Search results

AbstractRaking is widely used for categorical data modeling and calibration in survey practice but faced with methodological and computational challenges. We develop a Bayesian paradigm for raking by incorporating the marginal constraints as a prior distribution via two main strategies: (1) constructing solution subspaces via basis functions or the projection matrix and (2) modeling soft constraints. The proposed Bayes-raking estimation integrates the models for the margins, the sample selection and response mechanism, and the outcome as a systematic framework to propagate all sources of uncertainty. Computation is done via Stan, and codes are ready for public use. Simulation studies show that Bayes-raking can perform as well as raking with large samples and outperform in terms of validity and efficiency gains, especially with a sparse contingency table or dependent raking factors. We apply the new method to the longitudinal study of well-being study and demonstrate that model-based approaches significantly improve inferential reliability and substantive findings as a unified survey inference framework.

Panel studies typically suffer from attrition. Ignoring the attrition can result in biased inferences if the missing data are systematically related to outcomes of interest. Unfortunately, panel data alone cannot inform the extent of bias due to attrition. Many panel studies also include refreshment samples, which are data collected from a random sample of new individuals during the later waves of the panel. Refreshment samples offer information that can be utilized to correct for biases induced by non-ignorable attrition while reducing reliance on strong assumptions about the attrition process. We present a Bayesian approach to handle attrition in two-wave panels with one refreshment sample and many categorical survey variables. The approach includes (1) an additive non-ignorable selection model for the attrition process; and (2) a Dirichlet process mixture of multinomial distributions for the categorical survey variables. We present Markov chain Monte Carlo algorithms for sampling from the posterior distribution of model parameters and missing data. We apply the model to correct attrition bias in an analysis of data from the 2007–08 Associated Press/Yahoo News election panel study.

Care coordination may be more challenging when the specific physicians with whom primary care physicians (PCPs) are expected to coordinate care change over time. Using Medicare data on physician patient-sharing relationships and the Dartmouth Atlas, we explored the extent to which PCPs tend to share patients with other physicians over time. We found that 70.7% of ties between PCPs and other physicians that were present in 2012 persisted in 2013, and additional shared patients in 2012 increased the odds of being connected in 2013. Regions with higher persistent ties tended to have lower rates of emergency room visits, and regions where PCPs had more physician connections were more likely to have higher emergency room visits. The results point to potential opportunities and challenges faced by health care reforms that seek to improve coordination.

Abstract
Multiple imputation (MI) is a popular and well-established method for handling missing data in multivariate data sets, but its practicality for use in massive and complex data sets has been questioned. One such data set is the Panel Study of Income Dynamics (PSID), a longstanding and extensive survey of household income and wealth in the United States. Missing data for this survey are currently handled using traditional hot deck methods because of the simple implementation; however, the univariate hot deck results in large random wealth fluctuations. MI is effective but faced with operational challenges. We use a sequential regression/chained-equation approach, using the software IVEware, to multiply impute cross-sectional wealth data in the 2013 PSID, and compare analyses of the resulting imputed data with those from the current hot deck approach. Practical difficulties, such as non-normally distributed variables, skip patterns, categorical variables with many levels, and multicollinearity, are described together with our approaches to overcoming them. We evaluate the imputation quality and validity with internal diagnostics and external benchmarking data. MI produces improvements over the existing hot deck approach by helping preserve correlation structures, such as the associations between PSID wealth components and the relationships between the household net worth and sociodemographic factors, and facilitates completed data analyses with general purposes. MI incorporates highly predictive covariates into imputation models and increases efficiency. We recommend the practical implementation of MI and expect greater gains when the fraction of missing information is large.