Regression analysis of count data
2002
Wulu, J.T. | Singh, K.P. | Famoye, F. | Thomas, T.N. | McGwin, G. (Department of Health and Human Services Bureau of Primary Health care, Bethesda (USA))
Three nonlinear count models, Poisson Regression (PR),Negetive Binomial Regression (NBR), and Generalized Poisson Regression (GPR) areused for assessing the effects of risk factors on agricultural injuries from farm injury data. A sample of 1,322 respondents who participated in the farm safety/injury baseline survey in nine rural countries in Alabama and Mississippi aged,18 years and older are considered for analysis. The dispersion parameters in theNBR models. Estimated disperssion parameters in the NBR and GPR models were positive and significantly different fron zero. Estimated goodness-of-fit measures showed that GPR models outperformed the NBR and PR models.
显示更多 [+] 显示较少 [-]