|Community Assessment of Natural Food Sources of Vitamin A, Guidelines for an Ethnographic Protocol (International Nutrition Foundation for Developing Countries - INFDC, 1997, 141 pages)|
This manual is designed for collecting sufficient data on diet and food use patterns to enable the design of an intervention program to improve vitamin A in the area to which the data apply. Policymakers and planners at various levelsin the health ministry, other interested ministries, and NGOswill want to be assured that these data are credible and useful for planning a vitamin A intervention. Your research report should describe the ways in which these data meet standards of reliability and validity at a level sufficient for developing an intervention.
The following are some typical questions that administrators and other policy makers may raise concerning the type of data produced by using this manual.
i. How do you know that these data are valid and reliable?
The major strength of this type of ethnographic research is in the idea of triangulation. The concept of triangulation refers to the fact that the important data are obtained from multiple sources. That is, specific points about food use behaviors, attitudes, and beliefs are obtained from in-depth interviews with key-informants and also from the structured interviews with samples of respondents. For some items, the researchers' direct observations further strengthen the level of evidence.
When there are discrepancies or conflicts between the different sources of information, the researcher explores those points in more depth with key-informants. For example, if a key-informant says that "people don't eat _____________(vitamin A-rich food), even though it is available...," the food frequency data from the sample of families provides a check on the accuracy of that statement.
ii. Isn't the sample of twenty-five or thirty respondents too small to permit any generalization about a community or population?
The suitability of sample size depends very much on the degree of homogeneity or consistency in the phenomena studied. Generally speaking, people's food behaviors and diets in most areas are strongly patterned. Meal patterns and diets in developing countries, especially in rural areas, are very consistent, or even monotonous as compared with food patterns in industrialized regions. Knowledge about food, as well as attitudes about it, are also quite consistent, though we find differences between the wealthier and poorer families and also between different ethnic groups. The manual is designed to be used primarily in communities that are ethnically (culturally) homogeneous. In areas with multiple ethnic groups, the sample size will have to be increased.
People in given communities, particularly in rural areas, not only share cultural values and behavior patterns to some extent, but they are also under the same ecological constraints. They buy food from the same markets and stores. They have generally similar crop-growing conditions. All those environmental factors contribute to the tendencies toward food-use similarities, even among ethnically distinct groups. However, the researchers using this manual should be alert to main sources of intra-community variations in dietary practices and other features. All communities have some systematic variations in cultural beliefs and behaviors. Key-informants often mention such local variations (e.g., caste differences in Indian villages) and researchers should probe for those local differences. Even with small sample sizes we can get a sufficiently accurate assessment for intervention planning purposes.
iii. How strong in reliability and validity do our data have to be for planning effective intervention programs concerning food use and dietary practices?
Of course, program planners and policymakers should insist on having really solid data for developing intervention programs concerning vitamin A, as in all other program planning. On the other hand, many intervention programs are launched with very little advance data-gathering. It is important to establish a realistic middle ground by insisting that any intervention planning should be preceded by a reasonable, economically feasible data-gathering stage.
On the other hand, the background data for planning intervention programs do not need to be statistically and epidemiologically justified. Dietary intervention programs are continually in contact with the people from whom the original background data were collected. Continued data monitoring during the program provides an opportunity for upgrading the background data and correcting misinterpretations. Therefore, all intervention programs should have built-in data-gathering, along with other programmatic activities.
iv. What about the generalizability of these kinds of data to other regions?
Any data, whether from ethnographic studies or carefully quantified surveys, are, strictly speaking, generalizable only to the populations within which the studies were made. On the other hand, we expect main features of the research results to be applicable to communities and regions that are basically similar to the original research site, in terms of the cultural backgrounds of the people and the ecological/economic setting in which they live. For that reason, the communities selected for studies using this manual should be broadly representative of the region or province in which vitamin A interventions are planned.
Program planners are generally aware of the main ethnic (cultural) variations in their regions. They also need to pay attention to the main ecological zones, with different crops, different food resources, different relationships to markets, and commercial food distribution systems.
Because of the high degree of cultural patterning of food practices (mentioned above), program planners can expect to apply the data from this research to other communities in the same ecological region, provided the people are of the same general cultural background. Where minor dialectical and subcultural differences are found, it is often possible to make some adjustments to the data, based on small numbers of key-informant interviews.