Application of odds ratio and logistic models in epidemiology and health research

M any researchers in the health field use the chi-square statistic to identify associations between variables. This edition of research notes will demonstrate that the odds ratio may be a preferred analysis to yield more useful and meaningful results. In epidemiological and health contexts, the outcome variable is often discrete, taking on two (or more) possible scores. For example, the outcome for cardiovascular diseases may be the presence or absence of the disease. Suppose we are interested in whether smoking is associated with cardiovascular diseases in a hypothetical group. We can construct a 2 x 2 crosstabulation table to examine the problem (Tabl