To motivate the discussion of weighted least squares methodology, consider the following example. Epidemiologists investigating air pollution effects conducted a study of childhood respiratory disease (Stokes 1986). Investigators visited groups of children two times and recorded whether they were exhibiting symptoms of colds. The children were recorded as having no periods with a cold, one period with a cold, or two periods with a cold. Investigators were interested in determining whether sex or residence affected the distribution of colds. These data are displayed in Table 13.1.
|Periods with Colds|