Yu, Tzy-Chyi2010-07-142010-07-142008-08https://hdl.handle.net/11299/91992University of Minnesota Ph.D. dissertation. June 2008. Major: Health Services Research, Policy and Administration. Advisor: Dowd, Bryan E. 1 computer file (PDF); viii, 191 pages.Objectives. Little attention has been paid to the connection between perceived access barriers to conventional medicine and CAM usage. The purpose of this dissertation is to quantitatively evaluate the associations between perceived access barriers to conventional medicine and CAM usage using population data. Method. This is a cross-sectional study using the 1999 and 2002 National Health Interview Survey. The target population is U.S. civilians age 18 to 64. The outcome variable is CAM usage classified into homogenous subgroups. Perceived access barriers to conventional medicine are grouped into distinct dimensions including affordability, accommodation, or accessibility. To deal with potential selection problems related to perceived access barriers to conventional medicine, recursive bivariate probit models are utilized to test the impact of individual perceived access barriers on CAM usage after controlling confounders. Moreover, multivariate probit and Boolean probit models are used to evaluate all three distinct perceived access barrier dimensions simultaneously on CAM usage. Results. In general, perceived access barriers to conventional medicine are positively related to CAM usage, to varying degrees. The various relationships found in this dissertation between each perceived access barriers to conventional medicine and CAM subgroup usage show a clear picture that neither perceived access barriers to conventional medicine nor CAM is homogenous. Moreover, several bivariate probit models are significant, indicating that perceived access barriers to conventional medicine are endogenous in these models. It can be seen from model comparison that serious biases arise from simple probit models when these variables of interest are endogenous. Furthermore, the impact of three perceived access barrier dimensions on CAM use differs in multivariate probit and Boolean probit models, each of which has its assumptions, as such strengths and limitations. Conclusions. Ignoring the endogenous nature of perceived access barriers to conventional medicine will bias estimates of the relationships between perceived access barriers to conventional medicine and CAM usage. People who perceive access barriers are likely to use CAM, especially perceived accommodation access barriers. These access barriers should be addressed to ensure that people have adequate access to health care services, both conventional and alternative, to improve their health.en-USBivariate probitMultivariate probitBoolean probitEndogeneityNational Health Interview SurveyConventional medicineComplementary and alternative medicineAccess to careBarriers to careThesis or Dissertation