In India, Son Preference Declines with Ideal Family Size, but Remains Strong
In northern India, women's desired fertility is positively associated with their level of son preference, according to analyses based on data from two successive rounds of a nationally representative survey.1 The smaller the reported ideal family size, the lower the likelihood of wanting more sons than daughters and the lower the proportion of sons in the reported ideal family. However, the sex ratio of wanted infants in northern India increased from 130 to 139 males per 100 females between surveys, while the wanted total fertility rate decreased from 3.2 to 2.4 births per woman, suggesting that during this fertility decline, the practice of female feticide increased and outweighed the effect of declining son preference.
Previous research showed that fertility decline in India is accompanied by an increase in the ratio of male to female children aged six and younger, and suggested that persistent son preference increases the rate of female feticide or infanticide at low parities, despite a reduction in the number of unwanted daughters at high parities. To further investigate the association between fertility and sex bias in India, researchers examined data from ever-married women who participated in the 1992-1993 and 1998-1999 National Family Health Surveys.
In all but one Indian state, ideal family size and son preference declined in tandem between the two surveys: The ideal number of children declined from 2.9 to 2.7, while the overall proportion of women wanting more sons than daughters decreased from 42% to 33%, and the average proportion of sons in the ideal family from 54% to 51%.
The researchers focused their analyses on northern India, where the overall level of son preference was particularly high: In 1998-1999, some 47% of women reported wanting more sons than daughters, whereas only 2% preferred daughters, and the proportion of sons in the ideal family was 56%. Although son preference was greater among women who desired odd numbers of children than among those who desired even numbers, it generally decreased with a decreasing ideal family size. For example, 75% of women who wanted five or more children reported preferring sons over daughters, compared with 67% of those wanting 3-4 children and 12% of those wanting 1-2. In addition, the proportion of sons in the ideal family decreased from 55-63% for an ideal family size of 3-4 to 49-50% for an ideal family size of 1-2. The actual proportion of sons among living children, however, generally increased with decreasing family size—from 49% at parities of more than five to 58% at a parity of two and 55% at a parity of one. According to the investigators, this finding indicates that in reality, smaller families are made up of more sons than daughters, and that women's reported ideal family compositions do not simply reflect a rationalization of previous births.
Multivariate regressions that controlled for variables related to reporting bias confirmed that for both surveys, ideal family size was positively associated with both the likelihood of wanting more sons than daughters and the ideal proportion of sons. Other factors that were positively associated with at least one of these outcome measures in at least one of the surveys were an odd-numbered ideal family size, Sikh religion, unpaid employment and the numbers of living and dead sons. Factors that were negatively associated with son preference were women's age, urban residence, regular exposure to the media, Muslim and Christian religion, membership in a scheduled tribe or caste, educational level, standard of living, paid employment and the number of living daughters. Hence, the analysts comment, in northern India, "the preference for sons declines with the forces of modernization and a decrease in the desired family size."
The investigators point out, however, that their results seemingly contradict previous findings showing a rising ratio of male to female children. To resolve this paradox, they examined the sex ratio at birth of wanted children and the sex composition of unwanted births. Between the two surveys, the sex ratio at birth increased slightly from 105 to 110 males per 100 females; the sex ratio of wanted children increased from 130 to 139, or from 125 to 130 after correction for rationalization bias. These findings, the analysts note, are "consistent with the reported practice of female feticide in some parts of northern India." Furthermore, comparisons between preferred and actual sexes of children born in the year before the second survey revealed that approximately 60% of unwanted births were those of daughters. Between the two surveys, the estimated total fertility rate for all wanted births decreased from 3.2 to 2.4.
The researchers conclude that "at any point in time, the number of unwanted daughters is more than the number of unwanted sons," despite the tendency for son preference to diminish as desired family size falls. They attribute the increase in the sex ratio during a period of fertility decline in northern India to "the increasing availability of prenatal sex-selection technologies," rather than to an intensification of sex bias per se, as previously assumed. In the authors' view, prenatal sex selection allows women to realize their "repressed demand for sons" and outweighs the influence of declining son preference. They call for the continued prohibition of the use of new sex-detection techniques in India, because "if parents were to be denied access to such technologies, a decline in fertility...should make the population less masculine."
1. Bhat PNM and Zavier AJF, Fertility dcline and gender bias in northern India, Demography, 2003, 40(4):637-657.