Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Introduction
and purpose
Underlying
assumptions or hypotheses for research in positive deviance
Relationship
of positive-deviance research to epidemiological methods
Definition
of terms and specification of research goals
Three-stage
research and pilot-project model
Research
design for stage 1
A
conceptual framework for the design of positive-deviance studies
Important
variables: results of the positive-deviance mail survey
Micro-level
variables measuring caretaker-child interactions
Variables
measuring maternal characteristics and socio-cultural support
Measuring
growth
Controlling
for socio-economic status
Limiting
the number of covariables: restriction by age and topic
Rationale
for existing behaviours and social structures
Timeline
for change
Nutrition
and infection
Management
of multidisciplinary teams
One of the purposes of this document is to share research approaches and solutions to methodological problems in studying positive deviance with the network of scientists working on this topic.
Most prior research has focused on problems. Positive-deviance research centres on solutions. From an applied perspective, the purpose of studying positive deviance is to design interventions that incorporate and capitalize on existing success factors - on the adaptative behaviours, social-support systems, and physiological mechanisms that already operate to protect well-nourished children in deprived communities.
From a scientific point of view, it is necessary to study beneficial adaptations to nutritional stress in order to discover the types of adaptation that occur and how they function. A similar adaptive capacity, moreover, would be expected to operate across socio-economic levels, generating healthful outcomes amidst overnutrition. The majority of existing health and nutrition surveys do not provide such information. Conducted primarily for needs assessment, most nutritional field research has described problems in nutritionally deprived or imbalanced populations in terms of biological reference standards, levels, or rates achieved in favoured reference groups. This research is indispensable. However, it sheds little light on the mechanisms that buffer and protect human beings from the negative effects of food scarcity or imbalance and highly contaminated environments.
Underlying assumptions or hypotheses for research in positive deviance
The study of positive deviance in young-child nutrition rests on the following assumptions:
1. In impoverished or less socio-economically developed communities there exist adaptive parental and child characteristics that distinguish between the households of children whose growth performance is at the top of the distribution and those whose children are malnourished. These characteristics may be behavioural, social, psychological, or physiological.
2. Certain of these characteristics will be useful for developing:
Relationship of positive-deviance research to epidemiological methods
The positive-deviance research approach falls within traditional epidemiological methods of studying prevention, and it is applicable to all types of diseases and risk factors. However, positive-deviance studies differ from the majority of epidemiological studies in three respects.
First, most epidemiological research focuses on identifying the agents and pathways of transmission that cause disease states and on preventing exposure to these agents and pathways. Positive-deviance research applies to cases where exposure already has occurred. It focuses on identifying sources and pathways of natural immunity or adaptive resistance. For example, positive-deviance research on AIDS (acquired immune deficiency syndrome) would work to determine why some individuals who have positive antibody tests to the AIDS virus do not come down with the disease and how others who are antibody positive can benefit from the same sources of resistance.
Second, the majority of epidemiological studies have yes/no outcomes. Positive deviance research, particularly in nutrition, tends to have a continuous range of outcomes. In most epidemiological research the individual either has an event or condition or doesn't have it (e.g. stroke, infection, heart attack, accident, burn, cancer, infant death). Such clean distinctions lend themselves to relatively simple retrospective case-control study designs.
Positive-deviant children, with respect to nutritional status or childhood illness, such as diarrhoea, do not fall into such either/or classifications. They rank at the high end of a continuous distribution of nutritional status or the low end of a distribution of sick days with diarrhoea, for example. Case-control methods that pool together the children at the worst end of the distribution (cases) and contrast them with a pooled group from the top (controls), as was done by Clemens and Stanton (1986) in a study of infant diarrhoea in Dhaka, Bangladesh, may lose critical information about individuals ranked intermediate between the cases and the controls. They also lose information from the variability between children within the top and bottom groups.
For example, a study of maternal feeding behaviours, directed towards 9- to 20 month-old children in Mexico, showed a highly significant positive correlation between active maternal feeding behaviours and growth status, when growth status was measured as a continuous variable (height-for-age). However, when tall children in the top third of the distribution (controls) were contrasted to short infants in the bottom third (malnutrition cases), the difference in active maternal feeding behaviour between the groups was much less significant (Zeitlin and Johnson, in progress). This drop in apparent significance occurred because the mothers of the tallest children were relatively less active in feeding them than those of children in the middle range. Examination of the top and bottom groups only would not have discovered this important information.
Third, positive-deviance studies tend to require complicated designs and analyses, because many psychological and behavioural factors contribute to resistance. Traditional epidemiological methods establishing the links between disease agents and disease states can be relatively simple. The main requirement for testing the relationship between dietary fat and breast cancer, for example, is to obtain large amounts of data on two variables: dietary intake of fat and breastcancer incidence. To identify the factors that protect women with high fat diets from developing cancer (positive deviance) would be much more complex and would involve psychological testing.
Definition of terms and specification of research goals
Behaviours and social structures may be either informal - or without cultural rules or institutionalized services - or they may be formally prescribed in cultural rules or social structures. Positive-deviance research seeks to discover functional informal supports for nutrition and health that can be transformed, by means of programmes, into formal structural supports. It also seeks so strengthen formal supports, such as child support legislation, that provide nurturing environments in which wholesome informal social interactions will normally develop.
The Transformation of Behaviours into Practices
In the realm of personal behaviour, practices are formalized behaviours. Practices are clusters of informal behaviours for which common words and cultural rules already exist. A practice is based on normative prescription, whereas a behaviour can be idiosyncratic. The set of rules governing infant feeding and care can be referred to as practices. For example:
At a more macro-level, patrilocal family structure is the practice of living with the husband's parents after marriage.
The word practice is sometimes used to described what is done; and the word behaviour, how it is done. The term behaviour is much more general, however, than this would imply and may be considered in its broadest sense to encompass all actions, including practices. The how part can be dissected into any number of individual actions. A video tape of people in action can be stopped any number of times and any of these still frames can be described. Behaviour can be dissected into infinitely small segments. Quality of mood or intent may not be captured by such dissection, however.
In the case of positive-deviance research. we are not looking for infinitely small hits, which we could not deal with usefully, but for clusters or units of behaviour that are:
Where personal behaviours are concerned, the goals of positive-deviance research are to discover beneficial behaviours and styles of behaviours and transform them into practices that can be generally recommended and supported by nutrition, health, and social services. The specific cluster or units that we discover and test to see if they can profitably be transformed in this manner may be:
As an example of behaviours not examined before, a study of mother-infant pairs shortly after birth (Pollitt et al., 1977; Pollitt and Wirtz, 1981) found that mothers who interrupted the baby's bottle-feeding in order to clean milk from his mouth or body had infants who gained less weight during the first month of life. This is one of the types of behaviour sought by positive-deviance research. It should not be difficult to test whether weight gain in the neonate could be improved by teaching both bottle- and breast-feeding mothers not to interrupt the nursing infant, who should be permitted to terminate the feed himself. If effective, the behaviour "not interrupting or distracting the nursing infant" could become a practice taught by nutrition educators.
Another example of a behaviour that has the status of a practice in some cultures but not in others is the mother's role in feeding toddlers - whether she should actively feed the 18-monthold or place the food beside the child and let it feed itself. In Bangladesh (Rizvi et al., 1984) there is a cultural practice of not hand-feeding the child, in the belief that the child will digest better what it picks up and feeds itself. This is suspected of being one of the destructive, culturally dictated weaning practices mentioned in part 1 (behind which lie hidden agendas of limiting population growth and reducing body size). In most other cultures this is a behaviour to which no rules have been attached. Evidence from Mexico (Zeitlin and Johnson, in progress) demonstrafes that active feeding by the caretaker plays a significant role in positive deviance and that active feeding should be elevated to the status of a practice.
An example of a positive-deviant modification of a traditional practice comes from the research of D'Alois (1980) in Liberia, where there is a traditional taboo against sexual intercourse for the duration of the breast-feeding months, based on the belief that semen poisons the milk and makes the baby ill. In traditional farming communities where the extended family exerts social control over polygamy, this taboo guarantees birth spacing without endangering the continuing marriage bond between the parents, who resume sexual relations two or three years after each birth. In newly urban areas, polygamy is no longer governed by traditional rules. The new father who wanders off with another woman may not return. D'Alois found that mothers in periurban Monrovia had re-interpreted the traditional taboo to mean that a sexual relationship with the baby's own father was safe, but that semen from any man who was not the father would poison the milk. This new practice served to reinforce a monogamous lifestyle which is better suited to urban living than polygamy.
The Transformation of Informal Functional Social Support into Formal Structural Support
Historically, in all societies the extended family has provided informal social support for its dependent members, including mothers with infants and young children, the elderly and the disabled. There are many reasons why such informal support fails to function for certain individuals and households. The thinning out of families that occurs with the reduction in family size that goes with the demographic transition forces society to institutionalize most of the informal-support functions of the extended family.
Typically, resource-scarce communities in developing countries have lost many informal supports without yet evolving new formal social structures to replace them.
In many areas of the developing world, grass-roots voluntary self-help movements with political or religious ideological backing are working successfully, despite local poverty, to replace old structures with new systems. Many of these movements support village-level health and nutrition programmes. Where such community-based programmes do not exist, government and private health and social-welfare projects also seek programme designs that will maximize social outreach to their clients.
With regard to social support, the goals of positive-deviance research are: (1) to discover the types of structural and functional support that are most indispensible for maintaining adequate child growth and health amidst poverty; (2) to develop and test programme models that can reinforce or provide these supports; and (3) to devise legislation that strengthens such socialsupport networks both with and without programmatic interventions.
Three-stage research and pilot-project model
This paper proposes a three-stage model for research and programme development comprising: (1) basic social-science nutrition research; (2) behavioural analysis and trials, or concept testing, and operational research; and (3) pilot implementation. Behavioural trials and operational research are methods used to develop, test, and perfect procedures before putting them into operation. The historic sequence in programme development has been to include only steps 1 and 3 in relative or complete isolation from each other. Step 2, which links 1 and 3, is a recent development which has been applied to nutrition and health-care programming since about 1975. Separation between these steps often is unavoidable for administrative and other reasons.
These steps are considered together in this paper because, under favourable circumstances, they can be linked, and much time and effort can be saved. When the purpose is to develop effective applications, concept-testing, using behavioural trials, and operations research should in fact substitute for basic research whenever possible. Industry developed these procedures because they were more rapid and efficient than basic research methods. Commercially they have proven immensely profitable. When the three stages are considered as part of an integrated series of procedures, issues of a practical nature that do not require large samples to permit the derivation of statistical confidence intervals can be solved by concept-testing and the answers can be provided in forms that are immediately usable in pilot-project applications.
Some of the factors contributing to positive deviance are sufficiently well known for programmers to incorporate them directly into projects and policies without a pilot project stage. in the majority of cases a pilot is needed to solve remaining problems of service delivery, to ensure feasibility, and to evaluate these applications.
The three stages should achieve the following objectives.
First, basic research identifies beneficial child-feeding and health-seeking practices that currently operate to protect positive-deviant children in spite of severely limited household income, food resources, maternal time availability, and community service structures. This research concurrently analyses nutrition problems of the community using traditional epidemiological, functional classification, dietary, and other approaches where appropriate.
These more traditional methods must be applied in any case in order to determine whether factors hypothesized as contributing to positive deviance really do improve nutritional status after confounding variables have been taken into account. An expected product of these more traditional approaches is identification of changes in behaviour and in programme structures along PHC lines that will be recommended for concept-testing and operations research. In locations where much previous research has already been done, traditional research may be limited to those variables needed to control for confounding factors in positive-deviance studies.
Second. concept-testing and operational research develop transferable sets of beneficial behaviours, programme procedures, and community activities that blend local mothering wisdom and local social-support structures with the new insights embodied in the accepted strategies for enhancing child survival. The behaviours would be developed into communications messages and materials, taught and distributed during health/nutrition worker training programmes, and disseminated using mass media. The procedures would be incorporated into programme routines. New community programmes or support groups might be tried on a preliminary basis.
Third, pilot projects apply the messages, programme-design features, and support groups developed in stage 2. Evaluation of these pilot programmes determines whether the newly developed approaches are effective in improving nutritional status within the context of normal programming procedures. Preferably, the pilot projects are imbedded in the national maternal child health (MCH) and nutrition programme, so that expansions of successful methods can proceed through normal channels.
The three stages noted above are logical phases that may overlap each other in practice, as in the positive-deviance research project recently conducted by Tufts University under the direction of the first author of this paper in collaboration with the Bangladesh Rural Advancement Committee (BRAC) and the Institute of Nutrition and Food Science (IFNS) of the University of Dhaka, in Bangladesh.
The Bangladesh study on diarrhoeal infection and malnutrition in 9- to 18-monthold infants selected matching basic research (stage 1) and behavioural trial (stage 2) sites from the beginning. The study began by rapidly collecting and analysing baseline data from both sites. This (simplified stage 1) analysis identified about 15 behavioural factors statistically associated with diarrhoeal infection and malnutrition in infants.
The results of the baseline were used to design a detailed 10-month study of behaviours, growth, diet, and morbidity (sophisticated stage 1) at the research site and a parallel behavioural trial, message development, and intervention project (stage 2) focused on diarrhoeal disease control at the trial site. A year after the baseline survey a cross-sectional resurvey compared health and nutritional status and knowledge, attitudes, and behaviours at the two sites, with the research site serving as a control for the intervention site. The intervention was found to improve hygiene knowledge and behaviours and reduce diarrhoeal infection (Ahmad and Ahmad, 1987).
During the following year messages from the diarrhoeal-disease control trials were incorporated into BRAC's primary health-care programmes (stage 3), and a second round of behavioural trials and concept-testing on supplementary feeding and other MCH nutrition messages took place at the trial site. Results of the year two data analyses will feed back into the development of new messages and interventions.
In this project the reason for not waiting for the results of carefully conducted basic research (sophisticated stage 1) before starting the concept-testing, behavioural-trial phase was that many of the simple baseline survey stage 1 results already were actionable. For example, the majority of infants were found to be crawling on dirt surfaces contaminated with animal and human faeces. They were also found to receive inadequate amounts of supplementary foods until the age of two years or later. Both of these conditions were known to be undesirable without need of further research results.
The comments here on research design are not intended to take the place of meeting with a statistician to work out the design best suited to the research. They are tips to fellow investigators based on experience in the field and in data analysis.
Design
For very low cost studies, a retrospective, case-control design probably is best, comparing wellnourished children with malnourished children matched for age. Such studies may either be indepth small-sample enthnographic investigations or larger sample cross-sectional surveys. In either case, they should use forms that combine questions to the child's caretaker with direct observation of the environment and of caretaking behaviours. The advantages of the retrospective case-control model is that the data can be collected and analysed rapidly, by hand if necessary. The design is simple and clear, so that the purpose of the study and its results will be clear to everyone involved. Such studies should yield an understanding of important factors associated with positive deviance. As noted earlier on pages 81-82, they will not reveal all the information it would be useful to know.
The next step up, at a slightly higher level of expense and sophistication, is a crosssectional study design that covers a representative sample of all of the children in the age-group and the socio-economic groups of interest. This model tends to take the form of a survey of at least 80 to 100 households, because the sample needs to be big enough to make statistical comparisons. It also requires an observational component to obtain behavioural and environmental data. One advantage of including children whose nutritional state is average is that it becomes possible to describe the full distribution of values within the population. The descriptive information can also be used as a baseline for designing interventions. Collecting a full continuum of values also permits multivariate regression, which is the analytic procedure most frequently ape plicable to nutritional data.
Longitudinal studies are desirable but many times more expensive and time consuming than cross-sectional designs. They require advanced computer capabilities for analysis. Funding agencies and researchers frequently do not take into account the fact that collecting and analysing six months' worth of data may cost up to ten times as much as the amount required to conduct and analyse a cross-sectional survey that collects the same information only once. Reasons for the additional expense include: extra time spent in designing longitudinal measures; the need for a long-term field office and full-time staff; and extra time and expertise needed for all aspects of the data analysis, as will become apparent in the section on measuring growth (p. 97).
Qualitative versus Quantitative Designs
Qualitative methods are needed to identify the types of behaviours and social networks that exist in households and communities. These methods include long openended interviews and observations as well as group interviews, known as focus groups. A problem is that results of qualitative studies tend not to be replicable in a strict scientific sense, and do not provide descriptive statistics. Policy-makers frequently do not regard qualitative results alone as sufficient evidence to justify a programme.
Therefore, it is usually desirable to combine qualitative and quantitative methods in positivedeviance studies. While this can be done in a variety of ways, the simplest and least expensive way to combine qualitative and quantitative research is to start with a three-week qualitativeresearch module, using a rapid-appraisal methodology (Scrimshaw and Hurtado, 1987). The results of the rapid appraisal are written up on their own. Simultaneously, the rapid-appraisal exercise functions as the design phase for developing and pre-testing the questionnaires and observational protocols to be used for quantitative-data collection.
For example, this qualitative module may have the following steps:
Sample Size and Unit of Analysis
Observational methods are frequently too resource-intensive to permit the collection of data from large samples. In basic ethnographic studies, for example, one observer may cover only 20 households per year (20 working days per month; one day per household per month). A full year may be needed in order to record seasonal changes and the sequential development of the child.
An obvious answer to the problem of insufficient sample size is to turn to behaviours or events as units of analysis. Instead of 25 infants, for example, one may wish to analyse the 350 feedings received by these infants during the observation period. Moreover, the feeding may be the theoretical unit of interest if one is studying the effects of the interactions or the amount of food eaten during the feeding. In moving to events or behaviours as units of analysis, it is imperative to involve a statistician to figure out whether the procedures used are statistically legitimate. Unfortunately, feeding events may not be stable units of analysis. Behaviours summarized at the event level in the Mexico study turned out to be highly non-normal in distribution (Zeitlin and Johnson, in progress). Different statistical methods used to cope with their non normality tended to yield different statistically significant results. Thus, the behaviours were finally summarized and analysed at the child level despite the small sample size.
It is also important to distinguish between statistical significance and biological significance. In the same Mexico study, the fact that the 25 well-nourished infants received on average oneand-a-half more bottles of cow's milk per child per day than the 25 malnourished was not statistically significant, but could well have been biologically significant.
Accuracy and Replicability of Observational Research
Replicable observational research must be based on structured methods of observation. Developmental psychologists and anthropologists have devised and tested numerous formats for structured observations. A time-saving strategy is to adapt and pre-test some of these formats during the same period when the qualitative research is being conducted. Once these new instruments have been developed, it is essential to test them for internal reliability and validity.
A conceptual framework for the design of positive-deviance studies
This section discusses the co-variables that must be considered in research on positive deviance in nutrition in the context of a general conceptual framework. This framework represents a broad statement of the causal relationships at work in the environment. It is presented because a general understanding of the nutrition system is required for the design of studies of this nature.
The variables under investigation are listed below in categories I to 4. Figure 12 represents the interactions of these categories or constructs. Box 1, parental and household characteristics, strongly affects boxes 2, parent's child-care "wisdom," and 3, child's characteristics, and also influences variables in box 4, child's nutrition and health status. Boxes 2 and 3 both interact and individually contribute to the outcome variables in box 4. Social and environmental factors impacting on the household have been subsumed under household characteristics.
The exact subset of variables within these constructs that will be selected for intensive study and the ways in which they will be operationalized will depend upon the area where the study is conducted and the focus of the study. The primary outcome variable is growth status, determined by anthropometric measurements. Indirect measures of the child's nutrition and health status will include dietary intake, morbidity experience, and developmental status.
Parental and Household Characteristics (Box 1 )
Parent's Characteristics, Health, and Nutritional Status
Each of these factors may affect the child's growth either directly or indirectly. For example, a mother's health and nutritional status will not only determine a child's size at birth but will also affect the amount of energy a mother has to devote to the care of
her child. Ill-health of the father, including conditions such as alcoholism, will have severe effects on the family's resources. A mother's age and childbearing history will affect her knowledge, attitudes, and behaviour toward child care. An older mother might provide better child care because of her experience; she may also have less time to devote to a new infant. The number of children that parents have, along with the length of the birth-spacing interval, also influences the child's chances of survival and nutritional status.
Parent's Educational /Psychological Status
Each of these variables is likely to influence the parents' child-care knowledge, attitudes, and behaviours, and thus ultimately affect the child. A mother's level of education has been shown to correlate with child mortality (Caldwell, 1981) and with child nutritional status. It is possible that education increases the parents' ability to deal with new ideas and judge the seriousness of illnesses. Radio and television can also be an important source of information concerning child care. Studies reviewed in part I have also indicated how certain personality characteristics and attitudes of mothers are associated with their tendency to have malnourished children.
Stress Factors in Parents' Lives
Certain stressful events, such as those outlined above, are likely to influence the parents' ability to provide good child care. Stressful incidences can influence both character and mood. A study in Chile found that mothers with malnourished children had a higher degree of dissatisfaction with their family life (Alvarez et al., 1982). In Jamaica, mothers of malnourished children have been shown to lead stressful, disorganized lives (Kerr et al., 1978).
Household Structure and Socio-economic Status
Austin and Zeitlin (1981) reviewed an extensive literature from developing countries attesting to the correlations between the above indicator categories and malnutrition.
Parent's Child-care "Wisdom ": Knowledge, Attitudes, and Behaviours (Box 2)
Nutrition and Health
The parents', and particularly the mother's, nutrition knowledge and practices determine what a child eats and thus affect the child's nutritional status and growth. It is important to investigate both traditional and new food practices, since both can be beneficial as well as harmful. According to Chavez et al. (1971), mothers of male nourished children in rural Mexico tend to carry out more traditional feeding practices. However, it is important to determine which of these traditional practices are harmful. The intrafamilial distribution of food is also important in determining the quality and amount of food that a young child receives. In Barbados, young children did not receive enough food although sufficient supplies were available (Pan American Health Organization, 1972).
A child's growth is not affected only by the amount and type of food eaten but also by disease. Gastrointestinal infections such as diarrhoea cause less food to be absorbed and can thus lead to a dramatic weight loss. A mother's health-care knowledge and practices will affect her child's health by making him more or less susceptible to diseases. For example, in Mexico the practice of boiling water seems to be strong indicator of a household's health status. Both traditional and new health-care practices should be investigated in order to identify those that are beneficial and those that are harmful. For example, the "hot and cold" classification system should be considered in Mexico, in order to determine which traditional beliefs are still held. The idea that it is important to health to maintain a proper balance of hot and cold foods is no longer believed by everyone (Molony, 1975). The use of traditional healers and modern health-care facilities is also an important consideration. Fathers often determine when the child is taken for treatment.
Mental and Behavioural Development
A mother's child-care knowledge and behaviour can greatly affect her child's mental and behavioural development, which is linked to the child's growth. The father may affect the child through direct interaction, or, primarily, through his economic and emotional support of the family unit. It is thus important that parents provide their infants with adequate care for optimal development.
Child's Characteristics (Box 3)
General Characteristics
This information is likely to influence the kind and amount of care that the child receives. In Mexico, a child's sex is important in determining the amount of food received. A study (Chavez et al., 1971) carried out in poor rural areas of Mexico suggested that far more girls were malnourished than boys. As a general rule in developing countries, children of higher birth order tend to be malnourished. The child's birth order can also influence the caretaker; the last-born child is sometimes the most spoiled.
Child's Nutritional and Health Status (Box 4)
Each of the variables in this box measures an aspect of the child's nutrition and health status. They will be investigated as dependent variables to determine whether the child's status is affected by the parent's background and " wisdom." It is also important to consider the degree to which the child's nutrition and health status influence his treatment. A child's health, physical appearance, or behaviour influences the care and food received by that child.
