The rates of herbal medicine use were assessed for the sample as a whole and by sociodemographic and healthrelated measures. Differences in the rates across independent variables were assessed using the chi-square statistic. Logistic regression was then employed to model the use of herbal medicine, first with sociodemographic characteristics and then for each of the health-related measures, controlling for the influence of the sociodemographics. All analyses incorporated weighted data that were adjusted for design effects
to produce results that were representative of the older Mexican- American population in the five-state region. Analyses were conducted using SAS (17), and confidence intervals were computed using SUDAAN (18).