Contents - Previous - Next

This is the old United Nations University website. Visit the new site at


Gender differences

The United Nations Children's Fund (UNICEF) has assembled anthropometric data by sex for children in a total of 44 developing countries. Proportions of underweight were higher for girls in 25 countries and for boys in 19 countries. Many of the differences were so small that they could easily be due to sampling error2 and should not be over-interpreted. Gender contrast in some countries, however, are substantial. The extremes of both male and female advantage are found in Africa. In Madagascar, 40.2 per cent of boys and only 33.5 per cent of girls were underweight in 1983-1984, for a female advantage of 6.7 percentage points. In a sub-national sample in Rwanda, 28.5 per cent of boys and 40.5 per cent of girls were underweight in 1982-1983, for a male advantage of 12 percentage points. Indonesia was the only other country with a gender difference of more than 4 percentage points: a 1987 study found 48.7 per cent of boys and 55.3 per cent of girls underweight. These data indicate that important gender differences in nutrition do exist in some populations. However, in many, rates of malnutrition are very similar for children and sometimes males are worse off.

Comparisons by gender of adult anthropometry are relatively rare but those present in the literature also do not show a consistent male advantage. For instance, a study of agricultural migrant workers in Brazil documented a marked female advantage on a range of anthropometric indicators. Although, on some of the indicators, women were clearly malnourished relative to the standard (Desai et al. 1980, as summarized in Hamilton et al. 1984 and reproduced in Kanbur 1991), men were more so.

A thoughtful review of African evidence for gender bias in anthropometry (Svedberg 1990) rejected the hypothesis of female nutritional disadvantage there and concluded that the different roles of African (as compared with South Asian) women provide them with greater control over food and enable them to avoid nutritional deprivation for themselves and their children. Proportions of children underweight, stunted, or wasted appeared to be slightly higher for males than for females in four studies. A larger set of studies gave average weights and heights of preschool children, schoolchildren, and adolescents by age and gender; ratios of these values to the medians of the National Center for Health Statistics (NCHS) weight-forage and height-for-age standards were typically slightly higher for females than for males, implying a modest female advantage. For adults, average heights were compared with those in the tallest European populations. Men were further below the reference heights in 11 out of 17 sample populations, and women in only 4. Desirable weights for actual average heights were also calculated, and actual average weights were expressed as a ratio to these desirable weights. In all 17 populations, men were thinner, relative to weights defined as appropriate for their heights, than were women. As Svedberg notes, these data are not consistent with a pattern in which females receive less than their share of household food supplies.

Svedberg does not question that females in South Asia are less well fed than males there; in his interpretation of the African data, he focuses instead on reasons why a pattern presumed to hold elsewhere might not hold in Africa: the "relatively favourable and autonomous position" of women in Africa is taken as explaining the absence of female nutritional disadvantage.

The conclusion, that females are discriminated against in access to food, rests disproportionately on studies from South Asia. The social status of women is especially low in South Asia. Their economic activities are restricted and their income-earning potential is far lower than men's. They take resources (labour and bride-price) out of their natal households. These factors explain why females are perceived as less valuable than males in the region. The female advantage in life expectancy that applies in most countries is reversed in Bangladesh, Bhutan, Nepal, and Pakistan (World Bank 1992). Until recently, this has also been the case for India. A female disadvantage in mortality is generally explained in terms of preferential treatment of males, with poorer nutrition for females often specifically cited as a likely part of such a pattern of discrimination. We focus closely on evidence from this region because nutritional discrimination is likelier to exist and also likely to be more severe here than elsewhere.


Chen et al. (1981), seeking to explain higher female than male mortality in Bangladesh, assemble both anthropometric and dietary evidence as well as data on morbidity and health-care utilization. This particularly thorough study was made possible by the extensive data collected in the middle of the 1970s as part of the ongoing research programme of the International Center for Diarrheal Disease Research, Bangladesh (ICDDR, B). Relative to the US-based Harvard standard of weight-for-age, 59.9 per cent of boys and 74.0 per cent of girls were identified as either moderately or severely underweight. Higher rates of underweight for girls than for boys in Bangladesh are also reflected in more recent data, although the contrast is less sharp: 64.8 per cent of boys and 67.8 per cent of girls were underweight in 1989-1990. Chen et al. (1981) find a similar pattern of female disadvantage in children's height-forage, indicating that girls were more likely than boys to experience the stunting that reflects a long-term history of malnutrition.

This gender difference in nutritional outcomes of children is at least partly explained by discrimination against females in the allocation of household food supplies. Chen et al. (1981) show lower caloric intake for females than for males within every age group. In early childhood, recommended intakes are the same for girls as for boys, so the lower intakes of girls demonstrate a less-adequate diet. At later ages, requirements differ by gender in a manner affected by body weight, pregnancy and lactation, and physical activity. Chen et al. (1981) adjust for these differences (although cautioning that the adjustments are crude) and conclude that female diets are markedly less adequate than male at ages 0-4 and somewhat less adequate than male at ages 45+ (see fig. 5.1). Girls actually do slightly better than boys at ages 5-14. At ages 15-44 the contrast is dramatically affected by which adjustments are made: women of 15-44 appear to be disadvantaged relative to men when the additional caloric requirements of reproductive status are included (panel C of fig. 5.1), but this disadvantage nearly disappears when adjustments for levels of physical activity are also included (panel D of fig. 5.1). When all adjustments are made, adequacy of the diet is quite comparable for prime-age adult men and women in this population. However, discrimination against females in food allocation is marked in early childhood and present to a lesser degree among older adults. The apparent parity in intake relative to need in the 15-44 age group may result in part from an underestimation of women's activity levels.3

Fig. 5.1 Bangladesh - ratios of male/female caloric adequacy by age groups: (A) actual;

Fig. 5.1 Bangladesh - ratios of male/female caloric adequacy by age groups: (B) adjusted for body weight;

Fig. 5.1 Bangladesh - ratios of male/female caloric adequacy by age groups: (C) adjusted for body weight, pregnancy, and lactation;

Fig. 5.1 Bangladesh - ratios of male/female caloric adequacy by age groups: (D) adjusted for body weight, pregnancy, lactation, and activity

(source: Chen et al. 1981)



A detailed examination of both anthropometry and food-consumption data from India's National Nutrition Monitoring Bureau (NNMB), provides little support for the expectation of general female disadvantage. The NNMB data are for large and representative samples of households in seven to nine states, depending on the round of their ongoing survey efforts.

Table 5.1 shows average caloric consumption, and adequacy of that average consumption, by state and requirement category. The standards with which the Indian dietary intake data are compared here are defined separately by physical activity level for both men and women, and individuals are assigned to activity-level categories corresponding to their occupations (NNMB 1981). The possibility remains that women's occupations are evaluated as involving less physical activity than they actually do. For instance, one might question whether a housewife in rural India really leads a sedentary life, although that is how she is classified (NNMB 1981)! None the less, this definition of dietary adequacy relative to activity-level-specific requirements is a serious attempt to eliminate the arbitrariness and possible bias implicit in applying the same requirements to all members of the same age and gender category regardless of their level of physical activity.

For the moment, we will focus on the gender comparisons in table 5.1, which start at age 13. Average consumption for females in most of rural India is less than they need. However, to justify a conclusion that females get less than a fair share, we should see caloric adequacy for males at a higher level, but females are less undernourished in most of the possible comparisons. At ages 13-16, girls receive a higher percentage of their requirements than boys in seven of the nine states covered. In the later teen years, females are less undernourished in all nine states. In adulthood, comparing males and non-pregnant, non-lactating females at the same activity level, 12 of the 18 comparisons show females consuming a greater percentage of their requirements. We return below to the question of nutrition during pregnancy and lactation.

Table 5.1 Mean caloric intake, expressed as proportion of requirements (Indian definition), by State and requirement category: rural India, 1975-1978

Requirement category


Kerala Tamil Nadu Karnataka Andhra Pradesh Maharashtra Gujarat Madhya Pradesh West Bengal Uttar Pradesh Requirement (calories)
Age (years): 1 0.44 0.52 0.50 0.35 0.47 0.56 0.47 0.59 0.55 1,200
2 0.47 0.73 0.84 0.65 0.58 0.77 0.76 0.59 0.64 1,200
3 0.55 0.85 0.91 0.74 0.76 0.84 0.89 0.75 0.63 1,200
4 to 7 0.56 0.67 0.93 0.69 0.77 0.78 0.71 0.67 0.69 1,500
7 to 10 0.55 0.69 0.94 0.65 0.75 0.76 0.72 0.64 0.71 1,800
10 to 13 0.55 0.77 0.92 0.67 0.70 0.74 0.64 0.62 0.72 2,100
13 to 16 (boys) 0.48 0.69 0.91 0.60 0.67 0.70 0.62 0.69 0.81 2,500
13 to 16 (girls) 0.65 0.76 0.93 0.74 0.73 0.82 0.73 0.63 0.63 2,200
16 to 18 (boys) 0.53 0.66 0.79 0.67 0.63 0.69 0.68 0.62 0.74 3,000
16 to 18 (girls) 0.60 0.84 0.97 0.92 0.73 0.74 0.74 0.69 0.82 2,200
Adult males: sedentary 0.77 0.88 1.12 0.86 0.87 0.91 0.77 0.85 0.93 2,400
moderate activity 0.61 0.82 1.00 0.77 0.79 0.79 0.67 0.71 0.77 2,800
heavy activity - 0.54 0.56 - 0.44 0.48 - 0.60 0.66 3,900
Adult females: sedentary 0.73 0.92 1.23 0.95 0.96 0.92 0.94 0.84 0.91 1,900
moderate activity 0.52 0.88 1.11 0.90 0.82 0.78 0.77 0.66 0.85 2,200
sedentary and pregnant 0.55 0.59 1.15 0.82 0.76 0.78 0.66 0.87 0.62 2,200
sedentary and lactating 0.45 0.65 0.87 0.68 0.69 0.68 0.73 0.58 0.61 2.800
moderate and lactating 0.50 0.67 0.67 0.80 0.55 0.74 0.90 0.55 - 2,900

Source: extracted and constructed from NNMB (1980), tables 27-44

The possibility remains that girls aged 12 or younger are undernourished relative to boys of the same ages; since the NNMB does not publish food-consumption data separately by gender at these ages, the possibility cannot be tested with these data. However, the NNMB does provide anthropometry - a nutritional outcome indicator - by gender for children. The prevalence of underweight for both boys and girls is very high, but there is no female disadvantage.

Relative to the Indian growth standard, the anthropometric data given in table 5.2 show consistently and substantially higher rates of malnutrition for boys than for girls. This observation holds for rural children in all nine states covered by the NNMB, and for urban children in each of five socio-economic status groups. However, as argued in chapter 2, the Indian standard of weight-for-age incorporates a bias in favour of males, and the apparent overall female advantage here is an artefact of this biased measurement (see also table 5.3).

Table 5.2 Percentage of children aged 1-5 with weights less than 75 per cent of Hyderabad standard median weight-for-age (percentage moderately or severely malnourished)

  Percentage malnourished
Region and period Male Female Average Male - Female(= female advantage)
Rural India, 1979
Kerala 39.7 25.2 32.5 14.5
Tamil Nadu 53.2 30.8 42.0 22.4
Karnataka 52.8 39.3 46.1 13.5
Andhra Pradesh 44.8 24.6 34.7 19.2
Maharashtra 49.5 39.9 44.7 9.6
Gujarat 49.6 40.3 45.0 9.3
Madhya Pradesh 61.8 52.6 57.2 9.2
Orissa 54.6 41.1 47.9 13.5
West Bengal 43.8 33.0 38.4 10.8
Uttar Pradesh 33.4 19.7 26.6 13.7
Average 49.3 35.3 42.3 14.0
Urban India, 1975-1980
High income 12.0 10.1 11.1 1.9
Middle income 19.4 12.2 15.8 7.2
Low income 36.7 27.2 32.0 9.5
Industrial labour 38.6 30.0 34.3 8.6
Slum 52.1 39.9 46.0 12.2

Source: extracted and calculated from NNMB (1980,1981,1982).

Even though the apparent female advantage cannot be believed, variations in the pattern of gender contrast may still be meaningful. Most informative is the contrast across groups within India's cities. The groups are ranked from best-off to worst-off, as is apparent in the monotonic rising series of combined male and female proportions of underweight (column 3, table 5.2). Underweight rates for boys are clearly more sensitive to socio-economic status (SES) category than those for girls (compare columns 1 and 2, table 5.2). As SES increases, boys benefit more than girls. What appears (given the bias in the standard) as a 12-point female advantage among slum children diminishes to near-equality among children in high-income households. It is often assumed that any bias against females will operate more strongly in poorer households and thus that improvements in household economic situations will benefit females more than males. These data, however, suggest that preferential treatment for males, relevant to children's nutritional status, operates more strongly in households that have more resources. This finding is not inconsistent with other studies of gender bias among Indian children (Des Gupta 1987; Miller 1997). In fact, Miller's review of South Asian studies revealed a consistent pattern of greater gender bias within higher social groups and highlighted the fact that discrimination against girls among the wealthy may be hidden in the aggregate figures.

India's NNMB has applied both the NCHS and the Hyderabad standard to data for 1975 and 1989 NNMB 1989). In comparison with the data shown in table 5.2, fewer states were covered, smaller samples were taken within each state, and urban areas were omitted entirely in the data subjected to this analysis. For these reasons, only summary figures for rural India are shown in table 5.3. As in table 5.2, female proportions of underweight appear lower than male, relative to the Hyderabad standard. In contrast, the proportions of underweight relative to the NCHS standard are virtually identical for boys and girls, at 61.9 per cent and 60.0 per cent, respectively, for the seven-state combined sample for 1989. When the presumably non-gender-biased NCHS standard is used, the female advantage that we see relative to the Hyderabad standard disappears. However, there is still no sign of the expected male advantage.

Table 5.3 Percentage of children underweight, rural Indiaa

Standard and year

Percentage underweight

Boys Girls Combined Female advantage
1975 55.5 44.2 50.3 11 3
1989 42.5 28.0 35.3 14.5
1975 71.6 71.6 71.6 0.0
1989 61.9 60.0 61.0 1.9

Source: extracted and calculated from NNMB (1989).
a. States covered are Kerala, Tamil Nadu, Karnataka, Andhra Pradesh, Maharashtra, Gujarat, and Orissa.

The Hyderabad standard, based on the anthropometry of the Indian elite, incorporates a bias against females and supports the point argued above that discrimination against girls occurs in relatively wealthy households. But this should not be generalized to Indian society as a whole.

Children's anthropometry is influenced by disease as well as by food intake, so we should not conclude from this evidence alone that girls' diets are as inadequate as boys' diets in India. However, to reconcile these data with a pattern of food allocation that discriminates against girls, one would have to postulate a female advantage in health strong enough to overcome the effects of less-adequate food intake. Medical anthropologists argue that males tend to be more vulnerable to physiological insult than females (Stinson 1985). On the other hand, preferential use of preventive and curative health care operates to the advantage of the preferred, in this case males. There is reason to believe that male children in some parts of South Asia do enjoy a favoured position with respect to these resources (see, e.g., Chen et al. 1981; Das Gupta 1987; Kynch and Sen 1983). The bottom line is that gender differences in anthropometry shown for India do not support the notion of substantial dietary discrimination against females in early childhood.

But anthropometric evidence is available only for surviving children. India has been known as one of the few populations in which males outlive females, and this reversal of the usual female advantage in life expectancy has itself been interpreted as evidence of poorer nutrition for females. If this pattern persists,4 however, these comparisons of food consumption and nutritional status suggest that the explanation does not lie in dietary but rather in health-care discrimination against females.

This conclusion must be qualified by highlighting the fact that not all of the states of India are covered by the data analysed here. Those omitted include some, such as Haryana and the Punjab, in which female status is believed to be the lowest. Kynch and Sen (1983) rank the states of India according to the ratio of females to males in 1981, interpreting lower values as indicating more-severe discrimination against females. By this measure, one might expect females to be worst off relative to males in Haryana and the Punjab and best off in Kerala and Orissa. Of the states covered by the NNMB data, the 1981 ratio of females to males was lowest for Uttar Pradesh, in which several of the gender comparisons within age and activity-level categories do favour males. State-level comparisons of male and female mortality patterns for 19691977 (Dyson 1989) also suggest that females may be more disadvantaged in Uttar Pradesh and the Punjab than in other Indian states.

Dietary-adequacy data from another source (M. Das Gupta and S. R. Millman, work in progress) are available for one of the two states in which analyses by both Dyson (1989) and Kynch and Sen (1983) suggest the status of women may be lowest. Figure 5.2 illustrates the age patterns of dietary adequacy for males and females in the rural Punjab. The curves intersect at several points, but adequacy is higher for females over much of the life cycle.

In this graph, female caloric adequacy figures during the reproductive ages are a weighted average of those for women who are pregnant, fully lactating, partially lactating (i.e. their youngest children are receiving other foods as well as breastmilk), or neither pregnant nor lactating. When these groups are disaggregated in figure 5.3, we see that dietary adequacy deteriorates sharply with pregnancy and further still with lactation. Male diets among 16-44-year-olds in this sample met almost exactly 100 per cent of their requirements. Therefore, reproductive-aged women who are neither pregnant nor lactating are actually better fed than comparable males. But women undergoing reproductive stress consume diets less adequate relative to need than men do. The slight female disadvantage we observe at ages 22-34 in the more aggregated data results from the high proportion of women at these ages who are either pregnant or lactating.

A similar pattern of deteriorating adequacy of the diet associated with pregnancy and lactation is observed for the multiple states of India covered in table 5.1. Pregnant women consume a lesser proportion of their requirements than women who are neither pregnant nor lactating, and in most states the situation during lactation is worse still. This pattern partly results from the increase in requirements during pregnancy and lactation. In some states, however, even absolute consumption declines during pregnancy; this is not just an insufficient increase. In others, average caloric intake is remarkably similar across women of differing reproductive status. At the best, we see increases that are less than proportionate to increased requirements, so that the average absolute caloric shortfall rises.

Fig. 5.2 Caloric adequacy of the diet by age and sex, rural Punjab: intake expressed as a percentage of requirements based on WHO/FAO recommendations (source: Das Gupta 1995)

Fig. 5.3 Caloric adequacy of the diet by reproductive status, women ages 16-44, rural Punjab (source: Das Gupta 1995)

Women spend a significant portion of their lives subject to the nutritional stresses associated with reproduction, if they bear many children and if each child is breast-fed for many months. This pattern of severe undernutrition during pregnancy undoubtedly contributes to the high prevalence of low birth weight in India. Similarly, inadequate intake during lactation may limit breastmilk production and thus the growth of children for whom breastmilk is an important part of the diet. But the nutritional stress that women experience in their role as mothers does not affect only their performance of this role: their own health, their ability to perform other functions, and the quality of their own lives are also at stake.


Contents - Previous - Next