摘要
當分析零過多且具過離散的個數資料時，許多文獻建議可使用零過多廣義卜瓦松 (ZIGP) 模型或零過多負二項 (ZINB) 模型。 本文研究指出ZIGP 和ZINB兩種模型的參數估計量均不具一致性，故建議以Royall和Tsou (2003) 強韌概似函數方法建立的強韌常態模型配適零過多且具過離散的個數資料。
並列摘要
Zero-inflated generalized Poisson distribution and zero inflated negative binomial distribution are models proposed for analyzing over-dispersed count data with excess zeros. We illustrate that inferences derived from these models are sensitive to model misspecification. Alternatively, we show that one can fix the normal model to accommodate data with the features of interest. The adjusted normal likelihood is asymptotically legitimate so long as the first two moments are correctly specified and that the 3rd and the 4th moments of the true distributions exist.
參考文獻
1. Cameron, A.C. and Trivedi, P.K. (1986) Econometric models based on
count data: comparisons and applications of some estimators and tests. Journal
of Applied Econometrics ,1, 29–53.
2. Consul, P.C. (1989) Generalized Poisson distributions: properties and
3. Consul, P.C. and Famoye, F. (1992) Generalized Poisson regression model.