| Health disparities associated with race, ethnicity, and socioeconomic status are consistently observed in mortality, morbidity, and other indicators of health. The slow progress in eliminating disparities in health demand that researchers consider new methods for studying these problems. Community based research has emerged as a promising new approach in the evolving field of public health as the missing link in the progression from research to practice. As such, community based interventions have the potential to be powerful tools in reducing health disparities.;Stepping outside the realm of the traditional scientific research model, and developing models that consider cultural and social factors to learn what works to prevent illness, raises a whole new set of statistical questions. These questions are often complex and diverse in their statistical methodological areas. In this dissertation, we propose new methods to answer statistical questions arising from the analysis of data from community based interventions and health disparities research.;The purpose of Chapters 1 and 2 is to develop methodologies that facilitate evaluation of multiple risk factor cancer inventions. In Chapter 1, we use conditional logistic regression to identify an optimal linear combination of multiple behavioral risk factors for cancer in a working class multi-ethnic population. In Chapter 2, we test the linearity assumption and propose methods to test the functional form of the covariates in the conditional logistic regression model, these methods are based on non-parametric smoothers. The methodology developed in Chapter 3 estimates the trend of the hazard function and when the changes in trend occur. Methods, like the ones proposed in this chapter, that estimate the overall trend of the hazard function for an entire population allow researchers and clinicians a better understanding of how changing medical practice affects the survival experience for a patient population. In Chapter 4, we propose a mixture model for ordinal outcome data with a longitudinal covariate that is subject to missingness. The proposed methodology is developed to examine the effectiveness of a tailored, telephone delivered, smoking cessation intervention for construction laborers. |
