Grant: Discovery Grants Program - Individual

The primary goal of the proposed research program is the development of methodologies for analyzingcomplex survey data, genomic data, genetic networks and forest fire data. As observational studies, surveys arecommonly used in many disciplines, such as health sciences and social sciences. Every year, the private sectorand government agencies invest lots of resources to conduct many analytical surveys. The collection processesfor surveys result in correlated data structures that provide challenges in analysis beyond that seen in othertypes of data. Traditionally, surveys are designed to provide descriptive statistics; however, in recent decades,survey data has been used for analytical purposes such as regression analysis. Hence, it is incumbent onresearchers to develop more methodologies to meet theoreticaland empirical needs. In my short-term project, I will provide empirical solutions to the bias problem forgeneralized multilevel models that are particularly suitable for analysis using survey data. Development ofempirical inferences will result in not only the variance estimates of unbiased estimators, but will also establisha new resampling technique which can be applied in other problems in survey sampling. Research in quantileregression techniques for survey data will fill in both the theoretical and empirical gaps in the literature.Facilitated by my short-term projects, my long-term project will show more theoretical results and practicalsolutions for analyzing survey data with stochastic regression models.