Study objectives. The objectives of this study were twofold. First, we used simulation to investigate three statistical methodologies for analyzing recurrent event data in the presence of environmental covariates to determine which procedure performs well under what situation. Secondly, we investigated the association between daily air pollution and hospital admissions of respiratory diseases by analyzing data from Vancouver, British Columbia. Settings and study population. Five air pollutants: carbon monoxide (CO), coefficient of haze (CoH), nitrogen dioxide (NO 2), sulfur dioxide (SO2) and particulate matter 10 microns or less in diameter (PM10) were considered in this study. The event of interest was daily respiratory hospital admissions of residents (65+ years) of Vancouver, B.C. from April 01, 1995 to March 31, 1999. The dates of hospital admission for each individual's repeated visits due to respiratory diseases were recorded. Statistical methodologies. Three statistical methods were studied in this study, namely, Dewanji and Moolgavker's model (2000, 2002) based on a Poisson process assumption (Model I), Nividi's model (1998, 2002) for bidirectional case-crossover designs (Model II), and the usual time series analysis using a generalized linear model with natural splines (ns) to smooth time (Model III). A simulation method was used to evaluate and compare these procedures. The mean square error (MSE) was the criterion used for evaluation. (Abstract shortened by UMI.)Dept. of Mathematics and Statistics. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2005 .K45. Source: Masters Abstracts International, Volume: 44-01, page: 0379. Thesis (M.Sc.)--University of Windsor (Canada), 2005.