In regression analysis, random errors in an explanatory variable cause the
usual estimates of its regression coefficient to be biased. Although this problem has
been studied for many years, routine methods have not emerged. This thesis
investigates some aspects of this problem in the setting of analysis of epidemiological
data....