**SPIDER**

Here is a simple spider to show some of these relationships involving immunization:

This is just one model that could be
considered. In fact, there could easily be more variables added, or fewer. Remember when
choosing the variables for testing that it is most important not to choose variables on
the causal pathway, but to try to select variables that are at the same level. As a
reminder, look again at the conceptual framework for **CAUSES OF MALNLUTRITION**. The method for analysis is basically the same. The idea is to
first look at the relationships between the potential causal variables to determine if
there will be likely be interaction or confounding in the model and then to set the model
up to test the relationships with the outcome variable (nutrition status).