Georgia Southern Examines Markov Chain Monte-Carlo Methods for Missing Data Under Ignorability Assumptions
Missing observations are a common occurrence in public health, clinical studies and social science research. Consequences of discarding missing observations, sometimes called complete case analysis, are low statistical power and potentially biased estimates. Fully Bayesian methods using Markov Chain Monte-Carlo (MCMC) provide an alternative model-based solution to complete case analysis by treating missing values as unknown parameters. Fully Bayesian paradigms are naturally equipped to handle this situation by augmenting MCMC routines with additional layers and sampling from the full conditional distributions of the missing data, in the case of Gibbs sampling. Here we detail ideas behind the Bayesian treatment of missing data and conduct simulations to illustrate the methodology. We consider specifically Bayesian multivariate regression with missing responses and the missing covariate setting under an ignorability assumption. Applications to real datasets are provided.
Dr. Haresh Rochani, Assistant Professor of Biostatistics and Director of the Karl E. Peace Center for Biostatistics, co-authored the chapter titled “Markov Chain Monte-Carlo Methods for Missing Data Under Ignorability Assumptions” in the ICSA Book Series in Statistics titled Monte-Carlo Simulation-Based Statistical Modeling.
Georgia Southern Explores Improving the Efficiency of the Monte-Carlo Methods for Missing Using Ranked Simulated Approach
This chapter explores the concept of using ranked simulated sampling approach (RSIS) to improve the well-known Monte-Carlo methods, introduced by Samawi (1999), and extended to steady-state ranked simulated sampling (SRSIS) by Al-Saleh and Samawi (2000). Both simulation sampling approaches are then extended to multivariate ranked simulated sampling (MVRSIS) and multivariate steady-state ranked simulated sampling approach (MVSRSIS) by Samawi and Al-Saleh (2007) and Samawi and Vogel (2013). These approaches have been demonstrated as providing unbiased estimators and improving the performance of some of the Monte-Carlo methods of single and multiple integrals approximation. Additionally, the MVSRSIS approach has been shown to improve the performance and efficiency of Gibbs sampling (Samawi et al. 2012). Samawi and colleagues showed that their approach resulted in a large savings in cost and time needed to attain a specified level of accuracy.
Dr. Hani Samawi, Professor of Biostatistics, authored the chapter titled “Improving the Efficiency of the Monte-Carlo Methods for Missing Using Ranked Simulated Approach” in the ICSA Book Series in Statistics titled Monte-Carlo Simulation-Based Statistical Modeling.
Georgia Southern Receives Grant to Study Nanoparticle Levels and Field Evaluation of N-95 Respirator Masks in Construction Sites
The Center for Construction Research and Training awarded a new pilot grant of $30,000 to Georgia Southern University through National Institute for Occupational Safety and Health (NIOSH) cooperative agreement OH009762 (PI: Dr. Atin Adhikari, Department of Epidemiology & Environmental Health Sciences, Jiann-Ping Hsu College of Public Health; Co-PIs: Dr. Aniruddha Mitra and Dr. Abbas Rashidi, Department of Mechanical Engineering). Construction workers experience high rates of respiratory diseases and exposure to ultrafine dust including silica nanoparticles potentially an important etiological factor. Exposure levels of nanoparticles in construction worksites are largely unknown and respiratory protection offered by commonly used respirator masks against nanoparticles in construction jobsites have never been investigated. According to the U.S. Bureau of Labor Statistics, approx. 6.7 million workers are employed in the U.S. construction industry and these construction workers are at an increased risk of respiratory diseases from exposure to respirable dust containing crystalline silica, gypsum, and other contaminants. General belief is that the dust generated during mechanical processes in construction jobsites are mostly large particles formed through crushing, drilling, grinding, or sawing. Consequently, little attention has been given to the generated submicron ultrafine and nanoparticles and their associated exposure levels. These data is, however, critically important because recent laboratory studies have demonstrated cytotoxicity of nanoparticles on lung epithelial cells. Because no data is available on the nanoparticle exposure levels among construction workers, the performance of particulate respirators generally used by construction workers were never evaluated in field conditions against particles of sub-micron and nano-size range. In this project, the multi-disciplinary research team will investigate nanoparticle exposure levels in several construction job sites by a novel newly developed nanoparticle monitoring device and also evaluate the real-time filtration efficiency of respirator masks against nanoparticles by a unique respirator evaluation set-up.
Dr. Marina Eremeeva, Associate Professor of Environmental Health Sciences at the Jiann-Ping Hsu College of Public Health Georgia Southern University, has received a $15,000 grant from the Lyme Disease Association, Inc. to perpetuate and further characterize an important reference collection of Borrelia isolates from the southeastern USA; one primary goal is to make them available to the wider research community investigating the ecology and etiology of Lyme disease. Our collection is a unique sampling of representative Borrelia isolates from ticks, animals and human. These Borrelia spp. Are known to circulate in the southeastern USA and can be used to further the development of accurate and comprehensive information about the spatial and temporal presence and prevalence of B. burgdorferi s.I. and other Borrelia spp. In the Southeast USA, and their contribution to the risk of acquiring Lyme disease in this region areas.
African American maternal caregiver support for prevention of childhood obesity may be a factor in implementing, monitoring, and sustaining children’s positive health behaviors. However, little is known about how perceptions of childhood obesity risk factors and health complications influence caregivers’ support of childhood obesity prevention strategies. The objective of this study was to determine if childhood obesity risk factors and health complications were associated with maternal caregivers’ support for prevention initiatives.
A convenience sample of maternal caregivers (N = 129, ages 22–65 years) completed the childhood obesity perceptions (COP) survey. A linear regression was conducted to determine whether perceptions about childhood obesity risk factors and subsequent health complications influenced caregivers’ support for prevention strategies. Caregivers’ perceptions of childhood obesity risk factors were moderate (M = 3.4; SD = 0.64), as were their perceptions of obesity-related health complications (M = 3.3; SD = 0.75); however, they perceived a high level of support for prevention strategies (M = 4.2; SD = 0.74). In the regression model, only health complications were significantly associated with caregiver support (β = 0.348; p < 0.004).
In conclusion, childhood obesity prevention efforts should emphasize health complications by providing education and strategies that promote self-efficacy and outcome expectations among maternal caregivers.
“Do Maternal Caregiver Perceptions of Childhood Obesity Risk Factors and Obesity Complications Predict Support for Prevention Initiatives Among African Americans?,” was published in the Maternal Child Health Journal.
Dr. Dayna S. Alexander, alumni of the Jiann-Ping Hsu College of Public Health Georgia Southern University (JPHCOPH) was the lead author. Dr. Moya Alfonso, Associate Professor of Community Health Behavior and Education and Dr. Alesha R. Wright, alumni both of the JPHCOPH were co-authors.