ABSTRACT
The easiest way of carrying out an analysis which takes several exposure levels into account is to analyze each exposure level separately. This means that the individuals in the highest exposure category are compared with the unexposed, then individuals in the next highest category, and so on for all categories of the exposed. When relative risks are compared across different exposure categories it is naturally important that they should be free from confounding. One way of ensuring that the weights really are the same for all exposure levels is to standardize, rather than to pool with weights that reflect the stratum-specific precision. It is often satisfactory to analyze each exposure level separately if one has a limited number of levels. There is no fundamental difference between dividing a material into exposed and unexposed and dividing it into two levels of exposed as well as into unexposed.
