This action might not be possible to undo. Are you sure you want to continue?
National Family Health Survey Subject Reports, No. 3
Son Preference and Its Effect on Fertility in India
Rangamuthia Mutharayappa, Minja Kim Choe, Fred Arnold, and T. K. Roy
National Family Health Survey Subject Reports Number 3 • March 1997
International Institute for Population Sciences Mumbai, India East-West Center Program on Population Honolulu, Hawaii, U.S.A.
2 National Family Health Survey Subject Reports, No. 3
India’s National Family Health Survey (NFHS) was conducted in 1992–93 under the auspices of the Ministry of Health and Family Welfare. The survey provides national and state-level estimates of fertility, infant and child mortality, family planning practice, maternal and child health care, and the utilization of services available to mothers and children. The International Institute for Population Sciences, Mumbai, coordinated the project in cooperation with 18 population research centres throughout India, the East-West Center Program on Population in Honolulu, Hawaii, and Macro International in Calverton, Maryland. The United States Agency for International Development provided funding for the project. ISSN 1026-4736 This publication may be reproduced for educational purposes.
Correspondence addresses: International Institute for Population Sciences Govandi Station Road, Deonar, Mumbai - 400 088, India Fax: 91-22-556-32-57 • E-mail: firstname.lastname@example.org East-West Center, Program on Population/Publications 1601 East-West Road, Honolulu, Hawaii 96848-1601, U.S.A. Fax: 1-808-944-7490 • E-mail: email@example.com
National Family Health Survey Subject Reports, No. 3
Son Preference and Its Effect on Fertility in India
Abstract. Numerous studies have found that most Indian couples have a strong preference for sons over daughters. In an effort to have sons, many couples continue to have children after achieving their desired family size. This practice may have retarded India’s fertility decline. Using data from the 1992–93 National Family Health Survey, this report assesses the prevalence of son preference in India as a whole and in the 19 most populous states. The state-level analysis is important because fertility levels, social and economic conditions, and the strength of son preference vary widely from one part of the country to another. The analysis compares the ideal number of sons and the ideal number of daughters mentioned by Indian women as well as contraceptive use by women with two sons and women with two daughters. It goes on to compare the situation for boys and girls in terms of immunization rates, period of breastfeeding, prevalence of three common childhood diseases and likelihood of treatment, prevalence of chronic undernutrition among children under age 4, and infant and child mortality rates. On nearly all these measures and in most states, male children have a decided advantage over female children. Son preference is particularly strong in northern and central India and somewhat weaker in the southern and western regions. Next, the analysis uses life-table methods and hazard models to examine the effect of son preference on fertility. It also examines how the effect of son preference interacts with a woman’s social and economic characteristics, in particular her rural or urban residence, literacy, and religion. The effect of son preference on fertility is found to vary substantially by region and state. In states where fertility is very high or very low, the effect is small, as expected, but in states with intermediate levels of fertility (between 2.0 and 3.0 children), it varies widely. It is highest in Himachal Pradesh, Punjab, Gujarat, and Maharashtra and lowest in West Bengal and most of the southern states. If gender preferences could be eliminated entirely, the fertility level in India would decline by about 8 percent. A decline of this magnitude would have a substantial impact on the population growth rate. Eliminating son preference would also have important social benefits.
Rangamuthia Mutharayappa, Minja Kim Choe, Fred Arnold, and T. K. Roy Rangamuthia Mutharayappa is an assistant professor at the Population Research Centre of the Institute for Social and Economic Change, Bangalore. Minja Kim Choe is a fellow at the East-West Center’s Program on Population. Fred Arnold is a senior population specialist at Macro International, Calverton, Maryland. T. K. Roy is a professor and head of the Department of Population Policies and Programmes at the International Institute for Population Sciences. National Family Health Survey Subject Reports, Number 3 l March 1997
3 . No.4 National Family Health Survey Subject Reports.
Kapadia 1966. C. the utility of having sons arises from the important religious functions that only sons can provide. Verrall. IIPS 1995). and other parts of South Asia. North Africa. and Kanazawa 1994. they typically become a member of their husband’s family after marriage and may have little continuing contact with their natal family. although there is some recognition that sons are no longer a dependable source of old-age support (Bardhan 1988. 3 A s in societies of East Asia. Cleland. Upon marriage. PRC. she is considered to be an economic liability to her parents mainly because of the heavy dowry payment demanded by the groom’s family (Kishor 1995). M. Miller 1981). Although a daughter provides help in housework before marriage. 1997. which is generally the responsibility of the bride’s family to bear. and P. This report examines the patterns of son preference and its effect on fertility in India overall and in 19 of its states. earn wages. Not surprisingly. sons are needed to kindle the funeral pyre of their deceased parents and to help in the salvation of their souls. Dyson and Moore 1983. economic. Although daughters are often considered to provide more emotional satisfaction to parents than are sons (Dharmalingam 1996. and many sons provide additional status to the family (J. According to Hindu tradition. Finally. Caldwell 1989. and economic factors influence the relative benefits and costs of sons and daughters and ultimately parents’ gender preferences (Arnold et al. Hechter. In the context of India’s patrilineal and patriarchal family system. a son brings a daughter-in-law into his family. and support their parents during old age. India exhibits substantial variations in its geographic. Karve 1965). Mamdani 1972. No. and she provides additional help around the house as well as an economic reward in the form of dowry payments. Another important advantage of having sons is their sociocultural utility. and Roy 1996. Studies in India have identified three major factors that underlie son preference. Vlasoff 1979). Reddy. Parents also bear a large burden in arranging a suitable marriage for their daughters and protecting their chastity before mar- . Choe. Lucknow University. as well as the high cost of the wedding. social. 1985. Vlasoff 1990. UN 1981. Espenshade 1977. One is the economic utility of sons.5 National Family Health Survey Subject Reports. Parents’ preferences for sons may have a significant impact on children’s welfare and may affect demographic behavior as well. A large and populous country. in India couples have been observed to have a strong preference for sons over daughters (Arnold 1996. Previous studies have found that a number of cultural. Bulatao 1981. Dharmalingam 1996. Sons are more likely than daughters to provide family labor on the farm or in a family business. having one son is imperative for the continuation of the family line. large differentials in the degree of son preference and in demographic behavior have also been noted among India’s regions (Arnold. Socially. the utility of having daughters is small compared with their cost. Williamson 1976) and to accord a low status to women (UN 1995). 1975. Caldwell. Friedman. and IIPS 1994). the Middle East. Basu 1989. and cultural environment. and Vaessen 1983.
3 riage. Some parents also cite the need to have a daughter to cry at the time of their death (Dharmalingam 1996). State-specific analyses are important in India. Moreover. In low-fertility societies. 1992. during the three years preceding India’s 1992–93 National Family Health Survey (NFHS). 1987. Parasuraman. Bairagi and Langsten 1986. in many cases the father of the bride has to assume a humiliatingly low posture in the presence of the groom and his family. most couples continue to have children regardless of the number of sons and daughters they already have. literacy. We focus on the parity progression of women with two or three surviving children because most decisions about whether to stop childbearing seem to occur after a couple has had two or three children.6 National Family Health Survey Subject Reports. is thought to be most pronounced in countries like India that are in the middle of the fertility transition. We also examine how the effect of son preference interacts with a woman’s socioeconomic characteristics. For example. as do socioeconomic conditions . 51 percent of births were of order 1–2. Existing studies. Srinivas 1977). The effect of son preference on fertility. Research on the relationship between son preference and fertility is confounded by the observation that the link is weak in both high-fertility and low-fertility populations. In high-fertility societies. specifically urban/rural residence. Haddad et al. No. Das 1984. The strength of son preference varies considerably from one part of the country to another. According to Hindu tradition. however. where decision making and program implementation are largely decentralized. and religion. fertility has declined dramatically in some countries where son preference is still widespread—for example. do not demonstrate a consistently strong effect of son preference on fertility (Arnold 1997. and only 19 percent were of higher orders (IIPS 1995. controlling for the survival status of the last child and selected sociodemographic characteristics of the woman. and Sureender 1994. 1996. Operations Research Group 1990. At the wedding ceremony. there is one important reason for having a daughter: her parents can earn religious merit by selflessly giving her away in marriage (kanya daan).11). in South Korea and China. Roy. 29 percent were of order 3–4. Koenig and Foo 1992. table 5. therefore. based on women with two or three surviving children. We use a modified version of conventional parity progression. however. 1989. Park 1986. rather than on the number of children they have ever borne. In this report we examine how son preference influences parity progression in India as a whole and in individual states. the influence of son preference is also weak because few couples want to have more than one or two children even if they do not achieve their ideal number of sons and daughters. A strong preference for sons may be an obstacle to fertility decline if couples continue having children after reaching their overall family-size goal because they are not satisfied with the sex composition of their children. The analysis estimates the effect of the sex composition of living children on the probability of a woman's having another child.
A majority of women are illiterate in all states except Delhi. The literacy rate among women is quite low in India. 3 and levels of fertility and mortality. These include Jammu. Nagaland. We have used appropriate design weights in both our analyses for the states without self-weighting samples and our analyses for the national sample. and several of the small northeastern states. Assam. and Tripura) are not shown separately in most tables. West Bengal. This report analyzes data from the 19 states having samples of more than 2. ages 13–49. Kerala. and Bihar. It was conducted in 25 states. Kanitkar and Murthy 1983. Uttar Pradesh. Bihar. Meghalaya. SOCIOECONOMIC AND DEMOGRAPHIC CHARACTERISTICS OF STATES Key socioeconomic and demographic characteristics show pronounced variations among India’s states (Table 1). In Punjab more than half of the population is Sikh. in a few states the proportion of Muslims ranges between 16 and 26 percent.500 women. Uttar Pradesh. No. stratified sample design. urban areas were purposely oversampled (IIPS 1995). but the percentage of the population living in urban areas in those states varies from only 10 percent in Himachal Pradesh to 49 percent in Mizoram. The NFHS had a systematic. SOURCE OF DATA We use data from the 1992–93 National Family Health Survey of India. In states where the urban population was not sufficiently large to provide a sample of at least 1.562 households. multistage. Goa. Female literacy is particularly low in Rajasthan. Muslims constitute the second largest religion.000 completed interviews with eligible women. and the households covered in the survey included more than 500. Of particular importance is the fact that son preference has been found to be consistently stronger in northern India than in the South (Basu 1992. the evermarried women interviewed in the survey come predominantly from rural areas.000 residents. The NFHS was a nationally representative sample survey of 89. A large majority of Indians are Hindus (82 percent).7 National Family Health Survey Subject Reports. Khan and Prasad 1983). In all of the states except Delhi and Goa. and four of the . Although nationwide only 11 percent of household heads are Muslim. Khan and Gupta 1987. Himachal Pradesh. where more than 70 percent of women are illiterate. and in Goa. in 88.777 ever-married women. Manipur. Mizoram. which include more than 99 percent of India’s population. Madhya Pradesh. but those states are always included in the totals for India. and Kerala. Estimates for the six small northeastern states (Arunachal Pradesh. Bhatia 1978. Kerala.
6 14.2 9.2 5.7 0.3 2.7 55.0 6.2 36.7 7.3 67.1 5.6 3.3 88.2 82.8 26.4 8. and household heads (percentages).9 11. 3 Table 1 Selected background characteristics of ever-married women.9 72. by state: India.3 93.9 79.0 5.1 62.9 15.6 19.1 22.8 15.9 81. Mizoram. Orissa.8 9.) Rajasthan. Buddhism.7 69. In Arunachal Pradesh 15 percent of household heads are Christian.7 77.6 6.1 8.9 * 17.8 16.3 83.7 0.2 18.5 23.0 20.7 78.9 4.0 79.1 37.2 97.2 28.0 94.4 75.3 33.4 10.5 4. and all of the northeastern states have substantial proportions of scheduled tribes.4 0. In four northeastern states (Arunachal Pradesh.1 Hindu 82.7 61.4 4.9 71.4 30.6 23.3 82.5 77.0 65.3 60.3 17.8 6.4 88. Uttar Pradesh.2 74.4 9.0 1.3 87.0 62. No.1 Muslim 11.1 8.7 88.4 43.3 7.7 1. or ‘other’. Meghalaya. ages 13-49.5 20.9 96.9 7.4 5.2 1.7 8.8 68.2 39.4 88.8 86.6 19.1 16.0 28.8 59.9 * 5.1 1.9 5.4 63.5 4.8 14.7 56. Madhya Pradesh.1 71.0 28.5 11.4 64.9 3.1 5.7 92.3 1.8 49.5 59.9 3.2 * * 0.6 2.6 4.9 Urban residence 26 92 26 10 34 28 20 22 20 15 15 27 15 12 32 19 49 21 20 50 35 41 26 33 28 35 Illiterate 63.6 34.2 2.7 0. Christians constitute sizable minorities.6 16. Gujarat.8 9.9 82.1 0.4 79.8 2. 1992–93 Religion of household head State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Arunachal Pradesh Assam Manipur Meghalaya Mizoram Nagaland Tripura West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu J & K—Jammu and Kashmir * Less than 0.6 82.3 80.7 * 4.4 92.0 89.8 National Family Health Survey Subject Reports.9 10.3 23. .0 41.8 76.1 96.3 50.0 50.3 4.3 58.1 82.7 4.6 5.8 79.4 50.6 70.1 * 0.4 96.3 47. (Scheduled castes and tribes are groups that the Indian Government identifies as socially and economically backward and in need of special protection from social injustice and exploitation.7 86. and almost half report their religion as animism.2 68.8 8.4 2.7 52.3 2.1 95.9 2.0 9.5 8.0 82.3 10.9 97.6 51.3 NonSC/ST 78.1 2.4 2.0 11.7 3.5 1.4 Other 6.6 69.6 66.5 northeastern states.3 27.5 1. and the southern states other than Kerala. Scheduled castes constitute substantial proportions of the population (between 12 and 30 percent) in all of the northern states except Delhi.8 77.6 86.2 5.2 31.3 1.6 0.3 94.05 percent Caste/tribe of household head Scheduled Scheduled caste (SC) tribe (ST) 12.
5 2.7 4.4 23.8 49. Bihar.0 2.9 2.6 2. states with high levels of fertility have a young mean age at marriage and low contraceptive prevalence.4 73.6 42.5 2.6 130.9 89. Maharashtra.6 2.7 42.6 60.3 141.1 Singulate Contraceptive Infant Mortality mean age prevalence mortality under at marriage rate rate age 5 20.1 47.6 2.0 99. Table 2.5 State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Arunachal Pradesh Assam Manipur Meghalaya Mizoram Nagaland Tripura West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu and Nagaland).0 88.6 17. and Arunachal Pradesh.1 75.5 2.7 109.7 58.7 104.2 3.9 4.1 63.9 22.3 83.3 18.6 4.4 18.4 23.6 85.9 20.8 3.3 49. showing selected demographic characteristics of the 19 states.9 3.0 21.4 3.8 45. The singulate mean age at marriage shows that Indian women marry at age 20.0.8 78. 1992–93 Total Mean ideal fertility number rate of children 3.4 20.6 38.7 69.1 children) in two states.6 2.0 21.4 65.7 3. The total fertility rate is below replacement level (2.3 57.0 20.9 2.6 4.8 4.7 2.0 142.3 3.1 19.5 2. 3 Table 2 Selected demographic characteristics. Exceptions are found in West Bengal.4 3.1 20.4 49. No.5 40.5 2.2 87.7 31.9 104.3 49.1 36.0 2.3 127.4 2.9 4.4 64.5.0 3.3 91.0 3. Only Himachal Pradesh and Tamil Nadu have a mean ideal number below 2.7 53.8 31.0 49.3 2.3 55.2 20.6 2.7 1.0 102.3 3. on average.1 3.2 75.2 112.6 25.1 98.9 68.8 34.5 19.7 2.3 53.0 56. where a combination of high contraceptive preva- .3 40.8 36. Uttar Pradesh.1 68.5 to 3.7 19.3 4.2 22.2 19.4 3.7 47.9 18.1 59.0 4.0 2.3 72.2 21.8 13.3 32.0 86. The mean ideal number of children reported by women in most states falls within a narrow range of 2. Fertility is particularly high in Haryana.0 20.7 50. indicates that in most states the total fertility rate ranges between 2 and 4 children per woman.0 3. Goa and Kerala. scheduled tribes constitute large majorities of the population.2 25. by state: India.9 2.0 70.5 70.4 21.9 3.0 2.1 18.9 National Family Health Survey Subject Reports.3 20.7 21.5 65.7 2.8 67.6 18.4 53.6 3.6 3.5 131.1 20.9 2.2 61.1 2.4 58.8 2.6 22.8 23.2 14.9 29. and Andhra Pradesh.0 2.4 17.2 99. In general.7 86.7 72.
an exceptionally high age at first marriage and a moderate level of contraceptive prevalence have been important factors in achieving below-replacement fertility. Bihar. and Himachal Pradesh). but the rates in the latter two states seem to be unreliably low. however. Orissa. The preference for sons is particularly high in most of the northern states. Another indicator of son preference is the extent to which the desire to continue childbearing depends on the current sex composition of a couple’s living children. and the treatment of sons and daughters. Kerala. indicating variations in the strength of gender preference. In those cases.10 National Family Health Survey Subject Reports. The percentage varies among states. The infant mortality rate is several times as high in Uttar Pradesh and Orissa as in Goa. For India as a whole. At the high end of the spectrum. A . Assam. more than 10 percent of children die before reaching their fifth birthday in Rajasthan. which have small sample sizes. on the other hand. In the remaining states (except for Delhi. In Kerala and Tamil Nadu. Infant and under-5 mortality rates are estimated from the birth histories of women. women do not care about the sex of about one-fourth of the children in their ideal families. 3 lence and young age at marriage has resulted in lower than average fertility. and some of the small northeastern states. The ideal number of sons exceeds the ideal number of daughters by 20 to 80 percent in all states. The low levels of contraceptive prevalence in Uttar Pradesh and Bihar are striking in a country where large-scale family planning efforts have been under way for more than four decades. and Gujarat. the proportion is 8 percent or less. Uttar Pradesh. Goa and Kerala share the lowest mortality rates with Mizoram and Nagaland. the respondents said that the children’s sex did not matter or that children of either sex would be fine. No. responses related to 7 percent of children in ideal families were of this type (data not shown). Table 3 shows selected indicators of son preference in every state except the small northeastern states. Madhya Pradesh. In Rajasthan. some women did not specify a preferred sex for some (or all) of their children. which include dates of birth for all children and the date of death for each deceased child. fertility and family planning behavior. Bihar. as well as in Uttar Pradesh. Haryana. Tripura. When asked about the sex composition of children in their ideal family. Once again. In South and West India. and Gujarat. EVIDENCE OF SON PREFERENCE We first examine the evidence of son preference from the NFHS data on the stated ideal number of sons and daughters. In Goa. women have a gender preference for nearly all children. the proportion of children in ideal families whose sex was not specified ranges from 12 to 26 percent. Under-5 mortality also varies substantially from one state to another.
06 0.7 1.6 1. This ratio is based on only two of the three diseases in Jammu.3 1.3 1.03 0.11 u 0.16 1.82 0. and Tamil Nadu have weak son pref- . Kerala. According to this measure.3 4.28 1.30 1.9 4. Except for Karnataka.9 2.01 1.4 1.3 1.10 0.90 0.07 1.4 1.8 2. Severely under-nourished children are more than 3 standard deviations below the median height-for-age of an international reference population. or acute respiratory infection.4 1.27 1.9 3. Diarrhea.7 1.9 1.17 0.0 9.86 0.09 1.97 1.3 2.87 0.10 1. 3 Table 3 Selected indicators of son preference.91 1.3 2.07 u 0.6 2.0 8. 1992–93 Proportion Median Prevalence Proportion Proportion Excess Ratio wanting Proportion Proportion duration of three taken severely female ideal no.99 1.93 0.97 1. Punjab.08 0.3 1.3 1. by state: India.7 1.83 0.0 5.04 1.5 2. son preference is strong in the northern states and is especially strong in Haryana and Rajasthan. c.24 1. Rajasthan stands out as having very strong son preference according to this measure.90 0.86 0.4 1.07 1.98 1. 1.87 0.95 0.2 1.94 1.95 1. women with two sons are less likely to want to have more children than are women with two daughters in every state.11 National Family Health Survey Subject Reports.7 1.16 1.07 1.84 0.04 0.80 1. D—daughters u—Estimates are unavailable because children’s height was not measured.07 1.00 1.6 1. women with two sons are more likely to use contraception than are women with two daughters.11 1.05 0.02 1.48 u 0.07 1.1 1.2 1.12 1.childhood to health underchild sons/ children contraception vaccinated feeding diseases facility nourished mortality (S/D)b (D/S)c (%) daughters (2S/2D) (2S/2D) (S/D) (S/D) (S/D)a INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—sons. no more using fully breast.09 1.13 1.85 1.24 1.90 0. In every state.90 0. Son preference can also affect family planning behavior.2 5. or acute respiratory infection b.20 1.08 1.11 0.99 1.07 1.3 1.1 3.8 2.05 1.97 0.08 u 1.4 1.11 1.24 1. fever.1 1. states in the southern region exhibit a low level of son preference.42 1.2 1.97 1.98 1.2 2.87 0. Andhra Pradesh.98 0.01 1. a.5 1.6 1.81 0.01 1.13 1.92 0.7 2. fever.19 1.2 2.5 1.07 0.6 1.98 1.21 1. No. Among children ill with diarrhea.8 1.27 1.00 0.10 1.7 1.3 1.9 2.7 1.3 1.34 u 43 56 135 44 69 81 59 22 70 55 45 63 13 11 42 24 28 30 –6 –20 simplified measure is used in Table 3 to compare second-parity women with two sons to those with two daughters.9 1.19 1.6 3.02 1.14 0.05 1.90 0. whereas Goa.91 0.95 0.7 2.6 1.32 1.91 0.2 1.08 1.13 1.14 1. and West Bengal because of the small number of children who were ill with the third disease.96 1.2 2.8 2.5 1.4 1. Not surprisingly.
Each indicator is defined as a ratio that has a value exceeding 1. Sen and Sengupta 1983). No. The first measure shows the extent of vaccination coverage of young children (ages 12–23 months). tetanus. On the basis of these three fertility-related measures. Boys are breastfed for a slightly longer period of time than girls in India as a whole.0 if boys receive more favorable treatment than girls and a value of less than 1. Although the intent of the parents may not always be to provide less adequate nutrition to daughters by weaning them earlier. and acute respiratory infection—in the two weeks before the survey. according to this measure. Poor feeding practices and frequent episodes of disease often result in the malnourishment of children. Madhya Pradesh. Girls are consistently less likely than boys to be reported as ill with any of these conditions. poliomyelitis. and Madhya Pradesh. India has one of the most serious problems of malnourishment of any country in the world (Sommerfelt and Stewart 1994). Das Gupta 1987.12 National Family Health Survey Subject Reports. possibly because their illnesses are given less serious attention by their parents. One reason for the shorter period of breastfeeding for girls is the parents’ desire to have another child sooner after the birth of a girl than after the birth of a boy in the hope of having a boy for the next birth. Male children are slightly more likely than female children to be fully vaccinated in India overall. the advantage for boys is observed in 14 of the 19 states. The NFHS estimated both the prevalence and treatment of three childhood diseases—diarrhea. and the male advantage is evident in all but five states. and when they become ill they are less likely to be taken to a health facility or health provider for treatment. 3 erence. particularly where medical care and feeding practices are concerned (Basu 1989. Once again. and measles). Rajasthan.0 if girls receive more favorable treatment than boys. Table 3 shows several indicators of the relative treatment of daughters and sons. Table 3 shows estimates of the ratio of female-to-male chronic undernutrition (stunting) among children under four years of age. Children are defined as being fully vaccinated if they have received all of the recommended vaccinations against six childhood diseases (tuberculosis. Kishor 1995. fever. we infer that son preference is strong in the states in the northern and central regions (except Delhi) and weak in Goa and most of the southern states. In fact. and West Bengal. Bihar. Orissa. whooping cough. The duration of breastfeeding is much shorter for girls than for boys in Haryana. the effect is the same. are more than 3 standard deviations below the median height- . Children who are severely undernourished. Girls face the most discrimination in a band of states from Punjab and Rajasthan in the West to Bihar and Orissa in the East. diphtheria. The strong tradition of son preference and the low status of women in India together produce an atmosphere in which female children may not be treated as well as male children in many cases. Discrimination against girls in medical treatment is particularly pronounced in the contiguous states of Rajasthan.
thereby biasing the indicators that are based on surviving children. 3 for-age of an international reference population (Dibley. This high level of excess female child mortality compares with an international average of only 2–3 percent. In every other state.5 times as high as the rate for boys. It should be pointed out. table 8. Figure 2 and the last column of Table 3 show the percentage by which female child mortality exceeds male child mortality in each state. and so the ratio for that state is based on small percentages. Preston 1990. Kerala and Tamil Nadu are the only two states in which child mortality is higher for males than for females. and Kerala. Dibley.000 births). and eastern India. Some of the worst cases of discrimination may have resulted in death during early childhood. Tabutin and Willems 1995). the mortality rate for India as a whole is slightly higher for males (89 per 1. but that result is entirely attributable to much higher neonatal mortality rates for males (see Figure 1). but it is at least moderately strong in all the states of northern. Child mortality (ages 1–4) is 43 percent higher for females (42. Rutstein. Girls are also somewhat disadvantaged in Punjab. Moreover. however. 1987. although it is relatively weak in Goa and South India (except for Karnataka). female child mortality exceeds male child mortality by at least 11 percent. some of the measures of nutritional status show that girls are actually favored. central.000 births) than for females (84 per 1. In the post-neonatal period. The disadvantage for girls is particularly notable in all the states in the northern and eastern regions. No. age-specific mortality rates in 1991–92 were higher for females than for males in every five-year age group from 0–4 through 30–34 (IIPS 1995. Karnataka. Pebley and Amin 1991. and Bicego 1994). Girls are much more likely than boys to be severely stunted in Haryana.000) than for males (29. In Haryana. based on comparable information from nearly 50 Demographic and Health Surveys around the world (Bicego and Ahmad 1996. The measures of medical care and nutritional status should be interpreted cautiously. Assam and Goa are the only states in which boys are considerably more likely to be severely stunted than girls. Son preference is especially strong in Haryana. et al. the indicators in Table 3 show that son preference is evident throughout India. In summary. and Maharashtra.13 National Family Health Survey Subject Reports. making the results inconsistent for those states. and Madhya Pradesh. Rajasthan. female mortality in South Asia is typically higher than male mortality throughout childhood and the early childbearing years (Ghosh 1987. girls experience slightly higher mortality than boys.4 per 1. girls face a child mortality rate nearly 2. Rajasthan. 1987). Female children are unusually disadvantaged in the next four years of life. In contrast to the usual pattern of higher male than female mortality at almost all ages. During the first year of life. et al. Sullivan. Goldsby. In the NFHS data. and in eight states the excess female child mortality is more than 50 percent. in both Karnataka and Kerala. .3).0 per 1. Staehling. that Kerala has by far the lowest level of severe stunting of any state.000).
1992–93 Son preference is comparatively weak. we estimate the extent to which the probability of having another child within five years depends upon the sex composition of older siblings. but still substantial.000 births 50 40 30 20 10 0 <1 <1 month 1–11 months 1-11 months 1–4 years 1-4 years Males male Females female Figure 1 Infant and child mortality during the five-year period preceding the survey. and how that relationship varies across socioeconomic and religious groups. Bihar. we examine the effects of son preference on fertility by estimating how the sex composition of surviving children affects a woman’s probability of having another child within five years. We compare the monthly probabilities of having an additional child for women whose living children have a given sex composition with the probabilities for a reference group. where fertility is low relative to the level of son preference.14 National Family Health Survey Subject Reports. and those with the weakest. namely her urban/rural residence. Our analysis is limited to parity progression during the period from 1980 until the time of the survey among women with two or three surviving children (referred to as parities 2 and 3 hereafter). The monthly probabilities . and we examine whether the relationship between family composition and parity progression depends upon the socioeconomic characteristics of a woman. The most prominent exceptions are Delhi. where fertility is high relative to the level of son preference. In the next section we examine how son preference is reflected in individuals’ fertility behavior in each state. by sex: India. 3 70 60 Deaths per 1. and Assam. and religion. low fertility. Specifically. PARITY PROGRESSION Using life-table methods and hazard models. in western and southern India. as well as in Assam. No. and Punjab. The degree of son preference in a state corresponds closely to the level of fertility there: those states with the strongest son preference tend to have high fertility. literacy.
these factors are mother’s place of residence. 3 India INDIA Delhi Delhi Haryana Haryana Himachal Pradesh Himachal Pradesh Jammu Region of J & K Jammu Region of R & Punjab Punjab Rajast han Rajasthan Madhya Pradesh Madhya Pradesh Uttar Pradesh Uttar Pradesh Bihar Bihar Oriss a Orissa West Bengal West Bengal Assam Assam Goa Goa Gujarat Gujarat Maharashtra Maharas htra Andhra Pradesh Andhra Pradesh Karnataka Karnatak a Kerala Kerala Tamil Nadu Tamil Nadu -50 -50 0 0 50 50 100 100 150 150 percentage excess female deaths Figure 2 Percentage of excess female mortality at ages 1–4 during the five-year period preceding the survey: India. For other groups the relative risks estimated by the hazard models are applied to the probabilities of the reference group’s estimated parity progression rates. year of birth of the last child.15 National Family Health Survey Subject Reports. Among women with two or three surviving children. The analysis model includes additional factors likely to have a strong effect on fertility. and the experience of child death. Choe. We also examine how the effects of son preference differ by these characteristics by including the interaction terms between sex composition of living children and socioeconomic factors. Parity progression by number of sons The effects of the number of sons on parity progression are summarized in Tables 4 and 5. 1992–93 are summarized to estimate the probability of having an additional birth within five years. religion of the household head. In the net-effects model. and Roy 1996). Table 4 shows the adjusted percentages of women who gave birth to the next . the reference groups are women with two sons. mother’s literacy. mother’s age at the birth of the last child. For the reference group the probability of having another child within five years is estimated by a life-table method applied to women in that group. because those women have been observed to have the lowest probabilities of having additional children in India (Arnold. No.
2* 67.6 57.1 51.2* 83. with p<0.63 1.6* 72.9* 76.5 80.1 52.6* 83. Women are least likely to have additional children if they have two sons and most likely to have additional children if they have no sons. maternal age.3* 77.3* 89.4* 81.7 1 sona 72.2* 72. mother’s literacy.9* 74.39 1.0 47.4 78. With few exceptions.6 67.2* 64. In five states (Delhi.33 1.4 0 sonsb 80.5* 54.0* 56.2 60.4 67. No.7* 76.37 1.3 64.14 1. and Tamil Nadu).5* 83.5* 81.7 74.3 71.7 63. S—son/s.2* 85. the differences in parity progression between women with one son and women with two sons are not statistically significant.09 1. classified by the number of surviving sons and whether the differences between women with two sons and one son.9 75.5* 83.4* 81.2* 82.6* 80.5 68.16 National Family Health Survey Subject Reports. year of birth of the second child.56 1.30 Note: Adjustments were made by using proportional hazard models controlling for the effects of child mortality.36 1. are statistically significant at the 5 percent level.29 1.1* Ratio (0S/2S) 1.23 1.5 76.9* 84. and between women with two sons and no sons. a. by number of sons and state: India.0 65.5* 48.38 1. The asterisks (*) in this column indicate that the underlying hazard model coefficient associated with no sons (compared with two sons) is statistically significant.14 1.1* 89.2 61.8 35. Goa.0 54.0* 50. In those states.03 1. A similar pattern is observed in India as a whole and in most of the states. with p<0.7* 83. The asterisks (*) in this column indicate that the underlying hazard model coefficient associated with one son (compared with two sons) is statistically significant.05.4* 67.21 1. the differences in parity progression between women with two sons and other women are statistically significant. mother’s residence. Andhra Pradesh. child within five years. 3 Table 4 Adjusted parity progression ratios for women with two surviving children.16 1.7 56.5* 65.1* 89. Karnataka. and religion of head of household.18 1.23 1.05. women seem to think that it is .7* 80.5* 82.0 68. 1992–93 Percentage having next child within 5 years State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu 2 sons 66.7 48.9* 73.8 76.31 1.7* 85. b.13 1.
97 1.5 65.05.0 47.28 1.26 1.0* 73.9* 61. important to have at least one son.1* 79. The last column of Table 4 shows an index of son preference that is the ratio of the parity progression ratios of women with two daughters and women with two . The asterisks (*) in this column indicate that the underlying hazard model coefficient associated with no sons (compared to two sons) is statistically significant.1 26.4 37.51 1. b. Adjusted parity progression ratios for women with three surviving children.1 28. maternal age.8* 54.8* 72.1* 56.5 67. by number of sons and state: India. mother’s literacy.1* 59.47 1.5 48. the sex of the other child does not affect future childbearing decisions.40 1.70 2.54 1.5 65.3* 84. Thus.1 47.32 1.5 22.8 38. with p<0.4* 69.54 1.8* 69. 1992–93 Percentage having next child within 5 years State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu 3 sonsa 55.7 40.94 1. but once they have had one son.33 2.10 1.00 2. parity progression is lowest among women with one child of each sex.1 71.3 40.8 47.4 52.7* 76. S—son/s a.2* Ratio (0S/3S) 1.8* 85.4 38.3* 48.1* 75.4* 56.11 1.10 1.53 2.56 1.8 51.7* 83.35 1.0* 74.8* 53.6 31.21 1.05.2* 50. mother’s residence.58 1.6 56.05.4* 72.8* 28.15 1.57 1.8 72.1* 60.1 51.32 1.3* 83.7* 80.9 78.2* 36.2* 31.4 38. women with no sons and women with no daughters have higher parity progression.41 1.3 35.2* 2 sons 50.4* 64. and religion of head of household.0 34. The asterisks (*) in this column indicate that the underlying hazard model coefficient associated with one son (compared to two sons) is statistically significant.61 1. with p<0.6* 74. women in Kerala exhibit no preference for sons or daughters and prefer to have one child of each sex.2* 61.18 1.8 39.90 2.8* 83.2* 60.4* 63.7* 48.3* 87. at similar levels. No.29 1.1* 85.1* 76.17 National Family Health Survey Subject Reports.23 1.31 1. The asterisks (*) in this column indicate that the underlying hazard model coefficient associated with three sons (compared to two sons) is statistically significant.7 1 sonb 65.09 Note: Adjustments were made by using proportional hazard models controlling for the effects of child mortality.8 54.15 1.8* 88.1* 63. In Kerala. year of birth of the third child.9 45.63 1.9* 74.5* 48.2 56.12 1.0* 64.37 1. c.0* 81.3 31.9 39.9 50.71 1.66 1.7* 0 sonsc 78.5 51.12 Ratio (1S/2S) 1.13 1. with p<0. 3 Table 5.
The first compares women with three daughters to women with three sons. Strong son preference is also noted in most of the states of the northern region. and Maharashtra stand out as having very strong son preference among third-parity women. and Tamil Nadu to more than 80 percent in Haryana. 3 sons. however. Assam. The first ratio may not be an ideal measure of son preference. Himachal Pradesh. among women with three children. Punjab. Assam. women with one surviving son are much more likely to have additional children than are women with two sons. Punjab. According to this index. for two reasons. and the second compares women with one son and two daughters to women with two sons and one daughter. Table 5 shows the adjusted percentages of women with three surviving children who gave birth to the next child within five years. Karnataka. Himachal Pradesh. No. Haryana. The parity progression rate is significantly higher among women with three sons and no daughters than among women with two sons and one daughter in only five states—Himachal Pradesh. Furthermore. that although the first measure is almost always higher than the second. and Tamil Nadu. namely Uttar Pradesh. States with high levels of fertility show low to moderate degrees of son preference. and the states of the western region. First. Second. Jammu. The effect of son preference is weakest in states with widely differing overall levels of fertility. Bihar. Madhya Pradesh. statistically significant in most states. and Tamil Nadu. women with two sons and one daughter are least likely to have a fourth child. exhibit relatively weak son preference in fertility decisions. the proportion with children all of the same sex is quite small. The second ratio does not have these problems but is problematic because both groups of women have sons. Kerala. Himachal Pradesh and Punjab stand out as having the highest effect of son preference on fertility behavior. however. In those states. women show a preference for having at least one daughter after having two or more sons.18 National Family Health Survey Subject Reports. classified by the number of surviving sons. Rajasthan and Uttar Pradesh. Orissa and West Bengal in the East. . Uttar Pradesh. Gujarat. Women with no sons have very high parity progression ratios in all states. and most women with two children go on to have more children regardless of the sex composition of their living children. Table 5 also shows two measures of parity progression ratios. ranging from about 50 percent in Goa. which show high levels of son preference as measured by fertility preferences and child mortality. and the estimated parity progression ratios may be unstable. parity progression among women whose children are all of the same sex is affected by the effort to have children of both sexes as well as by son preference. having one surviving son does not seem to satisfy most Indian women. The difference in parity progression among women with two sons and three sons is not. In India as a whole and in almost every state. In all states except Kerala. Both those states have high fertility levels. Rajasthan. Assam. and Karnataka. the two ratios give substantially the same results. Table 5 shows.
7 7. For all parities combined. Excess percentages of births of order 3.6 10.1 14.0 8.0 14.9 percent higher at parity 2 and 17.7 9. Arnold (1985) has developed a measure to quantify the effect of gender preference on fertility and family planning.7 38. Estimated excess percentages of births due to gender preference computed in this way are shown in Table 6.5 7..4 percent.3 12.9 4.e.3 16.0 12.1 Fourth child 17.3 16.1 7.4 7.2 18.3 All children 8.9 19.1 4.3 60.6 4.8 15. and all births due to parents’ gender preferences for children.9 6.7 20.1 8. The measure assumes that in the absence of gender preference.5 16.9 15. 3 Table 6.2 7.3 percent higher at parity 3 than it would be if there were no gender preference and women’s parity progression ratios did not depend upon the number of surviving sons. by state: India.2 26. order 4.1 34.4 8.4 5. Overall.3 21.6 8.19 National Family Health Survey Subject Reports.1 10.0 28.7 34.7 Table 6 shows another set of indicators of gender preference for children: the excess percentage of births attributable to gender preference.3 7. No. This measure is designed particularly for use in populations in which gender preferences are fairly homogeneous.8 4. those with the lowest parity progression ratio).9 24.1 20. the parity progression ratio of women in India is 8.7 112. The decrease in parity progression if .9 8. the excess percentage of births due to gender preference is estimated to be 8.6 6.9 8. all groups of women at a given parity would proceed to have an additional child at the same rate as women at that same parity with the most desirable sex composition of children (i.7 10.0 4.2 45.8 16.9 58.4 14. 1992–93 Excess percentages of births State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu Third child 8.0 8.6 6.3 14.0 11. estimated from adjusted parity progression ratios.4 16.4 8.4 12.
the effect of gender preference depends upon some characteristics of the states other than the level of fertility alone. In the following sections we examine in more detail the role of urban versus rural residence. No.5 total fertility rate Total fertility rate 3.0 4 4. where fertility is low. Residence and son preference Does the effect of son preference on fertility depend upon whether women live in urban or rural areas? Tables 7 through 9 show parity progression ratios and . A particularly high level of gender preference is observed in Haryana. 3 25 20 Excess percentage of births 15 10 5 0 1 . within India.5 4 . Figure 3.39 to 3. Himachal Pradesh.0 5 Figure 3 Relationship between total fertility and excess births due to gender preferences for children: 19 Indian states.20 National Family Health Survey Subject Reports. a scatter plot of the relationship between total fertility rates and the excess percentage of births due to gender preference.5 4.5 1.0 2 2. Although moderate. the effect of gender preference on fertility is low or moderate.5 5. indicates that in states with very high and very low levels of fertility.13 children per woman.5 3 . but not all. 1992–93 gender preference were absent is estimated to be approximately equivalent to reducing the total fertility rate from 3.5 2. even a decrease of this magnitude would have a noticeable impact on India’s rate of population growth. The effect of gender preference on fertility is moderately high in states in the western region. and Punjab according to this measure. It is relatively low in states in the southern region. women’s literacy. A large effect of gender preference on fertility is observed in some.5 2 . states with an intermediate level of fertility. and in states where fertility levels are high. and religion in determining the effect of gender preference on fertility. The pattern suggests that.0 3 3.
3 74.17* 1.9 75.7 75.6 57.6 75.15 1.4 58.15 1.27* 1.6 53.25 1.8 87.1 81. and by state: India.4 56.0 87.3 78.28* 1.2 63.9 80.7 67.0 82.73* 1.4 48.12* 1. 1992–93 Urban residence 2 sons 62.9 73.12* 1. parity progression ratios are somewhat lower in urban areas than in rural areas.24* 1.3 56.4 75.28* 1.1 72.9 75.2 86. however.9 81.8 56. by residence and whether they have two or no surviving sons.51* 1. The last column of Table 7 indicates whether the coefficient associated with the interaction term between the number of sons (two versus none) and urban/rural residence is statistically significant or not.18* 1.5 0 sons 81.1 64.32* 1. Parity progression ratios for women with no surviving sons are similar in urban and rural areas.3 83.48* 1.9 43.2 68.5 75.1 86. in terms of the ratio of parity progression between women with two sons and no sons.0 91. is statistically significant.8 81. at the state level only Gujarat shows a statistically significant difference in son preference.7 51. If it is statistically significant.7 87.36* 1.6 79.0 62.3 0 sons 80.49* 1.8 86.5 67.4 68.09* 1.6 51.29 1.0 62.1 59.1 80.2 85.10 1.9 66.4 90.1 43.8 46.0 75.30* 1.9 83. Although a statistically significant urban/rural difference is observed for India as a whole.1 65.4 68.21 National Family Health Survey Subject Reports.7 83.4 65.0 Ratio (0S/2S) 1.2 70.4 61.05 1.0 Ratio (0S/2S) 1.6 44.24* 1.2 75.6 50.4 80.27* Rural residence 2 sons 68.34* 1. 3 Table 7 Adjusted parity progression ratios for women with two surviving children.7 54.97 1.37* 1.78* 1.8 82.1 72.23* 0.20 1.55* 1. the urban/rural difference in son preference. Among women with two surviving sons.9 75.05 related statistics for women classified by urban/rural residence and the number of surviving sons.9 84. .3 65.3 90.19* 1.25* 1.1 41.13* 1.7 76.50* 1.0 79.33* 1. Table 7 shows that the fertility effect of son preference is slightly higher in urban areas than in rural areas at parity 2.4 55.33* Interaction term significant? Yes No No No No No No No No No No No No No Yes No No No No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s * p <0.7 58.19* 1.3 64.7 82.27* 1. No.3 88.43* 1.
4 50. The same is true for urban women in most states.6 45.0 53.7 82. At the state level the effects of son preference on parity progression are virtually the same among urban and rural women. 1992–93 Urban residence Three sons 52.2 46.47* 1.11 1.9 29. In the comparison of women with three sons and those with no sons (Table 8).61 1.2 64.6 27.0 Ratio (0S/3S) 1.1 72. 3 Table 8 Adjusted parity progression ratios for women with three surviving children.1 51.9 70.39* 1.5 33.8 82.77* 1.22 National Family Health Survey Subject Reports.33* 1.9 51.8 80.50* 1.0 87.42* 1. and by state: India.6 65.11* 1.39* 1.9 47.2 56.5 u 41.7 56.5 55.7 62.5 85.5 76.7 45.1 39.75* 1.8 55.8 53.6 84.0 u 85.6 53.70* 2.61* 1.4 86.9 78.97* 1.6 50.02* 1.0 No sons 75.7 48. but only about half of urban women with three daughters continue .8 87.77* 2.4 40. * p <0.16 Rural residence Three sons 56.5 75.6 85.90* 2.9 39.10 Interaction term significant? No u No No No No No No Yes No No Yes No No No No No No No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s u—Estimates are unavailable because of small samples.42* 1.1 49.92 1.1 84.70* 2.9 37.0 76.12 1.3 66. with few exceptions.13* 1.9 81.49 1.50* 1.90* 0.90 1.68 1.7 58.5 93.05* 1.5 79.09* 1.60* 1.9 Ratio (0S/3S) 1. Among women with three children in Haryana.0 49. by residence and whether they have three or no surviving sons.68* 1. son preference is more important among rural women in West Bengal and less important among rural women in Uttar Pradesh.3 60.0 75.4 76.9 33.40* u 2.4 75.4 51.7 28.5 86.4 80.61* 1.6 58. a large majority of rural women with three daughters continue childbearing within five years.3 78.11 1. the son-preference index measured by the ratio of parity progression ratios for women with one son and those with two sons is significantly higher among rural women than urban women.3 34.1 87.30* 1.7 51.2 70. Haryana shows a pattern different from all other states in the comparison of women with two sons and those with one son (Table 9).46* 1.5 44.1 74.5 51.05 At parity 3 the place of residence does not matter at all at the national level (Tables 8 and 9).5 43.57* 1.0 77. No.4 40.7 No sons 79. In every state except Kerala.5 57.8 87.
54* 1.62* 1.4 47.5 60.04* 1.2 75.29* 1.1 38.6 58.37* 1.26* 1.16 1.0 66.1 43. Among women with two surviving children.3 51. The difference in the index of son preference between literate and illiterate women is statisti- .7 81.6 54.1 52.9 52. and by state: India.54* 1.5 51.68* 2.6 61.6 30. Education and son preference Tables 10 through 12 show how the effect of son preference on fertility differs by women’s literacy.5 29.5 25.9 32.95* 1. A majority of urban women in most states curtail their childbearing after they have had three sons.5 50.8 63. which means that the effect of son preference on fertility is greater among literate than among illiterate women.38* 2.6 47. where 60 to 70 percent of these women have another child within five years.0 34.0 1 son 66.1 75.0 1 son 62.6 45.12* 1.51* 1.09 1.5 80.8 28. and Tamil Nadu. by residence and whether they have one or two surviving sons. No.32 1. 1992–93 Urban residence 2 sons 48.2 72.58* 1.12* 1.6 41.3 35.05 1.2 60.6 68.1 54.11* 1.8 23.5 67. Goa.7 74.5 25.7 49.7 29.8 74.7 45. Kerala.7 63.09* 1.34* 1.3 37.29* 1.2 61.5 59.8 29.9 52. But the difference is much larger if women have two surviving sons.1 33.5 68.12* 2.8 46.8 44.37* 2.8 61.5 57.8 60.7 35. Bihar.7 66.14* 1.2 76.8 37.7 79.7 64.08* Interaction term significant? No No Yes No No No No No No No No No No No No No No No No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s * p <0.9 56.1 45.8 35.33* 1. Andhra Pradesh.5 21.18* 1.05 childbearing in West Bengal.31* 1.6 70.15* 1.78 1. 3 Table 9 Adjusted parity progression ratios for women with three surviving children.00* 1.7 18.16* 1.1 33.7 43.05 Rural residence 2 sons 51.05* 1.4 70.45* 1.9 53.8 41.6 72.7 72.7 48.26 2.4 62.14 2.3 39.8 Ratio (1S/2S) 1.23 National Family Health Survey Subject Reports. and Assam. Uttar Pradesh.67* 0. parity progression is lower among literate women than among illiterate women.35* 1.7 Ratio (1S/2S) 1. Notable exceptions are Madhya Pradesh.1 62.
1 82. in seven states—Haryana.2 48.0 83. No.02 1.38* 1. and Maharashtra— literate women with no sons are more than twice as likely to continue childbearing as literate women with three sons. in every case the effect is greater for literate than for illiterate women. In all of those cases.9 62.19 1.1 52. Himachal Pradesh.99 1. Among women with three surviving children.69* 1.39* 1.33* 1.46* 1.24* 2 sons 72.51* 1.9 80.8 62.1 79.9 60.40* 1.40* 1.0 44.10* 1.17* 1.6 86.2 46.7 86.1 57.05 cally significant for India at large and for six of the 19 states (Table 10).8 77.5 61.7 Ratio (0S/2S) 1.36* 1.5 65. When we examine the ratios of parity progression among women with one son and two sons (Table 12).4 83.9 57.38* Interaction term significant? Yes No No Yes Yes No No Yes No No No Yes No No Yes No No Yes No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s * p <0.9 85.0 78.10 1. Rajasthan.71* 1.7 66.5 80.3 73.42* 1.4 70.6 70.48* 1.1 88.3 78.9 69.2 86.6 66.15* 1. Once again.23* 1.6 53.2 87.8 56.12* 1.31* 1.4 65. .2 83. 1992–93 Literate 2 sons 57.7 60.4 90. The difference is statistically significant for seven of the 19 states as well as for India as a whole (Table 11).3 66.04 1.42* 1. Gujarat.6 76.3 66. where the effect of son preference is stronger for illiterate women.5 65.5 86.1 48.28* 1.5 90.2 43.2 82. and by state: India.1 45. son preference has a greater effect among literate women. Punjab.36* 1.9 77.26* 1. we find the interaction with literacy is significant only in Punjab.4 69.4 62.7 48. the ratios of parity progression among women with three sons and those with no sons show large differences by women’s literacy.11* 1. Madhya Pradesh.5 84. In fact.4 Illiterate 0 sons 83.09 1.5 74.6 45.7 61.4 81.9 82.18* 1.4 72.2 70. by women’s literacy and whether they have two or no surviving sons.1 0 sons 76.0 56.7 72.05 1.3 72.9 76.15* 1.7 87.83* 1.6 82.6 71.6 Ratio (0S/2S) 1. 3 Table 10 Adjusted parity progression ratios for women with two surviving children.0 87.9 77.39* 1.24* 1.0 83.5 83.14* 0.24 National Family Health Survey Subject Reports.2 74.29* 1.6 87.32* 1.3 67.7 81. but there are no states where the differential is that large for illiterate women.3 85.
72* 1.78* 1.6 88.7 Illiterate 0 sons 79. for Muslims the ratio of parity progression among women with two sons to women with no sons is not statistically significant in most states.86* 1.2 51. son preference has some effect on fertility for Hindus.25* 1.45* 2. where women from other religions (mainly Sikhs) have .8 72.35 1.9 Ratio (0S/3S) 1.7 78.83* 2.0 88.7 65.0 70.6 36.85* 1.6 54.2 Ratio (0S/3S) 1.44* 1. and by state: India.47* 1.4 58.4 27.0 83.6 49.1 47.05 3 sons 62.2 30.0 69.0 29.6 76.8 87.2 48.73* 3.7 92.5 85.0 58.50 2.12* 1.7 26.06* 1.1 46.78* 1.57 1.2 55.6 73.7 35.08* 2.5 40.9 84.09* 2.41* 1.9 91.7 82.8 31. by women’s literacy and whether they have three or no surviving sons.70* 1. 3 Table 11 Adjusted parity progression ratios for women with three surviving children.3 87.85* 1.1 46.19* 1. 1992–93 Literate 3 sons 43.1 38.9 0 sons 75. In every state except Kerala. In fact.8 53.0 27.57 1.3 63.68* 1.2 45.8 29.17 Interaction term significant? Yes No Yes Yes No Yes No Yes No Yes No No No No Yes Yes No No No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s * p<0.28* 1.0 48.05 Religion and son preference Tables 13 through 15 summarize the relationship between the effect of son preference on parity progression and women’s religion. No.28* 2.0 48.7 82. and the effect is statistically significant in 14 states.9 83.0 75.7 88.4 48.9 70.61* 1.9 42.64* 1.6 84.25* 1.5 83.8 83. In general.0 60.9 54. Women who are not Hindus or Muslims often exhibit strong son preference in parity progression in states with sufficiently large samples of these women.9 42.5 70.0 78.25 National Family Health Survey Subject Reports.5 44.8 80.6 49.5 56.4 64.8 67.74* 1.5 83.9 43.2 40.1 46.3 83.27 1.7 75. Most notable is the case of Punjab.06 1.48* 1.32* 1. Only in Jammu and Rajasthan do Muslims exhibit a significant son preference.8 83.9 82.44* 1.24* 1.88* 1.1 60.9 75. the effect of son preference is weaker among Muslims than among Hindus when we consider the parity progression of women with two surviving children (Table 13).61* 1.9 61.7 59.
9 23.4 35.56* 1.65* 1.21* 1.2 71.02 1.30* 1.2 70.29* 1.2 63.0 36. Hindu women without any sons have a much higher propensity to continue childbearing than those with one or more sons.5 26.74* 1. Hindu women with only one son and two daughters are much more likely to have another child than are women with two sons and one daughter (Table 15).4 45.8 31.5 Ratio (1S/2S) 1.48* 1. and by state: India. 1992–93 Literate 2 sons 40.9 49.9 72.10* 1.26* 1.68* 1.0 Illiterate 1 son 71.7 58.9 Ratio (1S/2S) 1.2 32.3 43.1 1 son 54.7 38.3 55.7 58.9 28.52* 1.1 53.8 60.1 41.6 26.4 55. 3 Table 12 Adjusted parity progression ratios of women with three surviving children.9 71.59* 1.8 38.4 37.7 64.9 64.02* 1.5 27. These findings suggest that it will be difficult to eliminate entirely the effect of son preference on fertility in India in the near future as fertility falls to moderate levels in the populous northern Indian states. by women’s literacy and whether they have one or two surviving sons.8 45.2 81.0 61.7 45.2 68.09 0.4 74.6 48.55* 2. In Bihar and Karnataka.2 76.71* 2.43* 1.8 32.05 a very strong preference for sons. No. In summary.19* 1.37* 1. Although Kerala does not show much son preference overall.6 78.2 62.2 63.8 18.2 47.7 32.2 58.4 29.2 71.3 53.1 47.28* 1.10* 1.2 62.4 26.1 33.39 2.6 64.6 53.6 50.9 69.9 63.22* 1.06* 2.4 77.30* 1.8 38.57* 0.0 27. .3 23.2 23.8 47.88 1.7 56.27* 1.6 63.24* Interaction term significant? No No No No No Yes No No No No No No No No No No No No No No State INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s * p<0.6 40.5 58.7 43.3 46.3 76. the effect of son preference on fertility is generally strongest for groups of women who have a moderate level of fertility.26 National Family Health Survey Subject Reports.23 1.35* 1.92* 1.8 68. Another interesting result is the difference in parity progression ratios for women with three children in Kerala (Table 14).14* 1.33* 1.4 41.2 82.19* 1.23* 2.52* 1.88 2 sons 56.3 48.2 74.47* 1.0 58.
5 51.6 77. by religion of the head of household and whether the women have two or no surviving sons.5 u 56.23* 78.26* 0.29* u 1.28* u 1.9 u u 83.9 2 sons INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s u—Estimates are unavailable because of small samples.6 u 88.12 1.5 52.98 1.5 1.6 72.32* 1.8 u 1.9 74.3 62.89 1.6 u u 92.36 1.4 83.14* u 88.30 u u 1.9 75.3 75.10 1.3 u 76.0 u u 64.5 67.8 u 74. 1992–93 Hindu 0 sons 80.1 60. 3 * p<0.72* u u u u u u u 1.9 64.6 89.3 u u 46.9 40.02 1.00 u 1.3 73.0 81.05 63.08 1.3 85.08 1.5 u 1.5 64.2 1.37* 1.1 u 71.32* 1.08 1.6 85.6 1.2 84.36* 1.05 .9 68.8 85.65* 1.4 75.0 81.3 88.8 u 81.3 1.23* 84.9 58.0 75.9 40.9 83.02 0.04 u 1.7 82.8 69.07 1.1 u 69.9 81.8 76. No.4 63.5 1.9 u 89.6 90.3 47.19 Ratio (0S/2S) 2 sons 0 sons Ratio (0S/2S) 2 sons 0 sons Ratio (0S/2S) Muslim Other Interaction term significant? Muslim Yes No u u No u No u Yes No u Yes No u Yes No No No No No Other No u No u No Yes u u u u u u u No u No u u No No State 65.30 1.0 81.2 55.41* u 68.9 u u 59.34* 76.19 1.8 52.1 u u u u u u 55.13* 66.1 85.6 u 75.7 1. and by state: India.1 70. 27 National Family Health Survey Subject Reports.0 84.45* u 1.7 84.7 78.40* 1.2 56.40* 79.19* u 70.23 1.4 86.2 60.6 85.8 u 80.7 1.0 84.9 1.2 1.3 87.01 u u u u u u 38.Table 13 Adjusted parity progression ratios for women with two surviving children.44* u u 1.31* u 1.1 57.14* 1.8 82.6 51.7 74.0 82.0 58.
3 48.9 u 65.64* 1.53* 61.4 74.1 u 55.45* 72.21* 1.7 85.0 u u 1.2 44.60* 80.6 71.01 u 1.06* 2.34* 0.5 61.2 83.3 74.9 61.59* 1.0 58.76 1.1 44. and by state: India.93* u 1.11 74.6 52.18* 53.11 u u u u u u u u u u 1. by religion of the head of household and whether the women have three or no surviving sons.2 77.8 u 90.9 u u 22.0 37.7 1.7 1.1 54.17 84.2 81.0 u 30.21 u u 1. No.2 38.2 86.4 52.4 53.09 u 1.5 83.0 39.42* u u 2.1 80.8 u u 44.9 u 1.55* 1.8 87.30* 1.9 40.6 u u 75.9 41.3 50.2 85.03 2.17* u 79.6 82.1 70.3 47.42* Yes No u No u No No u u u u u u u No u No u u No u Ratio (0S/3S) 3 sons 0 sons Ratio (0S/3S) 3 sons 0 sons Ratio (0S/3S) Muslim Other Muslim Other Interaction term significant? State 53.21* 1.0 u 89.5 71. * p<0.37* u 42.11* u No u u No u No u No No u No No u No No No No Yes u 1.4 u 56.37* 1.8 u 1.0 46.6 96.05 u u 39.7 79.6 1.05 .2 1.3 87.59* 1.1 47.8 u u 62.2 u 1.58* 2.1 75.71* u 2.42 23.19* u 1.0 u 1.3 75.48 1.1 74. 3 North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s u—Estimates are unavailable because of small samples.5 u 74.4 u 1.6 60.2 u 76.9 u 84.9 82.3 63.4 u 95.5 84.5 49.8 72.72* 2.90 u 1.49* u 59.3 54.2 31.06 u u u u u u 1.8 87.58* 1.71* 2.4 u u 81.4 3 sons INDIA National Family Health Survey Subject Reports. 8 1.2 u 68.4 1.28 Table 14 Adjusted parity progression ratios for women with three surviving children.8 88.1 17. 1992–93 Hindu 0 sons 77.
11* u u 0.4 38.8 64.36* 54.1 61.1 21.9 40.12 1.0 58. No.02 u 33.2 u u 77. 3 * p<0.15* 68.3 63.5 0.54* u 2.4 79.6 69.9 79.7 74.7 u u u u u u u 33.6 63.10 u 1.3 u 44.0 68.0 58.6 58.13* u 78.6 u 75. by religion of the head of household and whether the women have one or two surviving sons.2 76.11 48.90 1.0 55.Table 15 Adjusted parity progression ratios for women with three surviving children.4 44.56* 1.32* u 2.5 56.7 52.1 18.00 u 1.4 37.8 68.09 1.8 75.99 u Ratio (1S/2S) 2 sons 1 son Ratio (1S/2S) 2 sons 1 son Ratio (1S/2S) Muslim Other Interaction term significant? Muslim Yes No u u No u No u No Yes No No No u No No No Yes No u Other No u No u No No u u u u u u u No u No u u No u State 48.3 26.2 u 1.8 u 1.6 u 1.8 51.2 36.1 55.4 72.39* 85.86* 2. and by state: India.5 33.81* 1.2 61. 1992–93 Hindu 1 son 64.3 55.20 1.8 50.0 36.0 u 29.0 61.0 u 1.35 u u 1.35* 1.36* 1.2 u u 14.8 47.8 56.23* 1.5 70.45* 1.5 1.19* 1.65* 2.8 u 61.1 u u 62.5 18.6 73.08 u u u u u u 22.4 1.4 79.9 u 1.4 u 84.4 46.5 u 80.36* 2.05 .15* 76.4 37.59* 0.8 u 67.8 79.32 1.45* 1.5 54.32* 69.29* 1.6 75.1 58.3 73.7 27.7 u 77.0 1.8 77.2 2 sons INDIA North Delhi Haryana Himachal Pradesh Jammu Region of J & K Punjab Rajasthan Central Madhya Pradesh Uttar Pradesh East Bihar Orissa West Bengal Northeast Assam West Goa Gujarat Maharashtra South Andhra Pradesh Karnataka Kerala Tamil Nadu S—son/s u—Estimates are unavailable because of small samples.0 u u 68.4 38.33* u 1.9 64.1 1.5 29. 29 National Family Health Survey Subject Reports.29* 1.0 76.99 1.6 1.8 57.14* u 56.86* u u u u u u u 1.2 83.12 u 1.32* 1.0 u u 14.
This rather modest effect. in the face of a strongly articulated preference for sons. The survey finds less evidence of systematic discrimination against young girls with respect to feeding practices or nutritional status. female mortality exceeds male mortality by more than 40 percent. Gujarat. In that age group. No. Kerala. of a desire to have one daughter. as expected. Bihar. Daughters face discrimination in the medical treatment of illnesses and in the administration of vaccinations against serious childhood diseases. whereas the effect is much smaller in others (notably West Bengal and most of the states in the southern region). and personality development. A skewed sex ratio. an interesting pattern emerges.30 National Family Health Survey Subject Reports. In states where fertility is very high or very low. Assam. some (notably Himachal Pradesh. the effect of son preference on fertility is small. If gender preferences for children could be entirely eliminated. Punjab.0 and 3.0 children. both attitudes and behavior are vitally influenced by a long-standing preference for sons that is deeply rooted in the cultural traditions of the society. but much weaker. The most troubling problem evident in the NFHS data is the unusually high excess mortality for girls who are 1–4 years old. effect on fertility in such diverse states as Uttar Pradesh. In India as a whole. There is some evidence. In this regard. Son preference has a significant. . as well. and often at least two sons. and Maharashtra) exhibit very large effects of son preference on fertility. Women express a strong preference for having at least one son. among their children. Our study has focused on the impact of son preference on reproductive behavior. we estimate that the level of fertility in India would fall by about 8 percent. however. In states with intermediate levels of fertility. population policy needs to pay special attention to the possible consequences of high son preference that go beyond the retardation of fertility decline. in turn. but having more than one daughter is generally not considered desirable. the effect of son preference on fertility varies widely. the marriage and labor market. Several states in northern India (particularly Himachal Pradesh and Punjab) exhibit especially strong effects. is due both to a rather weak effect of son preference on fertility in some of the most populous states and to the fact that by sheer biological chance most couples are able to have their minimum desired number of sons without exceeding their overall family-size goals. and Tamil Nadu. Among the states with total fertility rates between 2. Studies in countries with persistent high levels of son preference and very low levels of fertility document that such a combination can result in a skewed sex ratio at birth if women have access to methods of identifying the sex of fetuses and to induced abortions (Park and Cho 1995). has worrisome implications for future population structure. 3 CONCLUSION Our study confirms that a preference for sons is widespread in India and that son preference affects fertility behavior in every part of the country. If the states are arrayed according to their overall level of fertility. which we have found to vary substantially across India’s states.
however. Gender preferences for children: Findings from the Demographic and Health Surveys. Demography 22:280–88. We also gratefully acknowledge computer assistance provided by Lixia Qu and Jonathan Chow of the East-West Center. Son preference in South Asia. from April to June 1996. ACKNOWLEDGMENTS We express our appreciation to the United States Agency for International Development for providing funding for both the National Family Health Survey and the research on which this report is based. Beijing.31 National Family Health Survey Subject Reports. 3 Although son preference in India will be difficult to change dramatically in the short run. it is unlikely that son preference will decline sufficiently to transform fertility behavior in the near future. Sex preference and its demographic and health implications. who was able to conduct some of the research for this report during a fellowship at the Program on Population. 1996. Islamabad. and Roy (1996) have recommended a number of programs that could expedite the transition toward more gender-neutral attitudes and behavior. Paper presented at the International Union for the Scientific Study of Population (IUSSP) Seminar on Comparative Perspectives on Fertility Transition in South Asia. ———. REFERENCES Arnold. Without a basic change in the dowry system and the other traditions that engender a preference for sons. . 17–20 December. particularly with respect to medical care. Measuring the effect of son preference on fertility: The case of Korea. ———. No. 1992. 1997. as well as support for Professor Rangamuthia Mutharayappa. These include expanded efforts to encourage equal education for daughters and sons (such as educational savings accounts for girls and the provision of free midday meals at schools) and education and motivational programs to encourage equal treatment of daughters and sons. Arnold. 11–17 October. East-West Center. ———. 1985. International Family Planning Perspectives 18:93–101. Fred. Paper prepared for the 23rd General Population Conference of the International Union for the Scientific Study of Population (IUSSP). Choe.
Sex preference pattern and its stability in India: 1970–1980. Fred. Honolulu: East-West Center. Madison: University of Wisconsin Press.. Studies in Family Planning 17:193–210. 1996. Rodolfo A. Population and Development Review 13:77–100. Arnold. and T. Selective discrimination against female children in rural Punjab. Population Series. Fred. Honolulu: East-West Center. 1989. ———. 1987. 1988. Comparative Studies. Comparative Studies. Ideal number and sex composition of children in India. Reddy. Basu. Langsten. and Omar B. Demography India 13:108–19. Caldwell. and Pat Caldwell. India. Srinivasan and P. 1978. A simulation model to study the effect of sex preference on current fertility. Ahmad. pp. 472–82. Narayan. 1. Bairagi. Voorburg. Das. Bulatao. the familybuilding process. ———. The value of children: A cross-national study. 3 Arnold. Is discrimination in food really necessary for explaining sex differentials in childhood mortality? Population Studies 43:193–210. and child mortality in India. 1983. K. Alaka M. Roy. 1981. eds. 1984. In T. John C. Demography India 18:49–72. the status of women.32 National Family Health Survey Subject Reports. and Ray L. Rural poverty in South India. N. Son preference. Toshio Iritani. Bardhan. Netherlands: International Statistical Institute. Bulatao. World Fertility Survey. 20. Fawcett. George. Sex preference for children and its implications for fertility in rural Bangladesh. Demographic and Health Surveys. 1996. K. and Martin Vaessen. Monica. Preferences for the sex of children and their influence on reproductive behavior. Values and disvalues of children in successive childbearing decisions. East-West Center Working Papers. No. Bhatia. 1975. No. ———. Betty Jamie Chung. Journal of Family Welfare 24(2): 3–16. No. Minja Kim Choe. Bicego. 1989. Sex disparity in child survival in rural India. John. 1987. 1989. No. Sex preference and fertility behavior: A study of recent Indian data. 1986. 85. Radheshyam. Demography 18:1–25. Das Gupta. Jane Verrall. Culture. and Tsong-shien Wu. Maryland: Macro International. Rodolfo A. Vol. Introduction and comparative analysis. Demography 24:517–30. Bardhan. Oxford: Clarendon Press. Pranab K. H. Calverton. James T. Sung Jin Lee. The causes of demographic change: Experimental research in South India. . P. Chalio Buripakdi. J. Cleland. 1992. Infant and child mortality. and demographic behaviour. Oxford: Oxford University Press. 27.
The value and cost of children. Ghosh Dastidar. L. and C. Murthy. Bulletin of the Nutrition Foundation of India 8(4). pp. Bombay. and F. On kinship structure. M. V. Rajasthan and Madhya Pradesh. and S. Population and Development Review 9:35–60. N. Gupta. 1987. female autonomy. Espenshade.. B.: International Food Policy Research Institute. S. Goldsby. I.. 165–98. 1965.. A theory of the value of children. Food security and nutrition implications of intrahousehold bias: A review of literature. Debra. Staehling. 1996. 1987.. Agnes Quisumbing. Demography 31:375–401. K. The female child in India: A struggle for survival. V. 3rd ed. Discussion Paper No. 1996. Interpretation of z-score anthropometric indicators derived from the international growth reference. Khan. Factors associated with contraception in Bihar and Rajasthan: Findings from recent sample surveys. and demographic balance. D. Performance of family planning in India: Observations from Bihar. Trowbridge. Haddad. Population Bulletin 32:3–47. Christine Peña. E. W. B. Kapadia. Dynamics of population and family welfare. Kanitkar. E. Prasad. Tim. National Family Health Survey (MCH and Family Planning). The social context of family size preferences and fertility behaviour in a South Indian village. and R. American Journal of Clinical Nutrition 46:749–62. S. Kinship organization in India. and Alison Slack. Srinivasan and S. Thomas J. M.33 National Family Health Survey Subject Reports. J. 1966. J. . Marriage and family in India. Ghosh. Baroda: Operations Research Group. In M. 1994. Family planning practices in India: Second All India Survey. Food Consumption and Nutrition Division. K. American Journal of Clinical Nutrition 46:736–48. Dibley. Development of normalized curves for the international growth reference: Historical and technical considerations. IIPS (International Institute for Population Sciences). 1983. Dyson. Karve. M. Khan. Mukerji. Shanti. Neiburg. J. Bombay: Himalaya Publishing House. Friedman. L. No. Dibley. Bombay: Asia Publishing House. and B. and F. 1983. M. 1987. 1977.. N. Chizuru Nishida. Lawrence. T. Familial values. 3 Dharmalingam. Trowbridge. R. 1987. C. P. Uttar Pradesh. and Mick Moore. Michael Hechter. Prasad. In K. M. Khan. B. Washington. M.C. eds. Staehling. E. 19. 1995. 1983. W. contraception and utilization of MCH services in rural Uttar Pradesh. Delhi: Himalaya Publishing House. Genus 52:83–103. Gupta. Bombay: Oxford University Press. and Satoshi Kanazawa. India 1992–93. Arunachalam. eds.
and Gillian H. Samuel H. Annual Number. Patriarchy. Krishnan. Maryland: Macro International. 1989. Bombay: Oxford University Press.. India. Koenig. Chai Bin.34 National Family Health Survey Subject Reports. 1995. Mamdani. Demography India 21:145–66. Bombay: Himalaya Publishing House. Calverton. A. 1991. pp. Park. M. Chen. pp. M. N. 12. London: Monthly Review Press. eds. 1990.. No. Sex composition of children and fertility behaviour in rural Maharashtra. and Sajeda Amin. In Monica Das Gupta. and IIPS (International Institute for Population Sciences). 57–71. Mortality in India. Sinha. 1972. Paper presented at the Annual Meeting of the Population Association of America. Kathryn Stewart. eds. and T. The endangered sex: The neglect of female children in rural North India. Ithaca: Cornell University Press. Liège: International Union for the Scientific Study of Population (IUSSP). 1990. Bombay: IIPS. Sureender. and Sunil Sengupta. Health Transition Review 1:143–69. Barbara D. Family planning practices in India: Third All India Survey. Sommerfelt. Vol. Srinivas. Roy. 1992–93. and S. 19–54. No. 4. T. Population and Development Review 21:59–84. Baroda. How many births are attributable to preference for sex of children? A simulation analysis. 81–6. Dynamics of population and family welfare. pp. New Delhi. K. Culture and human fertility in India. Sunita. 1983. Gender differentials in child mortality: A review of the evidence. Preston. 1977. N. Sulabha. Lincoln C. P. Pebley. The myth of population control. Miller.-H. women’s status and reproductive behaviour in rural North India. Children’s nutritional status. Economic and Political Weekly. Sen. Foo. Madras: Oxford University Press. Elisabeth. and N. In K. 1981. May.. San Francisco. 1995. Demographic and Health Surveys. Malnutrition of rural children and the sex bias. 1992. Pathak. C. Chai Bin. Amartya K. PRC (Population Research Centre). In International Population Conference. The impact of a public-health intervention on sex differentials in childhood mortality in rural Punjab. U. and Arvind Pandey. Park. Operations Research Group. 1986. Parasuraman. 1994. National Family Health Survey (MCH and Family Planning): Uttar Pradesh. Comparative Studies. B. Women’s health in India: Risk and vulnerability. Michael A. Consequences of son preference in low-fertility countries in East Asia: Rising imbalance of sex ratio at birth. and M. 1994. 1994. 3 Kishor. Cho. Lucknow University. Anne R. . 3–5 April.
Michael. California: Sage Publications. Fertility preferences: Selected findings from the World Fertility Survey data. Selected factors affecting fertility and fertility preferences in developing countries. Dominique. No. Comparative Studies. Vlasoff. New York: Department of Economic and Social Affairs. No. 15. Population Bulletin of the United Nations 39:45–78. and Michel Willems. 1979. 1981. Bicego. United Nations. Demographic and Health Surveys. Carol. United Nations. Jeremiah M. UN (United Nations). Calverton. Tabutin. Nancy E. 1995. 3 Sullivan. New York: Department of International Economic and Social Affairs. Population Studies 33:415–28. Population Studies 44:5–20. Williamson. Beverly Hills. 1976. 1994.. and George T. ———. Vlasoff. Maryland: Macro International. Excess female child mortality in the developing world during the 1970s and 1980s.35 National Family Health Survey Subject Reports. . The value of sons in an Indian village: How widows see it. Sons or daughters: A cross-cultural survey of parental preferences. Shea Oscar Rutstein. 1990. 1985. Infant and child mortality. Labour demand and economic utility of children: A case study of rural India.