was more prevalent among rural wives, it was unlikely to interpret the lower prevalence of contraception among them, as had been stated in Hypothesis 5. We performed a logit regression analysis for this hypothesis. Selection of the Best-Fitted Logit-Regression Mbdel In the tg§t_gf_flypgthg§1§_§, wife's literacy status and deficiency in fertility (predictor-variables) were added to measures of her residence and her stage in the reproductive life cycle to predict whether she was currently protected by 91 Table 4.5: Percent Contraceptive Use Among Currently Non-Pregnant Wives by Family-Building Stages and by Place of Residence in Pakistan, PCPS 1984-85. Family-Building Stage / Place Of Residence Contraceptive Use Urban Rural % % Early Middle Late Early Middle Late Yes 11.4 9.3 22.1 2.5 3.2 7.4 No 88.6 90.7 77.9 97.5 96.8 92.6 Total 100% 100% 100% 100% 100% 100% (N=517) (N=69l) (N=1224) (N=757) (N=1210) (N=1737) Note: Chi Square: Urban Rural 65.048, D. 38.287, D. = F 2, P < 0.05. = F 2, P < 0.05. 'See note to Table 4.2 and also exclude 2 missing cases on contraceptive use. The study sample = 6136 (=6138-2). 92 contraception. Of course, the test was confined to nonpregnant wives of reproductive age. Contraceptive use assumes a dichotomous response, the values of 1 = yes: 2 = no. The odds of getting a 'yes' response (p=percent or proportion of affirmative replies) are (p / (1 - p)). It can be estimated by maximum likelihood estimation (MLE) procedures (as described earlier in Chapter 3). By using the MLE procedure an attempt is made to solve the following equation: ln(P/(l-PH = L + BX where, ln(p/(l-p)), is called the log-odds (logit) of a yes. The above equation can be solved for p as: p = 1 / (1 + e (L+BX)). To solve for both outcomes y = 1, or y = 2, the combined solution can be made. In a similar way, the probability of whole sample can be obtained. Therefore, the likelihood function of (L, B) is that pair of parameters which minimizes the difference between the observed and expected cell means. We estimated the cell means expected under a given logit regression model using an iterative procedure defined by Fienberg (1981:37-40). The SPSS computer program ceased the computations when the change in the estimated cell means from 93 one cycle to the next became very small. The Wis; (Ga) of a logit model, is assumed to follow approximately a chi-square distribution. 62 tells the association between the observed and the estimated (fitted) cell means. The smaller value of G2 provides the better fit (Aldrich and Nelson, 1984) . A saturated model is one that contains all the dependent, independent, and control variables in the study, plus all possible interaction terms linking them. The G2 of a saturated model is zero, and there are no degrees of freedom. However, it is preferable to base interpretations on a smaller (reduced, unsaturated) model, provided that the discrepancies it produces between observed and expected cell means are very small. There are two steps in selecting the best-fitting reduced model: (a) Define a reduced model by eliminating a term or set of terms from the saturated model. The G2 and the number of degrees of freedom will be nonzero for the reduced model, because terms have been given up and the expected and observed cell means have become discrepant. The difference in the Gz, between the more and less inclusive models is compared to the difference in their degrees of freedom. A lack of statistical 94 significance of these changes in G2 would mean that the reduced model offers parsimony without impairing the correspondence between observed and expected cell means. (b) Compare the G2 and the number of degrees of freedom for the reduced model itself. These two values should be almost identical. The best-fitted model implies that the expected cell frequencies are not significantly different than the observed cell frequencies. We proceeded to eliminate terms from the saturated model and invoke these two criteria until the most parsimonious model was found. Let us illustrate this two—step procedure by showing how we chose Model 9 (Table 4.6) as the best-fitting reduced model from which to test Hypothesis 5. HQ§§l_11 The first reduced model, Model 1, includes contraceptive use(1=dependent variable), deficit fertility (2=intervening variable), life-cycle status (3=control variable), place of residence (4=independent variable) and literacy status (5=intervening variable), plus all six possible three-way interaction terms linking the dependent variable with a pair of the other three kinds of variables, together with all possible two-way interactional relatives. With a statistically nonsignificant likelihood chi-square ratio (G2 95 = 10.99, D.F = 9 and P > 0.05: see Table 4.6), Model 1 implied a minimal difference between the observed and the expected cell means. IHence, Model 1 formed the basis of comparison for the rest of our reduced.models~ The several models, and their comparison with Model 1 is shown in Table 4.7, and the procedures are illustrated below. HQQ§1_Z} In W we kept all 2-way interaction terms and omitted one 3-way interaction term [123] assuming that this interaction may not be very meaningful. With 11 degrees of freedom, the G2 of 13.76 for this model indicated a very good fit to the data (Table 4.6). The increase in G2 was only 2.77 above that for Model 1 (see Table 4.7); and at a gain of only 2 degrees of freedom, this change in G? fell below the critical value of 5.99 needed to reject the hypothesis that the change in G2 followed a chi-square distribution. Thus, both.criteria indicated.that.the model was good; therefore, we could not reject Model 2. 825121;: Model 3 excluded 3-way interaction term [124] from Model 1. The model did not appear to represent the best fit to the data (Table 4.6): and the comparison of Model 1 and Model 3 was done by computing the difference between the values of G2 (89.66) and. degrees of freedom (=1), a disparity which was 96 Table 4.6: Goodness-of Fit Statistics (G?) for Various Log-Linear Models Applied to Test Hypothesis 5 O Models G2 D.F P Value MQQ§1_1 10.99 9 0.276 [1][12][13][14][15] [123][124][125] [134][135] [145] Egdel_z [11(12][13][14][15] [124][125] [134][135] [145] n2§§l_l [1][12][13][14][15] [123][125] [134][135] [145] nodel_4 [1][12][13][14][15] [123[124] [134][135] [145] n2d21_§ [11(12][13][14][15] [123][124][125] [135] [145] MQQ§l_§ [1][12][13][14][15] [123][124][125] [134] [145] MQ§§1_Z [1][12][13J[14][15] [123][124][125] [134][135] 13.76 20. 65 14.02 13.09 18.77 11.52 11 10 10 11 11 10 .247 .024* .172 .287 .065* .319 Continued. 97 Continued Table 4.6: nggg1_§ 35.89 18 0.007* [11(12][13][14][15] uggg1_2 20.12 15 0.167 [1][12][13][14][15] [124][135] ‘ The numbers in parentheses, for each model, denote the variable number included in each of the model as stated in the following list of variables: * p < 0.05 (statistically significant difference). Li§t_gf_yariable§: 1 . Contraceptive use: yes = 1, no = 2. 2 . Deficit fertility: yes = 0, no = 1. 3 = 3. . Residence; urban = 1, rural = 2. J ] ]. Life-cycle status: early 2 1, middle.= 2, and late 1 ]. Literacy status: yes = 1, no = 2. Study sample size = 5713 (see footnote to Table 4.4). 98 highly significant statistically and represented a departure from a chi-square distribution (9.66 is much larger than 3.84): (see Table 4.7). Thus, we rejected Model 3. 11952214: Model 4 omits the term [125] from Model 1. Comparing the differences between the values of G2 (14.02 - 10.99 = 3.03) and the degrees of freedom (10 - 9 =1), indicated that the difference of 3.03 was less than the critical value of chi- square distribution (3.84) . In addition, the G“2 for Model 4 (14.02) was not much different from its number of degrees of freedom (10) (see Table 4.6). These two criteria led us to accept Model 4 as a good-fitting and more parsimonious model in relation to Model 1. 119.0214: In this model we excluded the term [134] from Model 1. We found the difference between G2 of Model 5 and Model 1 as (13.09 -10.99 = 2.10) and the degrees of freedom (11 - 9 - 2). This conformed to a chi-square distribution since 2.10 was less than the critical value of 5.99 (Table 4.7). In addition, the 62 of Model 5 (13.09) was close to the number of its degrees of freedom (11) , an indication that the model itself ‘was not statistically significant" The lack. of statistical significance of G2 (p > 0.05: Table 4.6) means that this reduced model did not impair the correspondence 99 Table 4.7: Comparison of2 Difference in Goodness-of-Fit Statistics (62) for Various W LL£§§I_MQQ§l§ with the Large_ngdel Applied to Test Hypothesis 5. Models GL2 D. FL Critical Value LQIQ£_H2§213 W 10 . 99 9 16 . 9 19 [1][12][13][14][15] [123][124][125] [134][135] [145] W: (BL2 - Gaz D.FL - D.F" Critical Value HQQ§1_Z 2.77 2 5.99 [1][12][13][14][15] [124][125] [134][135] [145] uggg1_; 9.66 1 3.84* [1][12][13][14][15] [123][125] [134][135] [145] nodel_1 [1][12][13][14][15] [123[124] [1341(135] [145] HQQ§l_§ [11(12][13][14J[15] [123][124][125] [135] [145] mm 7.79 2 5.99* [1][12][13][14][15] [123][124][125] [134] [145] Continued. 100 Continued Table 4.7: gggg1_1 0.53 1 3.84 [1][12][13][14][15] [123][124][125] [134][135] MQQ§1_§ 22.35 9 16.92* [1][12][13][14][15] nggg1_2 9.13 6 12.59 [1][12][13][14][15] [124][135] Note: 'P < 0.05 (statistically significant). Li§t_gf_yariablesz [1]. Contraceptive use: yes = 1, no = 2. [2]. Deficit fertility: yes = 0, no = 1. [3]. Life-cycle status: early = 1, middle = 2, late = 3. [4]. Residence: urban = 1, rural = 2. [5]. Literacy status: yes a 1, no 2 2. Study sample size = 5713 (see footnote to Table 4.4). 101 between observed and expected cell frequencies. Thus, we can not reject Model 5. W: We excluded the term [135] from Model 1 to get Model 6. When Model 6 was compared with Model 1, it showed a statistically significant increase in G2 (7.79), since the degrees of freedom rose from 9 to 11 (Table 4.7). Hence, the exclusion of the term [135] impaired the fit between observed and expected cell means and had to be put back in the model. In other words, Model 1 is to be preferred over Model 6. Thus, we reject the Model 6. HQQ§l_Z= This model excluded the interaction term [145], and the difference between the Model 7 and Model 1 was found not to be statistically significant. This was discovered on the basis of the comparison (Table 4.7), of G2 of Model 7 and Model 1 (11.52 - 10.99 = 0.53), which at one degree of freedom (10 - 9 = 1) was not statistically significant (p > 0.05). The Model 7 itself (see Table 4.6) appeared to be well fitted {G2 (11.52) and degrees of freedom (10) are very close, and p (31.9%) > 5%) and we accepted this model. 102 W: The difference in G2 of Model 8 and Model 1 is 22.35 (35.89 - 10.99 - 22.35), which at 9 degrees of freedom (18 - 9 = 9) was highly statistically significant (Table 4.7). By exclusion of all 3-way interaction terms, and retaining all of the 2-way interaction terms, [12] [13] [14] [15] this model produced a highly significant 62, as well (Table 4.6) . Thus, we rejected this model. This model indicated that some of the three-way interaction terms needed to be reincorporated. We decided that the essential three-way interactions might be captured by the terms [124] and [135], since those had been the excluded terms from the ill-fitting Model 3 and Model 6. W Model 9 included four 2-way interaction 'terms [12][13][14][15] and two 3-way interaction terms [124] and [135] , and the dependent variable [1] . The difference between G? of Model 9 and Model 1 (20.12 - 10.99 = 9.13) and between the degrees of freedom (15 - 9 = 6) was statistically not significant (p > 0.05) (see Table 4.7). With a statistically nonsignificant G2 (=20.12 at 15 degrees of freedom: Table 4.6), Model 9 appeared to be our best-fitted, most parsimonious model, and it became the basis for testing Hypothesis 5 (Table 4.8). 103 Logit Regression Analysis of Hypothesis 5: WWW: Squaring the parameter estimate, beta (B) and taking the antilogarithm of the results yields the expected odds-ratio, which represents the effect of being in the first (rather than the second) category of independent variable on the odds of being in the first category of the dependent variable. If the odds ratio equals one, it means that being in any category of the independent variable does not tip the odds toward/ against being in any category of the dependent variable. This situation is shown by a Z-score not statistically significant at the five percent level. By the same token, an odds ratio that is much larger than unity means that being in the first category of a predictor variable significantly boosts the risk of being in the first category of the dependent variable. As such, we see that women who could read a newspaper and write a simple letter had higher odds (1.273) of current contraception than did the illiterate (Z value = 2.669, p < 0.05) (see Table 4.8). Similarly, a woman whose number of surviving children was below what she considered appropriate for a family like hers had much lower odds (0.699) of current contraception than did a woman without such a deficit in fertility. Contrary to Hypothesis 5, the addition of these two predictor variables to the model stated in Hypothesis 4 did not cause the 104 Table 4.8: Logit Regression of Current Contraceptive Use (1=Yes; 2=No) upon Wife's Residence, Literacy and Control Variables. Variables Coefficients Antilogarithm Z-Value Intercept -1.999 0.135 -20.380* MAIH_EEEE§I§= -Deficit Fertility (0=Yes: 1=No) -0.358 0.699 -6.440* -Residence (1=Urban; 0.520 1.682 9.755* 2=Rural) -Literacy (1=Yes: 2=No) 0.241 1.273 2.669* -Life-cycle (1=Early: 2=Middle -0.455 0.634 -3.021* 3=Late) -0.176 0.839 -1.442 IEIEBAQIIQH_EEEEQI= -Deficit Fertility By Residence 0.176 1.192 3.320* -Life-cyc1e By Literacy (1=Early: 2=Middle -0.326 0.722 -2.175* 3=Late) 0.164 1.178 1.349 Note: Likelihood Ratio Chi-square = 20.12 D. F. = 15, P value = .167 ' p < 0.05 (statistically significant). 105 relationship between place of residence and current use of contraception to vanish. Indeed, the odds that a woman was currently contracepting were still 68.2% higher (odds = 1.682) if she lived in an urban than a rural place. We investigated the interactive relationship between deficit fertility and residence on current contraceptive use by nonpregnant wives (Z-value = 3.320 in Table 4.8). The odds of current contraception by an urban woman with a deficit in actual number of living children were 7.0% / 93.0% - 0.075 (Table 4.9). The odds of current contraception.by‘urban‘woman with no deficit fertility were 25.7% / 74.3% = 0.346. The odds of current contraceptive use for a rural woman were 0.021 if she had a deficit in number of living children and 0.104 if she had no deficit. Thus, we see that having deficit fertility greatly reduced the odds that any woman--urban or rural--would be using contraception. Notice also that urban women--with or without deficit fertility--were more likely than rural women to contracept currently. In fact, the odds of current contraception were about even for urban.women with a deficit in fertility (0.075) and for rural women without such a deficit (0.104). This equality produced the statistical interaction between deficit fertility, residence and contraceptive use. .A jpossible explanation of this equality may be the higher costs (market and psychological) of contraception which discourage rural women from adopting 106 Table 4.9: Percent Contraceptive Use Among Wives by Deficit Fertility Status and by Place of Residence in Pakistan, PCPS 1984. Place of Residence/ Deficit Fertility Contraceptive Use Urban Rural Yes No Yes No Deficit Deficit Deficit Deficit % % % % Yes=l 7.0 25.7 2.1 9.4 No=2 93.0 74.3 97.9 90.6 Total 100% 100% 100% 100% (N=1050) (N=1226) (N=1937) (N=1500) Note: Chi-square: Urban - 138.767, D.F = 1, P < 5 %. Rural - 89.421, D.F = 1, p < 5 %. Study sample size = 5713 (see footnote to‘Table 4.4). 107 family' planning ‘when. they' no longer' have a deficit in fertility. The following is the interpretation of the interaction between contraceptive use, literacy and life-cycle status (Table 4.10). Of the 53 women in the early phase of the family life cycle, only two (both illiterate) were currently contracepting. Thus, contraception does not become a means to an acceptable end until women enter the middle phase. The odds of currently contracepting by literate women who are at their middle or late stage of family-building life-cycle, were 0.056 (= 5.3% / 94.7%) and 0.212 (= 17.5% / 82.5%), respectively. The odds of current contraception by W women at their middle or late family-building life-cycle stages were 0.011 (- 1.1% / 98.9%) and 0.105 (- 9.5% / 90.5%), respectively, which were much lower than those for literate women. These relationships show that the propensity to contracept intensifies across the family life-cycle, that literacy strengthens this relationship, and that.the impact.of literacy is not constant for all life-cycle stages. The greatest similarity between literate and illiterate women in odds of current contraception is at the early stage, where the odds of both groups of women are nil: and the greatest dissimilarity is at the middle stage, where the odds of current contraception are five times greater for literate 108 Table 4.10: Contraceptive use Among Currently Non- Pregnant Women by Family-Building Stage and Literacy Status in Pakistan, 1984-85. Literate Illiterate Contraceptive Use Early Middle Late Early Middle Late % % % % % % Yes=1 0.0 5.3 17.5 2.0 1.1 9.5 No=2 100.0 94.7 82.5 98.0 98.9 90.5 Total 100% 100% 100% 100% 100% 100% (N=2) (N=l9) (N=382) (N=51) (N=364) (N=5318) Note: Chi-square: Literate: Chi-square = 2.36, D.F = 2, P > 0.05. Illiterate: Chi-square = 32.87, D.F = 2, P < 0.05. Study sample size = 6136 (see footnote to Table 4.5). 109 than for illiterate women. Put differently, the attainment of literacy increases not only the odds that women will adopt contraception but also that they will adopt it before reaching age 30. We speculate that literacy reduces the market costs of contraception by shortening the search for it and the psychological costs by legitimizing the innovative behavior among the literates. In other words, if, as I have posited, literate women are more open to Western concepts of personal autonomy, then one manifestation of this would be contraceptive use. In the next chapter, we explore what market and psychological costs might block contraception for rural and illiterate women even when their number of surviving children is no longer deficient. CHAPTER 5 COST OF FERTILITY REGULATION: A.RURALHURBAN COMPARISON Easterlin (1975) argued that lower costs of fertility regulation lead to a Ihigher’ effective demand for contraception, ggtgri§_paribu_. The usual costs associated with the use of the family planning program are monetary, physical and psychological. The government of Pakistan has sought to minimize the monetary costs of fertility regulation by subsidizing the price of clinic fees and of the contraceptive supplies doled out by the clinics. Advertisements promoting the family planning programs through messages printed on billboards and announced on radios have been aimed at reducing the search costs (monetary) for both literate and illiterate Pakistanis. However, the relevance of such a cost is interpreted differently by literates and illiterates. Moreover, the ratio of physicians to people is very low: and.physicians cluster in the urban.areas ((in 1975, there was only one physician per 3850 people, (Mamlouk, 1982: 30)). Consequently, trips to physician-attended clinics are time-consuming and exact a hidden price in time forfeited for work, especially by rural people who must walk. The physical costs include not only the fatigue of the trip but also the side effects which sometimes accompany birth control: such as, the morning nausea that often attends the adoption of the birth control pill and the breakthrough 110 111 bleeding or physical discomfort that sometimes follows the insertion of an intrauterine device (IUD). Both side effects could be interpreted as more troublesome by rural wives, who might see a need for repeated visits to the distant clinics until the side effects disappear. In a country like Pakistan, the last-mentioned side effects would subject the woman to a vexatious psychological cost: the inconvenience/embarrassment of having the menstrual taboo imposed on her for more than once in a month. A second psychological debt would be the stress of disobeying the wishes of a husband or a mother-in-law who disapproves of family planning. A third would be the hardship of keeping the disobedience secret: a travelling companion to the clinic must be sought, since most Pakistani women are not "permitted" to travel alone outside the home. In Chapter 2, I argued that these several psychological costs would be high all over Pakistan, but given the scarcity of clinics, especially heavy in the rural areas. In Chapter 4, we saw that non-pregnant rural wives who had enough or too many children were no more likely to be contracepting than were their urban counterparts who had too few. I surmised that a reason was the greater cost of fertility regulation faced by the rural wives. Below I explore the empirical evidence for /against this idea. 112 Current Contraceptors: A Profile When the market, physical, or psychological costs of trying a family-planning method are heavy, it is economically rational to adopt an irreversible, one-time-only method. Thus, the modal method in both rural and urban Pakistan was the surgical sterilization of*women (30.6% for rural wives and 28.7% for urban wives: Table 5.1). In urban Pakistan, one other method rivalled the popularity of female sterilization: condom (29.7%: Table 5.1). The fact that contraception by males was so much higher in urban than in rural Pakistan suggests that the greater literacy of‘women in the urban areas motivated male cooperation or that the several costs of male contraception were lower in urban areas. While pills were about equally popular among rural and urban women (19.9% and 16.8%, respectively: Table 5.1), it is noteworthy that rural women were more than twice as likely as urban women to use the IUD and almost twice as likely to use an injection. The latter two methods do not require frequent visits to the source of family planning and thus are more economical for women who have difficulty gaining access to contraception. The method of contraception is related to its source (Table 5.2) . In both rural and urban areas, but especially in rural areas, government hospitals were the main source of the contraception currently in use, because female sterilization must be performed by a doctor trained in Western medicine. 113 Table 5.1: Percent of Wives gggrengly Helge Specific Methods of Family Planning by Residence, PCPS 1984-85. Residence Methods Rural Urban % % EEMALE_MEIHQD§ Oral Pill 19.9 16.8 Vaginal 1.1 1.5 Injection 8.6 4.8 IUDs 14.0 6.9 F. Sterilization 30.6 28.7 MALE_MEIHQD§ Condom 9.1 29.7 M. Sterilization 0.5 0.3 QIEEB_HEIHQD§ Rhythm 0.0 2.0 Withdrawal 6.5 7.4 Abstinence and else 1.6 0.8 Total Using: 100% 100% (N=186) (N=394) 114 Table 5.2: Source of Contraception by Residence of Currently Contracepting, Non-Pregnant Wives: PCPS 1984-85. Residence Source Rural Urban % % Government Hospital 54.3 40.4 MCH Center 1.6 1.5 Rural H. Center 0.5 0.0 Basic H. Center 2.2 0.0 Family Welfare Center 11.8 7.1 Drug/Grocery Stores 8.6 31.7 Community/FP Worker 3.2 4.1 Others/Hakim/doctor* 1.6 3.9 Not Sure/No response 16.1 11.4 Total Using 100% 100% (N=l86) (N=394) * A ”hakim" is a traditional cure provider, basing the treatment of diseases by the use of medicines made out of traditional herbs and shrubs. Hakim has been very popular in Indo-Pakistan sub-continent for long time. 115 The popularity of condoms among urban husbands probably explains why drug stores or grocery stores were the second most important supplier of contraception to urbanites. About 96.8% of rural and 95% of urban women who were contracepting currently did not prefer some other method or else were unsure about their preference (Table 5.3). Of those stating a preference for another method, most named female sterilization (2.2% of rural female contraceptors and 2.0% of urban ones). An obvious reason for not using it at present was an openness to having more children. Since our interest was to investigate the problem of availability of family planning methods by residence, we examined whether a contraceptor had ever stopped the current method for more than a month (Table 5.4) and her reasons for stopping its use (Table 5.5). Table 5.4 revealed that among current contraceptors, 10.8% of rural and 13.5% of urban residents had stopped using their current method for more than one month. The major reasons among rural respondents were that they had stopped using upon doctor's recommendation (30%) or forgot to use (25%). Among urban respondents, the major reasons were: forgot to use (30.2%) and health (24.5%). The mention of health concerns or medical advice stresses the deterrent role of side effects of contraception in. the continuation of a particular method. However, 15% of rural 116 Table 5.3: Percent of Wives gnrrenrly_neing, fieeeng Preferred Methods by Residence, PCPS 1984- 85. Residence Preferred Method Rural Urban % % EEMALE_MEIHQD§ Oral Pill 0.0 0.3 Vaginal 0.0 0.0 Injection 0.5 0.5 IUDS 0.5 1.3 F. Sterilization 2.2 2.0 MALE_MEIHQD§ Condom 0.0 0.5 M. Sterilization 0.0 0.5 HQI_§HBE 0-0 0-3 HQ_B£§£QH§E 95-8 94-7 Total Using: 100% 100% (N=186) (N=394) 117 Table 5.4: Percent of Non-Pregnant Wives W, Have Stopped Family Planning for More Than a Month by Residence, PCPS 1984-85. Residence Stopping Status Rural Urban % % Yes 10.8 13.5 No 51.1 54.3 Not Sure 5.9 2.3 No Response 32.3 29.9 Total Using 100% 100% (N=186) (N=394) 118 Table 5.5: Percent of Non-Pregnant Wives 93mm, B§§§QQ§_IQI;§§2221flQ Family Planning for More Than a Month by Residence, PCPS 1984-85. Residence Reasons Rural Urban % % Health 0.0 24.5 Doctor Recommendation 30.0 7.5 Husband Dislikes 5.0 11.3 Out of Supplies 15.0 9.4 Used Another Method 10.0 3.8 Forgot to Use 25.0 30.2 Other 15.0 3.8 Not Sure 0.0 0.0 No Response 0.0 9.4 Total Using 100% 100% (N=20) (N=53) 119 respondents reported. that they ‘were out of supplies as compared to 9.4% of urban respondents. 0f the three respondents who reported they were out of supply, one rural woman and one urban woman had been using the pill and the other urbanite reported use of the condom (Table 5.6). Thus, a shortage of contraceptive supplies was found for both urban as well as rural residents. For current users of family planning, the reeeege regarding the difficulty in obtaining the method or reaching the source of family planning were explored in Tables 5.7, 5.8 and 5.9. The question "who usually obtains the method you are currently using?" was asked only from the current users of oral pills, vaginal methods and condoms. The response to this question was recorded as: wife, husband or someone else. In Table 5.7, we noted that quite surprisingly, the rural-urban difference in who obtains the method currently being used was almost negligible, except for the vaginal methods. For 2 of the rural women who reported use, one of them got it by herself and for the other one someone else (not husband) obtained it: for the 6 urban users, 83.3% obtained the method themselves and for the rest of them (16.7%) husbands got it. For the pill, the husband was about as likely to purchase it as the wife, regardless of residence. This fact symbolizes not only the difficulty that women face in.going out.in public Table 5.6: 120 Percent of Non-Pregnant.Wives Qgrrenrlyrueing Family Planning, Had Stopped Use for More Than a Month by Wed by Residence. pops 1984-85. Residence Methods Rural Urban % % Oral pills 100 50 Condom 0.0 50 Total 100% 100% (N=1) (N=2) 121 Table 5.7: Percent of Non-Pregnant Wives, W a t od tS't I! c> ' , PCPS 1984-85. Residence Rural Urban Current Method Pill Vaginal Condom Pill Vaginal Condom % % % % % % Wife 48.6 50.0 5.9 50.0 83.3 5.1 Husband 48.6 0.0 94.1 47.0 16.7 94.0 Else 2.7 50.0 0.0 3.0 0.0 0.9 Total 100% 100% 100% 100% 100% 100% (N=37) (N=2) (N=17) (N=66) (N=6) (N=ll7) 122 but also the importance of the husband's approval in her use of the pill. The husband was almost the exclusive purchaser of the condom whether his residence was rural (94.1%) or urban (94.0%) . Thus, it demonstrates that it is deemed more appropriate for a husband to purchase the male-method than it is for a wife. A woman will feel very embarrassed while purchasing a male-method, particularly from a grocery store (which is possible to do only in an urban area): she might even be suspected of being a very 'fast' woman who would commit adultery. The W. represented respondent's own difficulty in reaching the source of family planning, has been 0263mm here to reflect the cost (access and availability) of contraceptive use. The question, "Is it difficult or easy for you to get there (to your source of family planning)?" was asked from 580 rural-urban current users of any contraceptive methods offered by the national family planning program. The responses were recorded as: difficult/sometimes difficult: easy: and not sure/no response. The analysis of these responses from Table 5.8, revealed that the percent reporting difficulty in reaching the source of their family planning method in rural areas (30.6%) was much greater than that of urban areas (7.4%). We suspected that this difficulty must involve the problems related to the availability of transportation, among other things. 123 Table 5.8: Percent of Non-Pregnant Current Users by Difficulty in Reaching the Source of Family Planning by Residence, in PCPS 1984- 85. Residence Difficulty Status Rural Urban % % Difficult/Sometimes 30.6 7.4 Easy 48.9 76.4 Not sure 4.3 4.3 No response 16.1 11.9 Total 100% 100% (N=135) (N=394) 124 What was the mode of travel used in reaching the source of family planning method by the current users? Table 5.9 shows that the 74.1% of rural respondents reported the use of some transportation, which.is:much.greater'than.as compared to the 61.5% of urban respondents. The percent of those who walked to the source of the family planning was lower in rural areas (25.9%) than in urban areas (38.5%), a fact which must reflect a relatively less easy and convenient access to the sources of family planning for rural residents than urban residents. The reasons given by the respondents for their difficulty in reaching the source of family planning were addressed in Table 5.10. About 54.4% of rural respondents reported bad roads as their number one problem in reaching the source of family planning as compared to the 13.8% of urban residents. The percent reporting transportation problems was also much higher for ruralites (21.1%) than urbanites (10.3%). Quite surprisingly, the percent reporting that the travel time was too long in reaching the source of family planning was higher in urban areas (62.1%) than rural areas (19.3%). Also, the percentage of those who reported difficulty in reaching the source of their family planning method due the trip being too expensive was higher in urban areas (6.9%) than rural areas (1.8%). The likely reasons for greater time- and financial- cost among urban contraceptors, in spite of having relatively 125 Table 5.9: Percent of Non-Pregnant Current Users by Mode of Transport for Reaching the Source of Family Planning by Residence, in PCPS 1984— 85. Residence Transportation Mode Rural Urban % % Walk 25.9 38.5 Use transport 74.1 61.5 Total 100% 100% (N=116) (N=317) Missing cases: 116 126 Table 5.10: Percent of Non-Pregnant Current Users by Reasons for Difficulty in Reaching the Source of Family Planning by Residence, in PCPS 1984-85. Residence Reasons Rural Urban % % Bad Roads 54.4 13.8 No Transport 21.1 10.3 Travel time too long 19.3 62.1 Too expensive 1.8 6.9 No one to go with 1.8 0.0 Other 0.0 0.0 Not sure 0.0 0.0 No Response 1.8 3.4 Total 100% 100% (N=57) (N=29) 127 greater accessibility to transportation and roads, were the mental and physical discomforts of travelling in over-crowded buses and.waiting too long at bus stops, clinics, hospitals or drug stores. Moreover, if the respondent is travelling with a companion (friend or a relative) she is obliged to pay the cost of trip for the companion, as well. Additionally, if they travel by a taxi or'a scooter-rickshaw, the fare‘would.be much higher than the public bus fare. Thus, among urban contraceptors, the greater difficulty seems to be associated with accessibility costs (roads and transportation are available but the nature of problems is different). However, these findings clearly outline the greater nonavailability costs, such as, inconvenience associated with a lack of roads and transportation which must entail higher physical, financial and psychological costs for ruralites than urbanites. Current Non-Contraceptors: A Profile I was interested in exploring what future contraceptive method the currently noncontracepting eligible couples might use. Where would they seek it, and why were they not currently using it? Was it the cost of fertility regulation (nonavailability) or some other constraint against contraceptive adoption? We explored these phenomena in Tables 5.11 and 5.12, respectively. 128 Among the noncontraceptors who responded, oral pills, injections, female sterilizations and IUDs were the main future preferred methods for both rural and urban respondents (Table 5.11) . However, the reported percents for rural women were higher than for urban women for oral pills (27.2%), and for female sterilization (18.5%): and were even for injections (19.3%) . These percentages for urban women were 21.6% for oral pills, 17.9% for female sterilization, and 19.5% for injections. The responses for IUD, condoms, rhythm, withdrawal were higher for urban women (see Table 5.11) . However, it is remarkable to note that a higher percentage of rural (25.8%) than urban (20.7%) noncontraceptors indicated an intention of using any appropriate method. This did not mean that the appropriate method would be any other than the methods provided by the program but possibly that the respondent could not name any specific method or a method provided by the family planning program. Thus, it may reflect that lack of exposure, information and availability kept the rural residents more ignorant than urban residents. This bears an important implication for policy making in rural areas . W. if they decide to use contraception in future, the preferred primary source for obtaining a family planning method was also a government hospital (Table 5.12): 67.0% of rural and 63.0% of urban women 129 Table 5. 11: Percent of Women Currently Non-Contracepting Intending to Use in Future By Preferred Method By Residence, PCPS 1984-85. Residence Method Rural Urban % % EBEEEBBED_EEMALE_HEIHQD§ Oral Pill 27.2 21.8 Vaginal 0.4 1.2 Injection 19.3 19.5 IUDs 4.8 7.6 F. Sterilization 18.5 17.9 EBEEEBBED_MALE_EEIEQQ§ Condom 2.4 7.4 M. Sterilization 0.0 0.3 EBEEEBBED_QIHEB_MEIHQD§ Rhythm 0.4 2.0 Withdrawal 0.3 0.8 Any Appropriate 25.8 20.7 Other 0.8 1.2 Total Using 100% 100% (N=798) (N=605) 130 Table 5.12: Percent of Women Currently Non-Contracepting Indicating to Use in Future by Preferred Source By Residence, PCPS 1984-85. Preferred Residence Source if Use in Rural Urban Future % % Government Hospital 67.0 63.0 MCH Center 1.8 3.0 Rural H. Center 4.1 0.0 Basic H. Center 6.6 0.0 Family Welfare Center 6.6 12.0 Drug/Grocery Stores 6.6 9.7 Community/FF Worker 4.1 7.1 Others/Hakim/doctor 3.1 5.1 Total Not Using 100% 100% (N=678) (N=565) 131 identified this source as their number one preference. Family welfare centers, basic health centers and drug/grocery stores appeared to be the second preferred source for obtaining their future method among the rest of the rural noncontraceptors (6.6% each). Rural health centers, and community and family planning workers were identified as the next preferred source by 4.1% of rural respondents. Of urban women 12% identified family welfare centers and 9.7% identified drug/grocery stores as their second preferred source to obtain their future method: community and family planning workers were pointed out only by 7.1% as the next source. Other sources, including hakims and private doctors would be preferred by 3.1% of rural and 5.1% of urban women. These results suggest that if government hospitals or family welfare centers were more numerous, then a larger proportion of noncontraceptors could be recruited. Table 5.13 investigated the first and second reasons for not using any family planning method among noncontracepting respondents by residence. The desire to have more children was the number one reason for nonuse among rural (20.8%) as well as among urban (20.4%) women. To our surprise, the percent of women not using contraception for religious reasons was lower for rural women (11.1%) than urban women (14.1%). The likely reason for such an attitude among rural women could 132 Table 5.13: Percent of Non-Pregnant Wives Qurren§1y_fler Heine by Beeeene for Non-Use of Family Planning by Residence, PCPS 1984-85. Residence Reasons Rural Urban for % % Non-Use First- Second-4 First- Second— Reason Reason Reason Reason Want more children 20.8 1.6 20.4 1.8 Religious 11.1 4.1 14.1 3.8 Fear side effects 2.7 1.2 4.9 1.1 Past side effects 0.4 0.2 1.0 0.9 Breastfeeding 4.5 1.4 5.4 0.9 Nonavailability 2.4 0.7 1.8 1.3 No knowledge 0.3 0.2 0.3 0.2 Menopausal 1.9 0.1 2.5 0.0 Perceived sterile 0.7 0.5 0.3 0.3 Post partum abste. 0.0 0.0 0.5 0.2 Husb./fam. oppose 0.0 0.0 2.6 1.6 Husband Absent 0.0 1.7 1.4 0.6 Too lazy to do 0.0 0.3 1.0 0.0 Natural spacing 0.9 0.6 5.2 0.9 Pregnant 3.5 1.5 1.5 0.0 Other 0.6 0.4 1.1 0.4 Not sure 0.6 0.0 0.7 0.0 No response 46.6 85.4 36.8 86.6 Total Not Using 100% 100% 100% 100% (N=3518) (N=3518) (N=2038) (N=2038) 133 be, firstly, due to a lack of wider religious preaching through mass media which are more common in urban areas: secondly, the impact of the government's Islamization propaganda was much greater among urban than rural people. The nonuse of family planning for fear of having side effects was lower for rural women (2.7%) than urban women (4.9%), as well. The nonuse because of breastfeeding a child was also lower among rural women (4.5%) than urban women (5.4%) . However, the number 6 reason for nonuse appeared to be the W of family planning which was higher among rural women (2.4%) than urban women (1.8%). The most common second mentioned response for nonuse was associated with religious reasons (4.1% in rural and 3.8% in urban women). However, the Quraan does not expressly forbid family planning on moral grounds. Thus, a much greater proportion of couples would initiate contraception if they could be persuaded that it is not sacrilegious. CHAPTER 6 SUMMARY AND CONCLUSION In this dissertation research, the fundamental objective was to explain the lower level of current use of contraception by rural than urban Pakistanis. Since the professed goal of the Government of Pakistan is to reduce its extremely rapid rate of population growth through a national family planning program, the goal will not be reached and the funds invested to this end will be wasted if the 70% of the population who live in rural areas do not use contraception obtained from this program or private sources. Rurality and urbanity must be understood in the social sense rather than in a geographical sense. In Pakistan there are many differences in social and economic institutions between rural and urban spheres which structure different life-styles and behavior. We found that in rural areas of Pakistan most peoples' economic activities and livelihood were attached to the agricultural occupations and in urban areas to non-agricultural occupations” The traditional farming system is labor-intensive and does not require much formal schooling and training for the majority'of'people to be able to function as farmers in the rural areas. Modern agricultural mechanization has displaced some rural laborers: those few with large farms and large machinery are men who have been approached by male agricultural extension agents for training. 134 135 Women contribute to farming activities of Pakistan to a great extent, though the tasks performed by them are considered to be subsidiary to the major tasks which are performed and controlled by men: thus women's literacy is not considered important in order to function effectively in rural society. Nonetheless, the underestimation of women's work due to deficient definitions adopted in Pakistani censuses and surveys (Abbasi, 1987: Shah and Shah, 1980) remains an issue to be addressed. The techniques necessary to tending to ”women's tasks" inside and outside the home are passed on to them by word or deed from the elder generation. The school enrollment rates for rural females are lower than for urban females, and so is the governmental allocation of educational budgets (Sarmad, ML, 1988: and Tsang and Zaki, 1989). Thus, literacy levels are very low in the country as a whole, and they are even lower in rural than urban areas, particularly for women. Since women's economic roles are less valued than men's, sons are preferred to daughters: thus, a large number of children is desired. Because children/sons provide security to parents in their old age, social security for mothers in their widowhood, and insurance against divorce, wives are benefited as much or more than husbands from high fertility. Children enhance familial status and social power, and are 136 sources of financial support upon assuming their economic roles (Das, 1984: Farooqui, 1984: Mason, 1985: Miller, 1983). Whether men and women desire a certain number of children (or sons) for different reasons is not known in case of Pakistan. Thus, further research needs to be done to understand the reasons for differing fertility goals for husbands and wives. However, we noted earlier that in other developing countries, even when men and women differed in their reasons for wanting sons, they did not differ substantially in their total number of desired children (Arnold, 1975: Mason and Taj, 1987). The economic and social value of sons keeps many couples from using contraception until they have at least one: and in societies with high rates of child mortality, two sons may seem necessary to insure the survival of at least one (Mason, 1985: and Ahmed, 1981) . Rao and Beaujot (1986) found a positive relationship between the mean number of children ever born and the number of child deaths, both in rural and urban areas of Pakistan: it implied a tendency to replace lost children because it was more pronounced when the lost children were sons than daughters. Son mortality may very well discourage contraceptive adoption among Pakistani women. The authors noted further "that the mean number of subsequent births (to child deaths) were always higher among urban women than for rural women" (p.298). I would suspect it to be related to the shorter duration of postpartum amenorrhea among 137 urban than rural women resulting from the more frequent interruption of breastfeeding due to young-child mortality. Thus, it is obvious that further research is needed to study the possible effect of child mortality on contraceptive adoption or continuation in Pakistan. Because the child mortality rate is higher in rural than in urban Pakistan and because the demand for child labor is particularly high in labor-intensive agriculture, I reasoned that rural women in Pakistan would report a larger number of children as appropriate for a family like theirs. Rural fertility is generally higher than urban fertility, and so is the wanted number of children. As Johnson (1977) noted, even in the most advanced societies of the world, such as, the United States of America, wives of husbands in farm occupations have higher fertility than wives of husbands associated with nonagricultural occupations“ However, unlike many societies of the world, the rural Pakistani society has demonstrated, on the one hand, a lower actual fertility (measured in terms of number of children ever born and number of children currently surviving due to the longer breastfeeding norm) than for urban Pakistanis, yet the number of children considered appropriate by ruralites is greater than for urbanites. This would imply that rural women are farther from reaching their targeted number of living children and would explain why there is a lower level of contraceptive 138 adoption among rural people (Alam and Dinesen, 1984: Shah er e1. 1986: Khan and Sirageldin, 1977: 1979: Soomro and Ali, 1984). A very rapid rate of population growth (3% per year) and a very low usage of the government-supported family planning program has posed a great dilemma for policy makers of Pakistan, who deem the population explosion to be inimical to economic development. Thus, it is a considerably important matter to understand how Pakistani wives form their conceptions about appropriate numbers of births and how social factors create more pronatalist fertility orientations among rural wives. Previous research in Pakistan has demonstrated that literacy is an important factor in explaining the differentials in fertility and contraceptive adoption in the society (Syed, 1978: Shah, 1986: Sathar and Mason, 1989: Sathar er_el., 1988: Cleland and Hobcraft, 1985). I argued that female literacy affects fertility mainly in six ways: a) in developing countries, literacy at lower levels of schooling can have a pronatalist effect by lowering fetal and infant mortality through increased maternal health care and post- and pre- natal care: the antinatalistic effect is realized: b) through increasing the awareness and knowledge of modern family planning as a result of exposure to media campaign 139 (more often in urban areas): 0) through raising the female age at marriage, which shortens the family-building life-cycle span (in societies where marriage marks the initiation of reproduction period for a woman): d) through increasing the chances of gainful employment, which creates conflicts between work and childcare: e) through raising the status of women so as to make them more autonomous and less dependent on husbands and adult sons for protection, advice, and economic support: and, f) through breaking the norms of traditional wealth-flows from children-to-parents to one of parents-to-children and nuclearizing and ”Westernizing" the families. With all these changes, literate women were hypothesized to desire to have fewer but better-quality (better'health and education entails costs) children. Thus, it was hypothesized that literate women should desire fewer children and should have fewer children to die such that would more quickly attain the number of living children they deem appropriate for a family like theirs. Following Easterlin (1975) I reasoned that women who have reached or exceeded the number of living children they consider appropriate for a family like theirs have a motivation of contracept. Since literacy is much higher for urban than rural women in Pakistan, we hypothesized that literacy would interpret the higher current usage of contraception by urbanites. 140 The data from 1984-85 Pakistan Contraceptive Prevalence Survey formed the bases of analyses in this dissertation. Literacy is considered to play an important role in determining the quality of data, also. Overall, rural-urban data appears to be of a reasonable quality. The coverage rates for the sampled households were very good: actually the number of households missed for rural areas (1.4%) was lower than for urban areas (3.2%). However, the sample coverage appeared to be better for younger women (below age 30) as compared to the age distribution of women reported in 1981 census of population. The better coverage of sample women in a survey can be expected due to its small—scale manageable nature. A survey is expected to provide better quality of data than census because its undertaking is closely supervised as compared to the gigantic operation involved in a census. However, this better reporting of younger women for most of whom the number of currently surviving children may fall short of the number they considered appropriate for their families might have been the reason for the finding they have a lower motivation to contracept. Contraceptive use is a private sexual matter, and revealing information on such a matter becomes embarrassing for Pakistani women. Thus, the misreporting of contraceptive use was expected, especially among rural women. However, an approximation of "shy users" (underreporting) revealed a 141 higher percent of underreporting (15.3%) among urban couples than rural couples (12.9%), a fact which did not appear consistent 'with. our' expectation. of Ibetter reporting for urbanites due to literacy. The concepts of "appropriate" number of children and desired/undesired fertility, for assessing the information on preferences of individuals, are usually criticized in demographic literature. In order to establish the reliability of this measure in 1984-85 Pakistan Contraceptive Prevalence Survey, we performed a number of inter-item consistency checks, some of them similar to those applied to countries covered in the World Fertility Survey (Lightbourne, 1985) . We found a strong correlation, suggesting consistency in responses, to the. questions about. appropriate number' of children and the desirability of more births in view of the number of living children. Thus, it was concluded that the comparison of the number of currently surviving children with the number reported as "appropriate" gave a reliable measure of which couples were ready to stop childbearing and would have a motivation to contracept. This difference in reliability was not substantial between rural and urban women. I also found.the percents of currently contracepting rural and urban respondents to be higher, when neither of the spouses wanted more children (11.3% and 27.2%, respectively). This implied that the "wantedness" of the next possible birth was 142 a meaningful regulator of marital contraception but also that spousal consensus on the desirability of stopping increased the perceived cost of not doing so. Having established the reliability of "appropriate number of children" as an index of the motivation to contracept, I sought. to connect. rurality’ to illiteracy, illiteracy ‘to "deficit fertility" (a lower number of living children than deemed appropriate) , and deficit fertility to the avoidance of contraception. Indeed, urban women were more than twice as likely to be literate as were rural women: 9.79% and 4.45%, respectively, (Table 4.2). However, literate women were not significantly more likely than the illiterate to have reached or exceeded the number of living children they deemed to be appropriate: 17.99% and 15.20%, respectively (Table 4.3) . Thus, the higher rate of literacy amongst urbanites could not interpret 'why ‘urbanites 'were 'more likely' to be current contraceptors, contrary to the hypothesis of this dissertation. This fact remained true even after the wife's stage of the life-cycle was controlled. As hypothesized, women whose number of living children had reached or surpassed what they said was appropriate were more apt to contracept currently. The literate women had higher odds of contraception than illiterate women (Table 4.8) . However, the odds of contraception (based on Table 4.9) amongst women with no "shortage" in number of living children were still much 143 lower for ruralites (0.104) than urbanites (0.346) . In other words, literacy was found to be related to a greater extent to likelihood of contraceptive use, but it did not completely explain the lower rural than urban current contraception. A likely reason is the much higher psychological and market costs of fertility regulation for rural wives. I explored what market and psychological costs might block contraception for rural women even when their number of surviving children is no longer deficient (Chapter 5). we argued that when the above mentioned costs are heavy, it was economically rational to use an irreversible, one-time-only method: and we found that surgical sterilization was thetmodal method for both rural (slightly higher) and urban women. The use of the condom was more than thrice as common among urban men, which indicated lower costs of male contraception in urban areas than in rural areas. In both rural and urban areas, but especially in rural areas, government hospitals were the most common source of contraception. It is not surprising because a family planning method like female sterilization requires a doctor trained in Western medicine and an operation theater. The popularity of condoms in urban areas is also easy to understand because of its greater availability and the access to grocery and drug stores, which were the next most frequent suppliers of the contraception to urbanites. The rural'women were more than twice as likely as 144 urban women to use the IUD and almost twice as likely to use an injection: these two methods do not require frequent visits to the source of family planning and thus are more economical for women who have difficult access to contraception. The shortage (nonavailability) of supply was observed for both rural and urban women, who used the methods (pill and condom) supplied by the family planning program. Also, quite surprisingly, rural-urban difference in who obtains the method (husband or wife or someone else) of use was not all that great. An exception was found for the 8 users of vaginal methods, of whom 50% of rural wives and 83.3% of urban wives obtained it for themselves. For oral pills, it was equally likely for husband or a wife to get the method, regardless of the rural or urban location of their residence. This fact underscores the importance of the husband's consent in order for a wife to employ birth control. The percent of current users reporting difficulty in getting to a source of family planning was much higher for rural contraceptors than urban contraceptors. Rural women were more likely to cite bad roads or no transportation as reasons for this difficulty. More of rural than urban respondents used some sort of transportation to reach the source of family planning: the lower percent of rural women who walked certainly indicated that there were fewer sources 145 within walking-distance reach in rural areas than urban areas. Likewise, the percent reporting the reasons for the difficulty in reaching the source were higher among rural contraceptors: the most common reasons were bad roads, transportation problems and longer travelling time. Hence, this clearly outlines the higher physical and financial costs of contraception for ruralites than urbanites. The findings of this dissertation offer guidelines to policy makers wishing to increase the prevalence of marital contraception in order to reduce the rapid rate of population growth ( 3% annually) in Pakistan. Recall that literate women, defined as those who could read and write a letter, had higher odds of current contraception than illiterate women. The likelihood of contraception intensified across the life-cycle more so for literate women than illiterate ones. .Also, Sathar and Mason (1989) found that a college educated woman (completing 12 years of schooling) had "almost two children fewer than her counterpart who is without school experience" (p.2). Thus, it is apparent that.government programs aimed at boosting the literacy rate, especially for rural girls, would be followed in about two decades by an expansion of contraceptive use. However, the attainment of functional literacy, requires attending schools for more than a few years of primary schooling. 146 Literacy's meaning goes beyond the ability to read and write. It increases the social and economic statuses of the literates and of the parents who send them to school. The literate daughters' chances of marrying literate or occupationally well-placed men are increased, and this prospect rewards parents who send their daughters to school. However, even.in the.government schools‘where tuition fees are very low, the books, uniforms and travel to school entail some financial cost to parents in addition to what parents had to forgo in terms of their children's labor. The government should allocate higher budgets to subsidize the cost of books and. transportation ‘to school, particularly for' the less advantaged ones in the society. The government should open more schools in.areas making it more accessible to the poor in the community. The female literacy campaigns should also use mass media sources to attract parents to send their children to schools: this would certainly bring some positive attitudinal changes in the society towards female education. If these changes occur in tandem with economic development, the pool of upwardly mobile prospective bridegrooms will grow: and in hopes of gaining such sons-in-law, the parents of daughters will become increasingly motivated to send the girls to school. Thus, eventually the motivation to control fertility will follow for the younger generation. 147 To promote literacy the schooling system must be strengthened. Most rural primary schools lack resources which are essential to increase the quality and efficiency of the educational system. Shah (1986) noted that "of schools (during 1977-78) for girls, 87% of the primary and 78% of middle (and 26% of high) schools were estimated to be located in rural areas": she considered this concentration to be reasonable yet the opportunity, above the middle school-level, was limited because the location of the school in the area of residence is considered important for female attendance. The physical existence of schools at below-high-schools-level does not appear to be that much of a problem but the shortage of female teachers in rural areas has remained a major constraint. Where the teachers are available, their quality of work, commitment and "love" for students are‘criticized for several reasons that affect the quality of the education provided to students (p. 244). The government could promote new 'teacher's training' and. educational programs for’ the students from rural areas by offering them scholarships and making the service to their own home-schools mandatory for a specified period of time. The government could also promote the location of talented teachers in rural areas by offering them better salaries, government housing and assistance in securing support of local people. The newly appointed teachers' effectiveness could be increased by creating a favorable working environment for them in rural areas. The 148 government can institute exclusive units within the local district boards to deal with the female teachers' grievances (they may face greater hostility in a community which is not ready for change and also for the mere fact of being the teachers of a feminine gender). The expedited procedures for resolving those grievances would become a very effective support system. Shah also noted that "aspirations of rural parents for their daughters' education are very low": according to a study of Punjabi villages, 32% of the parents desired no education or only religious education for their daughters (p. 253). This poses a significant hindrance in female education. The contents of school curricula are criticized for "haphazard and vague objectives" (Shah, 1986). I believe that the school textbooks and other educational programs must emphasize the norms regarding smaller families, changing gender roles and the social power attained by wives and their husbands due to the education. These changes should minimize the importance of son preference and hence create a desire for fewer but healthier and literate children. Married couples in Pakistan might also be persuaded to use contraception not only for stopping but also for spacing their children's births. Even having the same number of births but having them farther apart would slow the rate of population growth. 149 Some policy makers may argue that increasing female literacy is a long-term.process for increasing the motivation among women or couples to desire fewer children and to use contraception. Thus, policies which can be implemented in the short run should be undertaken, too. For example, information about family planning could be sent through channels that do not require literacy: e.g., television and radio announcements, and pictorial billboards. Such messages would have a better chance of reaching the 84% of the women who are illiterate. When a subject is sensitive or taboo, literate women can get messages in way that does not require face—to- face contact. Oral media would dispel the disadvantage of illiterate women in this regard but would also serve to weaken the taboo against talking about such matters. The early family planning programs of the 1960s and the early 19708 had used the above-mentioned sources for carrying out the propaganda campaigns. We noted that in the 1968-69 National Impact Survey the use of any family planning program method was higher (3.9% among rural, and 9.8%, among urban women, respectively) than what it was found in the later period. In the 1979 Population, Labor Force and Migration Survey it was 1.8% among rural, and 7.7% among urban women, respectiveLy. This decline was attributed to the changing government policies under the new political regime with higher emphasis.on "religiouswmorality" of the late 19705 (Population 150 Welfare Division, 1986). That is the period, when the propaganda of "Islamization" caused all the pictorial billboards and family planning announcements to disappear: even the.name of the organization was changed to the one which did not directly address the concept of contraception. In any case, the message through pictorial billboards would benefit those few women who go outside the home on a regular basis, but a majority of men would definitely see those and could bring the message home. We must create ways for family planning messages to reach the masses, especially women. In addition to sending family planning messages in Urdu and English languages at the national level, the messages must be reproduced in Sindhi, Baluchi, Pushto, and Punjabi at each respective provincial level. This would increase women's exposure leading towards motivation to contraceptive adoption and in granting them a greater autonomy. More research needs to be done to investigate the areas of demand for children and demand for contraception. Research on "shy users" might reveal useful information for new family planning strategies. The hospital and clinical records on sterilization cases might provide more complete information than question-answer sessions (interviews). Couples should be provided with the methods of family planning they want. Development of non-physician-administered 151 methods, such as, anovulant pills and self-insertible IUDs, would accommodate shy women contraceptors by eliminating the need to visit a physician. Female sterilization appeared to be one of the dominant methods for stopping childbearing. Male sterilization (vasectomy) could be promoted as a one-time contraceptive act which might reduce the physical costs associated with female sterilization. Thus, the Government of Pakistan can reduce the cost of contraception not only by building more government hospitals but also by increasing the supply of non-physician-administered methods and addressing husbands as well as wives. In the area of family planning program, a more effective planned information system regarding the availability and use of contraception must be developed to reach the illiterate rural masses as by creating new distribution outlets situated within walking distance. The management, supervision and keeping up with the supply of family planning, with minimized cost in the rural areas must enhance the demand for contraception among rural women. Thus, a close scrutiny of the administrative structure of the family planning program is required, if the efficiency of its service-supply system is to increase. .An example to comprehend such a concern would be to undertake a community level, clinically based longitudinal study or a follow-up study of family planning clients, as well as the family planning personnel according to the source of 152 supply. The information on duration of actual effective use of’ contraception is important. Is ‘that. duration-of-use sufficient to prevent or postpone a birth for a woman? What has been the history of pregnancy/childbirth and birth-spacing for that period of duration-of-use for that particular client? Has she used this method of contraception successfully? What kind of complaints did she make about the method, if any? What was the source of supply, and what were the characteristics of providers? Were the providers of services "appropriate" personnel in view of the clients? Were the providers socially-distanced or accommodating personalities? Is the problem of contraceptive provision more of a structural-type or of a functional-type? Is it a problem of supply or management of the distribution system or of demand for contraception? The answers to above questions will mark an important leap forward in better comprehension of Pakistan's family planning program. Thus, a systematic diagnosis of the above factors can lead the program to discern what to improve, where to improve, and how to improve its services in order to raise the prevalence of contraception.and to make it more similar between rural and urban couples in Pakistan. LIST OF REFERENCES Abbasi, Nasreen. 1987 Urgenirerien_in_£ekieren. Islamabad: Pakistan Institute of Development Economics. Research Report No. 152. Ahmed, Ghyasuddin: William P. Schellstede and Nancy E. Williamson. 1987 "Underreporting of Contraceptive Use in Bangladesh. " WW Eerepeeriyee. Vol. 13, No. 4:136-140. Ahmed, Nilufer R. 1981 "Family Size and Sex Preferences Among Women in Rural Bangladesh. " W Vol. 12, No. 2:100-109. Alam, Iqbal and Betzy Dinesen, (eds.) 1984 wwl't sta : W ' ' ' . Voorburg, Netherlands: International Statistical Institute. Alam, Iqbal, Muhammad Irfan and M.N.I. Farooqui. 1984 Fertility Levels and Differentials in Pakistan: Evidence From the Population, Labour Force and Migration Survey, 1979-80. Pakistan Institute of Development Economics, Islamabad. £LM_2rejeer B§22£L_EQL_1L Aldrich, John H. and Forrest D. Nelson. 1984 W991} Sage Publications: Series No. 07-045. Beverly Hills. Ali, Syed Mubashir. 1989 "Determinants of Family Size Preferences in Pakistan- " 111W. Vol - 28, NO. 3:207-231. Arnold, Fred., R. A. Bulatao, C. Buripakdi, B. J. Chung, J. T. Fawcett, T. Iritani, S. J. Lee, and T. 8. Wu. 1975 W: W Introduction and Comparative Analysis. Vol. 1 Honolulu, Hawaii: East-West Population Center, East-West Population Institute. Benavot, Aaron. 1989 WW: 8 W. Michigan State University: Office of Women in International 153 154 Development. Working Paper #194. Bogue, Donald J. ' 1969 Brineinle§_gf_nemegraphy. New York: Wiley- Booth, Heather and Iqbal H. Shah. 1984 "The Data and their Quality. " inF _erriliry_ig £eki_ren: A 3e vie! ef Eigdings from the Pakistan Fertiliry §grv y. (eds. ) Alam, Iqbal and Betzy Dinesen. Voorburg, Netherlands: International Statistical Institute. Cain, Mead. 1980 "Risk, Fertility and Family Planning in a Bangladeshi Village-" .8tQdie§_in_Eamily;Elanningl Vol. 11, June: 219- 223. Caldwell, John C. 1980 "Mass Education as a Determinant of the Timing of Fertility Decline." 29pulatign_and_neyelepment Reyie_. v61. 6:225-255. Caldwell, John C. 1981 "The Mechanisms of Demographic Change in Historical Perspective." Beeglerien_§reg1e§. ‘Vol.?? 5-27. Caldwell, John C. 1982 The9ry_ef_Eertility_Decline. London: Academic Press . Caldwell, John C., I. Gajanayake, B. Caldwell, and P. Caldwell. ' 1988 "Is Marriage Delay a Multiphasic Response to Pressures for Fertility Decline? The Case of Sri Lanka." A lecture-seminar given by John Caldwell and Pat Caldwell, at the University of Michigan. Ann Arbor. Caplan, Patricia and Janet M. Bujra. 1979 E2men_unitedl_flemen_niyided= semparatiye_§tudy_9f ten_ggntemperary_sulture§- Bloomingtcn: Indiana University Press. Casterline, John. 1984 "Fertility Differentials in Pakistan." in Fertility_1n_gakistan= A_Beyieg_2f_fiinding§_frgm rhe Pakisten Fertility §grv y. Alam, Iqbal and Betzy Dinesen (Eds.). Voorburg, Netherlands: International Statistical Institute. Cleland, John and John Hobcraft (eds.) 1985 Bepr9d2ctiye_Qhange.1n_neyeleping_sgnntries:' 155 . Oxford University Press. Cleland, John and Zeba Sathar. 1984 "The Effect of Birthspacing on Childhood Mortality in Pakistan." Beeglerien_§rggiee. Vol. 38, No. 3. Cochrane, Susan Hill. 1979 W: W? World Bank Staff Occasional Papers: Number Twenty Six. Published for the World Bank by the Johns Hopkins University Press. Davis, Kingsley, and Judith Blake. 1956. "Social Structure and Fertility: An Analytic Framework." engage. Vol. 4, No. 3:211-235. Das, Narayan. 1984 "Sex Preference Pattern and its Stability in India." Qemegrephy_1nd1e. Vol. 13, Nos. 1- 2:108-119. Easterlin, Richard A. 1974 "The Effects of Modernization on Family Reproductive Behavior ." In The_£gpnlatign WW Papers of the World Population Conference, Vol. 2, Bucharest. New York: United Nations. Easterlin, Richard A. 1975 "An Economic Framework of Fertility Analysis." Weaning Vol- 6. No. 3: 54- 63- Easterlin, Richard A. and Eileen Crimmins. 1982 "An Exploratory Study of the 'synthesis framework' of fertility determination with World Fertility Survey data." WW. Scientific Reports No. 40:1-39. Farooqui, M. Naseem Iqbal. 1984 "Factors Affecting Rural Fertility in Pakistan." Demegrephy_1ngie. Vol. 13, Nos. 1-2:70-91. Fienberg, Stephen E. 1985 - e page. The MIT Press. Massachusetts: Cambridge. Graff, Harvey J. 1979 "Literacy, Education, and Fertility, Past and Present: A Critical Review. Repnlerien_en§ 156 Deyelepment_neyiex. Vol. 5. No. 1:105-140. Institute for Resource Development. 1990 Demographic and Health Surveys: Newsletter. Vol. 3, No. 1:1-12. Johnson, Nan Edith. 1977 Earmzngnfarm_n1fferent1al§_1n_Eert1lity= Effects of Composirienal egg fiex- gele Eeerors. Unpublished Ph. D. dissertation. Pennsylvania State University. Karsada, John D. 1973 "Economic Structure and Fertility." A Comparative Analysis." Qemegrephy 8, No. 3:307—317. Kent, Mary Mederios and Ann Larson. 1982 Family_§1ze_£reference§: Eyidence_frgm_the_flerld r ' i e . Reports of the World Fertility Surveys #4. Population Reference Bureau, Inc. Khan, Seemin Anwar and Faiz Bilquees. 1976 "The Environment, Attitudes and Activities of Rural Women: A Case Study of a Village in Punjab." The Eakistan_Deyelepment_Beyiew. Vol. 15. No. 3:237- 271. Khan, Ali and Ismail Sirageldin. 1977 "Son Preference and Demand for Additional Children in Pakistan." Qemegrepny. Vol. 14, No. 4:481-496. Khan, Ali and Ismail Sirageldin. 1979 "Education, Income and Fertility in Pakistan." Es9n2m1c_Deyel9pment_and_snltural_9nange- pp=519- Lee, Ronald D., and Rodolfo Bulatao. 1983. "The Demand for Children: A Critical Essay." Chapter 8 in s. (Eds.) Bulatao and Lee. New York: Academic Press. Lieberman, Samuel. 1982 "Demographic Perspectives on Pakistan's Development." Vol. 8, No. 1:85-120. Lightbourne, R.E. 1985 "Individual Preferences and Fertility Behavior." in John Cleland and John Hobcraft (editors), BenI9d22t1ye_Qhange_1n_Deyeleping_quntrie§: 157 . Oxford, W0 F PP. 165-199. England. Oxford University Press. Mamlouk, Maria. 1982 W e o Qeveloping Countriee. A publication of the population Reference Bureau, Inc. Report on World Fertility Survey No. 3. Mason, Karen Oppenheim. The_§tatg§_gf_flemen: A_Bey1eu_ef_1t§_Eelatign§hip§ t2_Eertility_and_ngrtality- The Rockefeller 1985 Foundation: New York. Mason, Karen Oppenheim. and Anju Malhotra Taj. 1987 "Differences between Women's and Men's Reproductive Goals in Developing Countries." £epg1e§ien_egg QQYELQQEEDLLBQYLQH- Vol. 13, No. 4:611-638. McDonald, Peter. "Social Organization and Nuptiality in Developing 1985 Societies." in John Cleland and John Hobcraft (editors), Bepredgetiye Change in Qeveloping ‘e’ ISO"! ‘ an 1‘ 0 0 ‘ ' ° 1‘ . 0-1t - -_ , _ Oxford,England. Oxford University Press. pp. 87-114. McHenry, John Persons. "A Critique of The Easterlin Approach to Micro- Population Bulletin of 1984 Level Fertility Analysis." the United Nations. ST/ESA/SERN/17. No. 17:80- 93. Mernissi, Fatima. 1987 Beygnd_the_Yeil: Male:Eemale_nynam1c§_1n_nedern ' ' . Bloomington and Indianapolis: Indiana University Press. Meyering, Anne C. 1990 Eertilit¥_lransitien§: An_derxiey_gf_Iheery1 Eyidenee egg Policy Implicatione. Department of History, Michigan State University, Michigan: East Lansing. (a typed Manuscript). Miller, Barbara D. 1983 _‘l ' ‘ 2, 1! ‘ .é_'! ‘ ‘° :- 21". _ 3 1 ". Beties: Pakisrag and gengladesh Cemeree. Michigan State University: Office of Women in International Development. Working Paper #30. Pakistan Planning Commission. 1986 Eak1stan_E2989m12_§uryey_12§§:128§o Islamabad: 158 Ministry of Planning and Development. Pearson, Karl. 1899 "On a Criterion that a Given System of Deviations from the Probable in the case of a Correlated System of Variables is such that it can be reasonably supposed to have arisen from Random Samplin9-" Eh1lesephy_nagazine. Vol. 50 No. 5: 157- 175. Population Reference Bureau. 1990. E9rld_29pelatign_nata_§heetl_1229- Washington D- C. Population Welfare Division of Pakistan. 1986 Eak1etan_Q9ntraQentiye_£reyalense_§uryey (PCPS) 1984-85. Government of Pakistan, Islamabad: Ministry of Planning and Development. Pullum, Thomas. 1983 "Correlates of Family Size Desires." Chapter 10 in Determ1nants_9f_£ertility_1n_nexelepins gednrriee. (Eds.) Bulatao and Lee. New York: Academic Press. Pullum, Thomas W. and Sajjad Latif Awan. 1988 "In Search of The Silent User: The Reliability of Reported Use and Non-Use of Contraception in Pakistan." in c d e s rese sem1nar_9n_the_finding§_9f_the_zak1§tan Q9ntracentiye_£reyalence_§nryey- August 11. pp- 35-49. Pullum, Thomas W. 1988 "Evidence for and Against The Emergence of Family Size Norms in Pakistan." in Selected Pepere Presented_at_the_Sem1nar_Qn_the_£1ndings_pf_the Pak1stan_Q9ntracentiye_£reyalence_§urxey- August 11. pp. 35-49. Rao, K. Vaninadha and Roderic Beaujot. 1986 "Effect of Infant Mortality on Subsequent Fertility in Pakistan and Sri Lanka." l29£fl21.2£.fii§§221§1 £21§22_§- Vol. 18: 297- 303. Retherford, R.D. 1985 "A Theory of Marital Fertility Transition." 2922181198_§tgdies- No. 39:249-268- Robinson, Warren C. 1978 "Family Planning in Pakistan 1955-1977: a review". Ihe_Eak19188.nexelepment_neyiey. Vol. 17. Na- 159 2:233-247. Robinson, Warren C. 1987 "The 'New Beginning' in Pakistan's Family Planning Programme" - WM- Vol - 6, No. 1:107-118. Ryder, Norman B. and Charles F. Westoff. 1971 "Wanted and Unwanted Fertility in the United States: 1965 and 1970." In, Charles F.‘Westoff'and Robert Parke. Jr- (Ed-). Demographis_and_§92111 Aspeste_of_£opulation_§royth. ‘Vol- 1. Commission on Population Growth and the American Future. Sanderson, Warren C. 1976 "On Two Schools of the Economics of Fertility." v . Nos. 2,3 and 4 (September-December): 469-477. Sarmad, Khawaja, Fazal Husain and G. M. Zahid. 1988 "Investment and Inequality in Pakistan's Education Sector." Ihe_Eakistan_neyelopment_neyiey- Vol. 27, No. 4:677-686. Sathar, Zeba Ayesha. 1979 "Rural-Urban Fertility Differentials: 1975." The_£akistan neyelopment_Beyiem. Vol. 18. No. 3:232-251. Sathar, Zeba Ayesha. 1984 "Does Female Education Affect Fertility Behaviour in Pakistan?” The_Pak1§tan_neyelopment_Beyie_- Vol. 23, No. 4:573-590. Sathar, Zeba, Nigel Crook, Christine Callum and Shahnaz Kazi. 1988 "Women's Status and Fertility Change in Pakistan." 0 o v . Vol. 14, No. 3:415-432. Sathar, Zeba A. and Afifa Akhtar. 1988 "Evidence of Fertility Decline in Karachi." The 2akistan_neyelopment_seyiey- Vol. 27. N. 4:659- 670. Sathar, Zeba A. and Shahnaz Kazi. 1988 Broductive end Reprodderige gheieee er Merrepelirendflemeh: Report of a Survey in Karachi. Islamabad: Pakistan Institute of Development Economics. Sathar, Zeba A. and Karen Oppenheim Mason. 1989 m ’v 160 . University of Michigan: Population Studies Center. Research Report NO. 89-194. Schultz, Theodore W. 1973 "The Value of Children: An Economic Perspective." ' WW- Vol- 81. N0- 2:82-13. Shah, Iqbal, Thomas W. Pullum and Muhammad Irfan. 1986 "Fertility in Pakistan During the 1970s." degree; eLEieeeeieljeieneee- Vol- 18:215-229. Shah, Nasra M. and Makhdoom Shah. 1980 "Trends and Structure of Female Labor Force Participation in Rural and Urban Pakistan." Weneh WWW- (Editor) Alfred de Souza. New Delhi: Indian Social Institute. Shah, Nasra M. and Makhdoom A Shah. 1984 "From Non-Use to Use: Prospects of Contraceptive Adoption." in Alam, Iqbal and Betzy Dinesen (eds.) W Voorburg, Netherlands: International Statistical Institute. pp. 149-162. Shah, Nasra M. (Editor) 1986 W: LaeeiefleenenieW £ref11e. Pakistan Institute of Development Economics and East West Population Institute. Shryock Henry S. and Jacob S. Seigel. 1975 Ihe_Hetnede_enQ_Heter1ele_ef_2emegrennx- U-So Department of Commerce: Bureau of Statistics. Singh, Susheela and John Casterline. 1985 "The Socio-Economic Determinants of Fertility." in John Cleland and John Hobcraft (eds. ) Repredueg1ve Chehge 1n Qevelop1ng Ceghrr1ee: W u . Oxford, England, Oxford University Press. Smith, Peter C. 1983 "The Impact of Age at Marriage and Proportions Marrying on Fertility." Chapter 13 in W e: Fert111ry 1h Developihg ngntr1 e. (Eds. ) Bulatao and Lee. New York: Academic Press. Vol.2. Snedecor, George W. and William Cochrane. 1977 firer1er1ee1_herhede. Iowa: The Iowa State University Press. 161 Soomro, G. Y and Mubashir Ali. 1984 WWW- Pakistan Institute of Development Economics: PLM Report No. 3. Statistical Package for Social Sciences. 1986 §£§§_:jfleer1e_§d1de. (2nd. edition) SPSS Inc. McGraw-Hill Book Company. Syed, Sabiha Hassan. 1978 "Female Status and Fertility in Pakistan." The W. Vol 17 No. 4:408-431. Tsang, Mun. and M. Javed Akbar Zaki. 1989 "Household Expenditures and Other Alternatives for Financing Primary Education in Pakistan." Paper prepared for Basic Research and Implementation in Developing Education Systems (BRIDGES). Michigan State University: College of Education. Tsui, Ami Ong. 1985 "The Rise of Modern Contraception." in John Cleland and John Hobcraft (eds. ) Begrgdder1ye Qnen9e_1n_Deyelening_§enntriee: Ine19n1e_fren_tne MW- England: Oxford University Press. Weiss, Anita M. 1985 W: ' WW- Michigan State University: Office of Women in International Development. Working Paper #78. Westoff, Charles F., Elliot G. Mishler, and E. Lowell Kelly. 1957 "Preferences in Size of Family and Eventual Fertility Twenty Years After." Amer1eeh_degrhdl ef_§ee1elegy. (March):491-497. Whyte Robert Orr. and Pauline Whyte. 1982 The_flemeh_er_fidre1_he1e. Boulder: Westview Press. Zaki, Khalida Parveen and M. Javed A. Zaki. 1981 "A Comparative Study of Age Reporting in Pakistan Censuses and Surveys, 1951- -81." Demegrephy_1hd1e. Vol.12,No. 1. APPENDIX APPENDIX Husband's Occupational Status by Residence We compared husband's occupational status in rural and urban areas of our study (Appendix Table 1.1). We found that the percent of husbands with agricultural occupations was much higher in rural areas (46.7%) than in urban areas (7.5%). As expected, the percentages of husbands who were salaried or self-employed was substantially higher in the urban sector (44.9% and 43.3%, respectively) than in the rural sector (25.3% and 24.0%, respectively). This association was statistically significant as well (Chi-square = 1269.373, D.F = 3 and P < 0.05). Hence, it is apparent that rural and urban residences imply substantially different forms of economic behavior and thus different degrees of vulnerability to the physical environment. Wife's Literacy and Husband's Occupational Status From Appendix Table 1.2, we found that both in rural and urban areas a higher percentage of literate women were married to those who were not affiliated to farm-related occupations. Percent of literate wives (43.6% in rural and 49.4% in urban) was the highest among the salaried-employees. The second most popular occupation was self-employment of husband among literate women, particularly in urban areas (44.2%) and in rural areas (25.5%), which did not differ greatly from the percentage in rural areas because 25% of rural literate women 162 163 Table 1.1: Husband's Occupational Status by Place of Residence in Pakistan, PCPS 1984-1985. Residence Husband's Occupational Rural Urban Status % % Farm- Related 46.7 7.5 Salaried- Employee 25.3 44.9 Self- Employed 24.0 43.3 Unemployed 4.0 4.4 Total 100% 100% (N=4505) (N=2870) Note: Chi-square = 1269.373 D.F = 3, P < 0.05. Total for all Pakistan = 7394. Missing cases = 19. Valid cases in this table = 7375. Table 1.2: 164 Percent of Wives Literate by Husband's Occupational Status in Pakistan, PCPS 1984-85. Husband's Literate Illiterate Occupational ' Status Rural Urban Rural Urban % % % % Farm- Related 25.0 2.6 . 47.7 8.0 Salaried- Employee 43.6 49.4 24.4 44.0 Self- Employed 25.5 44.2 23.9 43.2 Unemployed 5.9 3.7 4.0 4.5 Total 100% 100% 100% 100% (N=204) (N=269) (N=4301) (N=2601) Note: Chi-square: Rural areas = 51.824, D.F = 3, P < 0.05. Urban areas = 11.127, D.F = 3, P < 0.05. Total for all Pakistan = 7394. Missing cases = 19. Vali d cases in this table = 7375. 165 were married to husband who was employed in farm-related occupation. In both the rural and urban areas this relationship was found to be statistically significant (Chi- square statistic for rural area = 51.82 at d.f. = 3 and for urban area = 11.13, at d.f. = 3). Wife's Literacy and Mean Number of Children We conform to our presumption that literate parents tend to have fewer children than illiterate parents (Appendix Table 1.3). We noted that on the average the number of children deemed appropriate by a literate respondent (4.53) was almost 0.44 of a child less than that of an illiterate one (4.97). Accordingly, the mean number of children ever born, and surviving were lower for literate (3.76 and 3.29 children, respectively) than illiterate wives (4.37 and 3.57 children, respectively). Wife's Literacy and Age at Marriage Appendix Table 1.4 examines the relationship between wives' literacy and age at marriage as found in the 1984-85 Pakistan Contraceptive Prevalence Survey. As expected, we found that in the PCPS data (Table 2.4) a greater percent of illiterate women, 59.1% in rural and 54.3% in urban areas, married at earlier ages (below or at 17 years) than the literate women (51% in rural and 51.3% in urban) in the same age. This finding conforms to our expectation that literate 166 Table 1.3: Mean Number of Children Deemed Appropriate, Ever Born and Surviving to Pakistani Women by Literacy Status, PCPS 1984-85. Number of children Literacy Status Appropriate Ever Born Surviving Literate 4.53 3.76 3.29 Illiterate 4.97 4.37 3.57 Total 4.94 4.33 3.55 167 Table 1.4: Percent of Wives Literate by Age at Marriage in Rural and Urban Pakistan, PCPS 1984-85. Literacy Status/Residence Age at Literate Illiterate Marriage (years) Rural Urban Rural Urban % % % % <17 51.0 51.3 59.1 54.3 17+ 49.0 48.7 40.9 45.7 Total 100% 100% 100% 100% (N=204) (N=269) (N=4308) (N=2611) Note: Chi-square: Literate = 0.005, D. Illiterate = 15.39, Total all Pakistan = Missing cases = 2. F = 1, D.F = P > 0.05 1, P < 0. 7392. 168 women tend to marry at later ages. .Although.these percentsare not greatly different, the difference is statistically significant (Chi-square = 15.39, D.F = 1, and P < 0.05) for illiterate women. For literate women the statistical difference is not significant (Chi-square = 0.005, D.F = 1, and.P'> 0.05), which implies that literacy have.greater impact for rural women than urban women. "ITiliifliifi‘fl‘fli“!11111