Contents - Previous - Next

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

The literature and its policy and programme implications

Rationale for studying positive deviance
Conceptual approaches
Overview of findings from the literature
Overall conclusions and policy recommendations
Introduction to the detailed review of behavioural, psychological, and social correlates of child growth
Maternal/caretaker-child interactions
Child characteristics
Maternal characteristics
Socio-cultural support


The concept of "positive deviance" had already entered the nutrition literature by 1967, although serious field research in this area is more recent. Hegsted (1967), for example, advised that "we should pay a great deal more attention to those individuals who are apparently healthy while consuming diets which seem to us to be restricted. We should pay more attention to the reasons for nutritional success rather than nutrition failure." Wray (1972) advocated studying "successful mothers," while Greaves (1979) recommended that "another approach might be to identify in the village women who can cope: there are many who do manage to rear healthy and active children, and yet who belong to the same 'community' as some who cannot. How do they manage? Are they following some of the basic rules? What is their secret? Can it be shared with others? There would seem to be tremendous opportunity for research here."

Mata (1980), writing on child malnutrition in Guatemala and Costa Rica, observed that "maternal technology" is a distinct determinant of the malnutrition complex: "Maternal technology pertains to practices, traditions and beliefs relating to food preparation, feeding techniques, child care during illness and co,,nvalescence, handling of drinking water and of feces, and personal hygiene. It is important to note that some mothers exhibit a high level of technology, independent of their upbringing or schooling. That is why some infants and young children in a community thrive well under village conditions."

In addition to practices relating exclusively to feeding, health care, and sanitation, other modes of mother-child interaction have also been consistently related to the nutritional status of infants and young children. The mother's affect - whether she smiles and enjoys the baby - and the frequency with which she interacts with her child, verbally and non-verbally, can be used as examples (Alvarez et al., 1982). Genetic factors governing energy metabolism, immune function, response to stress, and activity levels also play an underlying and as yet poorly understood role in determining nutritional status (Danforth, 1983).

Rationale for studying positive deviance

It is widely acknowledged that socio-economic development will not alleviate poverty over wide areas of the developing world during the next few decades. Nutritional conditions have worsened between 1980 and 1985 and are continuing to deteriorate in most developing regions outside of South and East Asia (Cornia et al., 1987). With these disturbing trends in mind, international assistance agencies have supported the search for and implementation of feasible approaches to improve child survival. An important goal is the protection of the health and development of the next generation of young children in the world's poorest regions.

In presenting the rationale for seeking practicable solutions, it is also important to stress the importance of economic development and of equity of resource distribution, including access to health and other social services. Our focus on positive deviance should not divert attention and energy away from efforts to change the. economic, social, and political conditions of the poor. Nevertheless, needless death and retardation of young children are among the most distressing manifestations of socio-economic inequity. This paper is concerned with the broadening of affordable methods that could be used in reducing these inequities.

At this point in time, primary health care (PHC) forms the programmatic basis for feasible interventions in health. The essential elements of PHC are: health education, promotion of food supply and proper nutrition, safe water and basic sanitation, maternal and child care, including family planning, immunization against communicable diseases, prevention and control of locally endemic diseases, appropriate treatment for common diseases and injuries, and, lastly, provision of essential drugs.

The knowledge base for the development of these activities as priority components was derived from a large number of epidemiological surveys focused on the causes of malnutrition, morbidity, and mortality of young children in developing countries. These epidemiological field studies were based on the need, first of all, to establish patterns of distribution of morbidity and to identify some of the broader associated factors. Given that this has been done, there is now an opportunity to promote more in-depth studies, in order better to identify and understand the group-specific characteristics associated with morbidity and mortality (Rogers, 1985; Griffiths et al., 1984).

A variety of social-science research methods are beginning to provide a greater depth of insight for the purposes of policy formation. Operational research and behavioural trial methods have begun to integrate the research process into the design of effective interventions.

Nutrition communications research and action projects have demonstrated that a blend of community development and commercial and social marketing techniques can achieve major changes in maternal behaviour. As found in Indonesia (Zeitlin et al., 1984), such changes may significantly improve the growth status of young children in low-income environments, even in the absence of major investments in health care or economic development. The JNSP in Haiti, as well as projects in Honduras and the Gambia, have used these methods to promote oral rehydration therapy (Smith et al., 1982).

To date, operational and behavioural research methods have not attempted to draw extensively upon the local adaptive wisdom of the mothers and families of the positive-deviant children in low-income communities. Very few studies in nutrition have systematically examined the households whose children are at the top end of the growth performance curve in order to learn from successful adaptations.

From a purely scientific point of view, additional interest in positive deviance is motivated by the recognition that a focus on the malnourished only-the bottom tail of the distribution - lacks methodological rigour and fails to provide the basic information needed to understand the causes of malnutrition.

Similar research in psychology, looking at "invulnerability" or the reasons why some children develop into psychologically healthy adults despite poor nurturing or traumatic early circumstances, has been conducted for several decades (see Anthony and Cohler, 1987, for a review of this approach). In the epidemiology of diarrhoeal disease, a positive-deviance casecontrol approach is now popular for studying hygiene behaviours associated with greater and lesser diarrhoeal incidence in low-income communities (Clemens and Stanton, 1986).

Example from Burma

An example from a positive-deviance study conducted by the Burmese Department of Medical Research illustrates the value of this approach to nutritional research. It demonstrates that information gained from studying the top of the distribution curve may call into question the accuracy of previous conclusions based on study of the malnourished only. The Department of Medical Research in Burma collected socioeconomic, anthropometric, dietary, and morbidity data from 3,29X households of children aged 0 to 3 years in representative regions of Burma. Two-day weighed food intake, including test-weighed breast-milk, was taken from an intensive study subsample of 874.

According to the accepted theory concerning growth failure at the weaning age in developing countries (based primarily on studies of the malnourished), the quantity of breastmilk becomes insufficient after 6 months, and malnutrition occurs in infants whose mothers do not start supplements soon enough or in large enough quantities. The Burmese study conducted by Dr. Cho Nwe Oo and colleagues (Nutrition Research Division, 1985) suggests that supplementary feeding patterns were not of primary importance in determining those who were well-nourished and those who were malnourished from 0 to 12 months. Rather, the quantity of test-weighed breast-milk consumed by the children differed significantly between positivedeviant, average, and malnourished infants between 0-3, 4-6, and 7-12 months.

From 0 to 6 months positive-deviant children receive fewer calories from supplementary foods than the other two groups. In fact early introduction of larger quantities of supplements was associated with poorer nutritional status in this breast-fed population (who did not practice formula feeding). From 7 to 12 months, the well nourished received significantly more breastmilk but only slightly more calories from supplements than the average and malnourished groups. These findings are supported by research from Thailand (Van Aarsdale, 1983). In the Burmese case, a scientific test of positive-deviance hypotheses yielded findings with important implications for intervention.

The accepted theory concerning growth failure at the weaning age and the hypothesis based on the Burma study are contrasted in figures 1a and 1b.

The study raises the question of the degree to which the large quantity of breast milk the mothers produced was due to the fact that larger babies stimulated greater milk production by their more vigorous sucking and higher intake needs, as the proportion of low birth-weight ( < 2,500 g) was 4 per cent among the subsequently well nourished, versus 17.5 and 36.7 per cent among the average and malnourished.

If early supplementation and a larger quantity of supplements are not associated with good nutritional status between 4 and 12 months, this may be due to quality of food given, its hygienic standard, and the possible reasons for starting supplementation.

Fig. 1a. The logic of traditional weaning theory.

Fig. 1b. New implications based on Burma positive-deviance study.

Conceptual approaches

Given the exploratory nature of positive-deviance research, this review will refer broadly to positive-deviance factors, behaviours, and hypotheses. These terms will generally characterize or pertain to the families of children whose growth falls in the upper third of the nutritional status distribution within a homogeneous low-income population.

Table 1. Four types of adaptation to limited food availability

Type A Level Benefits
1. Increase in neo-natal and young child mortality Across generations 1. Control of population size relative to food supply and ecological balance
    2. Selective survival of genotypes. social structures. and behavioural models resistant to nutritional stress
2. Stunting in linear growth, other metabolic adaptations to limited or irregular supply of energy and other nutrients Across and within generations Modification of phenotypes (rarely genotypes) to reduce food nutrient requirements
3. Socialization of child to be efficient member of food production and distribution system Across and within generations Efficient exploitation of technology characterized by simple, monotonous agricultural labour
4. Positive deviance Household and individual 1. Healthy nucleus maintained under adversity
    2. Maintenance of vitality in social class/leadership structure
    3. Improved survival value across generations with development of more efficient behaviours, technologies, and social structures


Positive deviance can be viewed as one of several interrelated adaptations to limited food availability (table 1). This section will describe the types of adaptations that are known or presumed to occur in malnourished populations and will present a conceptual approach to research and programme design that takes confounding interactions between these adaptations into account. For a discussion of child development in the context of evolutionary theory, the reader is referred to chapter 1 of Chisholm's (1983) book on Navajo infancy. It should be noted that, as Beaton (1984) pointed out, such adjustment to a limited food supply generally entails costs to the individual and to society.

Positive Deviance as Adaptation

The first form of adaptation occurs at the population level across generations. Limited food supply or agricultural carrying capacity of the land leads to increases in neo-natal and young child mortality. This mortality serves two main functions: 1. Control of population size relative to food supply and ecological balance. 2. Selective survival of genotypes, social structures, and behaviour patterns resistant to environment limitations in food availability. Some of the mechanisms related to mortality are: decreased maternal nutrition, leading to premature births and intra-uterine growth retardation, hence to increased neonatal and infant mortality; decreased breast-milk production and food availability for the weanling infanticide and differential care (Scrimshaw. 1978. 1982). The term dif ferential care describes the fact that parents treat unwanted or less wanted children in ways that sometimes permit them to die without the parents taking conscious responsibility for infanticide. Cassidy (1980) explains and discusses at length the first three types of adaptation presented here using the term "benign neglect."

Mortality rates typically increase in times of food scarcity. In many situations, there is now reason to believe that the modification of certain child-feeding and -rearing practices has enhanced child survival, even under the most stressful of conditions. In certain circumstances, on the other hand, current beliefs and practices may be actually exacerbating what are, for a variety of reasons, high-risk situations. Practices which are believed to protect the child, but which actually increase exposure to insult, may serve the purpose of permitting the child to die while the mother pursues highly apparent measures to save him. Withholding all fluids from infants with diarrhoea may be such a practice, as may be withholding key supplementary foods because "infants cannot digest them." These behaviours, which may not be consciously recognized as life-threatening to children, may have the effect of enhancing the survival of the group as a whole (Scrimshaw, 1978, 1982) by allowing the weakest children to die. They may thus contribute to the first form of adaptation.

The second type of adaptation occurs both across and within generations. Malnutrition in early childhood leads to stunting in linear growth. This stunting produces child and adult phenotypes with reduced food requirements and lower likelihood of overt nutrient deficiency symptoms which may occur when growth is accelerated (Hepner and Maiden, 1971). Other metabolic adaptations to reduced energy availability also occur and are introduced below on page 29.

The third type of adaptation socializes young children not to expect favoured foods or special treatment because of their low position in the family. It also sacrifices individual well-being for the survival of the group. This adaptation is in response to the inefficient subsistence agricultural production and distribution system to which child labour must contribute, and to the high reproductive burden placed on mothers who must continue to bear children every two to three years.

In agrarian societies with high fertility and high mortality, parents are usually guided by cultural modes, the goal of which is to maximize the number of surviving offspring, but not to optimize the growth and development of each one. It can be pointed out that societies in which the concept of child-spacing has not been adopted simply may not have the time, knowledge, or motivation to deal with growth and development. These models for parental behaviour are themselves adaptations, culturally evolved over the centuries to conditions of (a) labour-intensive family agricultural production and (b) natural fertility. The first of these creates a demand for child labour in the home: the more children, the more land can be cultivated and the more work distributed.

The second condition, natural fertility, creates an expectable birth succession in which a mother concentrates her efforts for child survival on the initial risk period, namely the early period of breast-feeding, especially the first year of life when mortality rates are highest. The mother then turns her attention to the next baby once he is born. This postpones risks of malnutrition and neglect to the post-weaning period, but it probably resulted - during times of better ecological equilibrium than are often seen today- in favourable survival rates despite malnutrition, chronic morbidity, and lowered life expectancy. Given the need to enhance child survival or accommodate high mortality through high fertility, it may well be that the importance of pre-weaning care of the child has failed to gain its merited significance.

In their study of the Gusii in Kenya, LeVine and colleagues (forthcoming, 1989) documented special parental care during the first year of life. They found that mothers hold and feed more frequently those six-month-olds whose actual heights and weights were less than others. This suggests that they may be monitoring the child's growth and using compensatory measures available from local cultural models to bring faltering infants up to a standard. Their time schedule for phasing out intensive attention and devoting themselves to agricultural work was also dependent on the infant's attainment of motor milestones (e.g. sitting without help, walking unassisted) at normal ages. They used physical growth and motor development (along with disease patterns), then, as individually varying infant characteristics to which they responded flexibly during the first year. Indirect evidence suggests that mothers improve with parity in this kind of sensitivity. This type of monitoring, however, did not prevent the growth failure and high morbidity rates in the second half of the first year, features that characterize developing country populations.

The weaned child is enculturated to participate in subsistence agriculture in which the family is the production unit. He or she is being prepared for early induction into the family work force (Levine, 1974). As pointed out by Caldwell (1981), such a child should be weaned cheaply. Parental goals are to produce an undemanding, compliant worker, starting work from 3 to 5 years of age. Once past the dangers of early infancy, the child must accept its lowly rank as the least productive and youngest member of the production team. As a symbol of his entrylevel status, he may receive the poorest quality and the smallest portions of food, and must not question this.

Living in a Yoruba village in Nigeria in 1966, Zeitlin observed that children learned their rank starting from the age of two or three years. They were encouraged to apply a rule of conduct that permitted any older child to give orders to and exact obedience from any younger one. Cousins differing by a few weeks in age knew exactly which of them was senior. This "game" tended to give authority for food and play materials to older children. Its enforcement was a source of fascination for the children.

Dr. E. P. Y. Muhondwa (personal communication, 1984), discussing the same ranking system in Tanzania, has pointed out its benefits. In the absence of adults, each older child is held officially responsible for the welfare of the younger ones. Older children are taught that although they have the right to exact services or food from their juniors, it is their duty to see that the younger ones get their share to eat. These rules facilitate orderly child care and child labour in situations where adults are too busy to supervise the children closely.

Harkness and Super (1977) wrote that the Kipsigis of Kenya needed children who responded to adult speech with obedient action, rather than reciprocal speech. Mothers increased their commands and insults to the child after the age of two and the child spoke less frequently.

The highly active play, assertiveness, and curiosity of well-nourished pre-schoolers may be unwelcome and difficult to handle for busy caretakers in subsistence agricultural settings. Where agriculture is particularly demanding and food supply most limited, compliant submission to leadership becomes increasingly necessary. Nondemanding toddlers who sit quietly in one spot may be most valued, even though this behaviour may in fact be indicative of chronic infection or debilitation.

Whereas in some countries children appear to demand more attention during the second year than during the first year of life, Gusii mothers in Kenya expressed fears that if an infant's demands were allowed to escalate, he would be difficult to manage when the next child was born. Those children and toddlers left in the care of a grandmother were considered to be spoiled.

The values that enculturate the young child to accept its low rank under conditions of food scarcity may be elevated to the status of moral principles. The mother may feed her child less well in order to teach the child not to be selfish. In Java, Indonesia, self-restraint in eating is an important moral virtue. The pregnant mother may already deny herself food in order to train the infant in her womb in the virtues of self-denial. Similarly, the mother of the toddler may feel that she is spoiling her child if she feeds him full meals that include generous portions of fish and other attractive nourishing foods (Dr. R. Sobekti, personal communication, 1982). In Senegal and the Gambia children are trained "not to sacrifice themselves to their stomachs" (N'Doya, 1980). Chavez and Martinez (1982) described the ways in which the practices that induced malnutrition among the Indians in Mexico also prepared the children to live placidly as adults under impoverished and monotonous conditions.

A study in Nigeria of Yoruba students' memories of childhood found that many students' strongest single memory was the punishment they received when caught by their parents "stealing" meat from the cooking pot (Levine, 1974).

By limiting the quality and quantity of food offered to young children, this third type of adaptation also contributes to the detrimental weaning practices that increase the mortality and stunt the growth of children in traditional low-income communities. These negative weaning practices are the primary target of nutrition education and growth-monitoring interventions in almost all developing countries.

We may then ask ourselves: First, do we not wish to continue to reform the negative food beliefs and practices as part of an overall strategy of socio-economic development? - to replace the old methods of population control with family planning? - to provide sufficient food to make stunting an unnecessary price to pay for group survival? - and to promote production technologies and social systems that encourage parents to stimulate and educate their children? Second, given that we owe the good health and nutrition of our children to the most scientifically advanced adaptation on the planet, shouldn't our goal be to enable all to share its benefits? Finally, is it not paradoxical under these circumstances to start searching for positive behaviours that enhance growth overlaid against a backdrop of negative behaviours that retard it?

The answers to these questions are not simple:

  1. Yes, we do wish to keep on replacing formerly adaptive practices leading to high mortality, stunted growth, and impoverished lives.
  2. No, it is not strategically possible at this time to bring the full benefits of modern technology, i.e. the high standard of living of the developed world, to the majority of the world's poor. Therefore, we are looking for interim solutions.
  3. No, it is not contradictory to look for factors that foster good growth and health overlaid against cultural practices that diminish survival, growth, and vitality. It is not contradictory because these factors are known to operate in both human populations and other species as a fourth type or level of adaptation.

This fourth and last positive-deviant type provides the dynamic for cultural (and/or physiological) evolution within the micro-environment. This form of adaptation occurs at the individual, household, or, in our present case, the caretaker-child dyed level. It manifests itself in forms of adaptation that permit individual households or dyads to achieve better growth rates and lower mortality rates than are average for their population, and serves to:

The Need to Combine Positive-deviance and Child-survivai Approaches

This complicated overlay of biological and cultural adaptations implies that in many or most cases one cannot search for simplistic positive-deviant solutions to local problems that have been developed by a few families and then set out to teach these solutions to the community as a whole. Some solutions may be embedded in an environmental context that needs to be changed. If a behaviour is adaptive because it helps children to survive in spite of cultural practices that stunt growth and increase infant mortality, one may first wish to change these dysfunctional infant-feeding and -care practices. Aher such change takes place, behaviours that previously served a positive-deviance function of protecting the malnourished child may no longer be adaptive.

The following examples illustrate positive-deviant adaptations that would not be incorporated into a child-survival strategy. Chavez and Martinez (1982) found that unsupplemented Indian infants had low activity levels and body temperatures that were half a degree lower than their supplemented counterparts. Keeping such babies almost constantly tied to the caretaker's back could have protective benefits by maintaining body temperature, psychological well-being from the presence of the caretaker, physical movement with very little energy expenditure by the child, and protection from exposure to infection and other hazards. LeVine ( 1974) discusses such protection in African cultures, where the child on the back is kept from being burned by cooking fires.

It might be found that Indian mothers of positive-deviant infants, in households with lower levels of infant mortality, were more consistent than others in carrying their babies on their backs, at least during certain hours of the day and up to a certain age. Yet, the change agent would probably choose not to reinforce actively the behaviour of carrying the baby on the back. The well-nourished supplemented toddler was too heavy to "back" for long periods, and resisted the confinement of being tied by the caretaker (Chavez and Martinez, 1982), except when being rocked to sleep. In such cases the mother's fatigue must also be considered. Kenyan women commenting to researchers about a photograph of a woman tea-picker at work with a large baby on her back expressed concern and pity for both mother and child (Bifani et al., 1982).

In Pakistan, the Unani (Graeco-Arabic) food-belief system prohibits pregnant women and young children from eating meat, eggs, and fish in summer. These are considered intrinsically "hot" foods that may cause diarrhoea and vomiting, particularly if eaten in hot weather. Given temperatures of more that 40 degrees Celsius without refrigeration, the likelihood that these foods will be consumed in a contaminated state in summer is high. For mothers who do not understand the germ theory of disease and therefore have no theoretical basis for practicing simple hygiene, positive-deviant behaviour may consist of delaying introduction of these foods to infants until after the age of two, when the incidence of diarrhoea is reduced (Ahmed and Zeitlin, 1972). The change agent would not accept the Unani prescriptions for withholding foods as examples of positive deviance to be reinforced within the community.

The reason for providing these examples is to demonstrate to the reader the fact that adaptive behaviours of type 4, the positive-deviant variety, must not be viewed as adaptations that can be automatically transferred without modification (although some may be). Behaviours do not stand by themselves. They are part of a total package of adaptation and cannot be manipulated individually without setting off a chain reaction. The usual procedure must be to integrate positive-deviance factors into a consistent pre-tested package of approaches to change based on known methods for improving child survival. This package must be developed by scientific behavioural trials in the homes of the malnourished (concept testing) and with due consideration to all four of the adaptation levels discussed earlier.


Contents - Previous - Next