American Statistical Association New York City Metropolitan Area Chapter New York State Psychiatric Institute at Columbia University Medical Center Biostatistics Seminar

The Iowa Child Passenger Safety Survey is conducted annually to assess compliance with Iowa’s Child Restraint Law for children travelling in motor vehicles. The survey involves 36 communities distributed across the state. Surveyors complete records by approaching the drivers of vehicles with child passengers in the parking lots of convenience stores, and asking the drivers for their participation. The restraint status of each child passenger is recorded along with other pertinent information. The results of the survey are summarized annually in a report presented to the Iowa state legislature.

This presentation outlines our first attempt to formally analyze the survey data using statistical modeling. Our goal is to investigate, characterize, and interpret the relationships between potential explanatory factors of interest and the binary outcome of proper restraint usage. These factors include the age of the child, the restraint status of the driver (belted or not belted), the population size of the community (reflecting its degree of rurality), and the size of the vehicle cab. We employ the framework of generalized linear mixed models (GLMM). The GLMM formulation includes a random effect to adjust for (1) the correlation among responses collected within a particular community, and (2) the spatial correlation between responses collected in communities that are located in close proximity to one another.

Joseph E. Cavanaugh is currently a Professor in the Department of Biostatistics at The University of Iowa with a secondary appointment in the Department of Statistics and Actuarial Science. He obtained his Ph.D. in Statistics in 1993 from the University of California, Davis. Dr. Cavanaugh’s research interests include biostatistics, model selection, time series analysis, state-space models, linear models, mixed models, modeling diagnostics, discrimination and classification, and computational statistics.