Page  2

Types of Variables

Most of the collected data used in nutrition related analysis fall into one of the following categories:

1. OUTCOME Variables - measurements of the current status of the population  -- 'manifestations' in the UNICEF framework. These might include mean continuous scores of a measurement (e.g., height or arm circumference in cms; derived indices such as WAZ score; micronutrient deficiency measures such as serum retinol or haemoglobin) or prevalence of categorized outcomes (e.g., below -2 SD weight for age z-score; below 12g/dl haemoglobin; with/without goitre). For a more detailed description of outcome variables, click on the outcome variable link at the start of the paragraph. Typical nutrition outcome variables are:

   Anthropometry:
   Some examples of where to look for data are: HH surveys, clinical growth monitoring or school data, or weighing programs.

   	Birth Weight
   	Underweight (weight for age z-scores)
   	Stunting (height for age z-scores)
   	Wasting (weight for height z-scores)
   	Middle Upper Arm Circumference (MUAC)

   Indices such as weight, height, and age, are compared to the values of a reference population to see if they are worse than expected from the reference.  For the same age child, the height or the weight is compared to the height and weight of the reference population.  Chapter 6 Submodule takes you through hand calculations using the reference tables which have been calculated from the data collected by the National Center for Health Statistics (NCHS).

   The "Z-score" of a child uses the standard deviation of the reference distribution for a given measurement.  The index expressed in z-scores represents the difference between the observed weight and the median weight of the refernce population at the same age.
               Weight/height index = (Observed weight - Reference weight)
                                              Standard Deviation
   
                  

   Micronutrient Deficiencies:

   	Iodine -measure total goitre rate in school children; urinary iodine concentration, salt iodization surveys, and iodization of salt as a proxy of iodine deficiency
   	Vitamin A - clinical measure of night blindness (a word in the local language for nightblindness may be an indication that the community is aware of this clinical sign), Bitot's spots, serum retinol <70 ug/l, food frequency questionnaire, vitamin A capsule distribution
   	Iron - hemoglobin or hematocrit readings, anemia, distribution of ferrous sulfate, consumption of absorption inhibitors like phytate and enhancers like vitamin C

   For more details, click on 'outcome variables' at the start of this section.

2. CLASSIFYING Variables - these variables identify groups within a population based on biological, social, physical, political, economic, or other characteristics. They are used for targeting, sub-dividing the population when exploring causality, etc.  Examples are:

   	Area
   	Gender
   	Age
   	Occupation
   	Distance from health post
   	Rural/Urban location
   	Education level (e.g.,of mother); literacy
   	Socio-economic status (usually as proxy, like housing, assets)
   	Water source, sanitation (e.g., latrine)

3. DETERMINING Variables are usually interchangeable with classifying variables; the difference is in interpretation.  These variables relate to potential intervention options. Some examples are:

   	Breastfeeding behaviors - breastfed in first hour, exclusive breastfeeding, age breastfeeding ceased
   	Complementary feeding - age complementary feeding began
   	Immunization - age of immunization, individual types of immunization, percent receiving immunizations
   	Water and sanitation - access to safe water source, habits for water use, hand washing, food washing, access to latrine, type of latrine, etc.

   IMPORTANT REMINDERS

   Look for interactions between variables that might show different effects at different levels and also for confounding, which can mislead in the interpretation of results.

   Conceptually and methodologically, classifying and determining variables are similar and differ mainly in interpretation.

   ---

4. PROCESS Variables are used for following the process of program implementation.   Intermediate actions of the program can be compared to outcomes along the way and at a later point in the program implementation. Process variables are generally related to program delivery in this context, such as:

   	Percent receiving immunizations
   	Coverage of media message
   	Number of facilitators trained
   	Percent of children exclusively breastfed from 4-6 months

   Note: These will be numerous and varied depending on the goals of the program.   They may also be tested analytically in the same way as outcome variables. For example, you might have to know the distribution by area of immunization coverage, or the correlates of exclusive breastfeeding. 

   ---

   These variables types:

   	OUTCOME
   	CLASSIFYING
   	DETERMINING
   	PROCESS

   ... are often interchangeable depending on the precise research question being studied.   If this question is explicitly stated, confusion can be avoided!!

Return to top of page