ABSTRACT
In the previous chapters we have confined our attention to the simplified situation where the primary exposure variable of interest is limited to two levels: exposed (E) or unexposed ( ¯E). Such a crude assessment is rarely an effective measure and does not allow us to develop a detailed understanding of how disease risk changes over subtle shifts in exposure. Fortunately, even the most basic exposure measurements allow a finer level of gradation; for example, in capturing coffee drinking, the data of Table 7.2 have exposure measured by average daily consumption in cups per day with four distinct categories. We now want to expand the methods to look at more complex risk factors that allow for several differing levels of exposure. For the moment we do not draw a distinction between exposure variables where these levels may be ordered and those for which there is no natural ordering. This issue will be of considerable relevance, however, in Section 11.3.
