| |
|
|
|
The goal in analyzing data here is to answer relevant questions about malnutrition -- its distribution, possible causes and correlates, and related issues -- in order to design, monitor, or otherwise assist in activities to improve the situation. Examples of questions that you should become familiar with tackling are:
 |
Who are the malnourished? |
|
What services and programs do the malnourished have and what are the unmet needs? |
|
What causes of malnutrition can be addressed
(often by service delivery and community based programs)? |
|
A further question is often relevant, if more complex: what are the contextual factors that modify the effects of programs and services? |
Throughout the analysis, posing specific questions (like those just shown, but often more detailed -- write them down!) is essential. Propose a research question of interest, e.g. to explore a possible causal pathway, and always keep the question in mind. Keep track of progress towards finding an answer at each step of the data analysis, and carry on until an answer is found. If you get lost, remind yourself of the original question.
Who are the malnourished?
This is a fundamental first question and there is a series of steps to answer it.
|
First, a situation analysis can give an
impression of the overall numbers and prevalence of the malnourished. This can be seen
with frequencies of single variables, usually nutrition outcome variables (anthropometric
and measuring micronutrient deficiencies) and background variables such as education
level, socio-economic status (usually as proxies, such as housing quality), access to
health care, and water and sanitation conditions. The situation analysis is typically
one-way, showing a series of indicators in the country overall, or by a classifying
variable, such as by biological status (age, gender, pregnant/lactating), or by areas
within the country. In fact, as the situation analysis is broken down (or
'disaggregated') from national to area level, it readily gives pointers to targeting
priorities, and helps to get an overall first impression. |
|
Second, trends showing changes in the situation over
a period of time tell a more powerful story. A trend is measured through data collection
at more than one point in time to give estimates of change in extent and type of
malnutrition, access to care, prevalence of disease, etc. Trends are measurements of
change over time. Caution is needed to avoid making improper comparisons due to
differences in data types. |
|
Third, reassembling existing data can help to create
new categories that will give a greater depth to the analysis. It can simplify the data to
show the prevalence of malnutrition by different groupings, to examine the consistency of
different indicators, often by biological status, geographic area, and so on. |
|
Fourth, diversity is readily demonstrated in the the data when reassembling. There are almost always differences in the prevalence of malnutrition in different areas (districts, provinces, divisions, etc). As a rule of thumb, at least a doubling of the prevalence is usually seen between better and worse-off areas. This can be important for advocacy, making the point that malnutrition affects different groups differently, that the better-off are better nourished, and that a lot can be done about malnutrition (this point is easier to make about micronutrient deficiencies, but the diversity is still worth showing). |
To what services do the malnourished have access?
This question is especially relevant to program planning and evaluation. The data may be from a cross-sectional (at-one-time) survey, from growth monitoring or clinic data, from management information systems, and other data sources. Not only should access be measured but where possible also trends in access. Here are some key concepts as they apply to an area-specific or national program:
|
Coverage - refers to numbers with service or using
the program, and % out of the entire population targeted; may also refer to the
number and % in need (e.g., malnourished) covered in the population. |
|
Targeting - how far the program is actually, or is
planned to be, targeted towards the malnourished (or otherwise needy); can be estimated as
the prevalence of malnutrition in the covered group in relation to the prevalence in the
overall population. |
|
Intensity - resources applied by the program: can be
measured as $ per head, field workers (e.g., mobilizers -- village workers, often
volunteers) per population covered, facilitators (i.e., supervisors, often government or
agency employees) per mobilizer, and other similar indicators. |
|
Program content - in relation to causes of malnutrition, what activities to include in the program; for on-going programs, relevance is the concept, referring to whether the activities are addressing actual causes of the problems of malnutrition they are intended to solve. |
Of these, coverage, targeting and intensity use mainly one-way analysis. Determining program content and relevance requires assessment of causality, so the analysis will use one-way, two-way, and multi-variable methods, and addresses the question:
What causes of malnutrition can be addressed ?
This question relates to the activities that should be, or are being supported, by a program, and their relevance to malnutrition. It may also refer to contextual factors -- the status of women for example, or education and literacy -- which are less easily addressed by projects but may profoundly affect both program design, and success. In the nutrition policy and program planning area, these may be addressed by 'supporting policies' needed to make programs succeed -- but that's outside the scope here. What is within the scope is that such factors usually need to be examined as potentially modifying a program design because of their interactive effects; they are also relevant to targeting. One way in which possible causal factors and their interrelations can be mapped out is by sketching causal factors and nutrition outcomes in box-and-arrow diagrams, sometimes called 'spiders' for reasons of anatomical similarity.
EXAMPLE SPIDER: INFLUENCES OF CARETAKER COMPETENCE
(Source: Martorell and Habicht ...)
These diagrams can be as simple as showing relations with one central factor -- caretaker competence here -- but can grow far more complex as new variables are added, such as the influence of the level of economic development, education, and so on.
When you are building a causal diagram, it is helpful to use the UNICEF Conceptual Framework on the causes of malnutrition to give general hierarchy of the causes, and a common language or starting point for discussion.
Frequent reference to the conceptual framework is important in analyses that explore causality. For example, methodologically one would usually not include two variables on the same causal pathway in a model; but one does control for variables at the same level of causality. Again, interactions can usefully be specified in the spider before estimating them. This will be further explained in the two-way and multi-way sections.