ABSTRACT Zero-inflated Poisson models are frequently used to analyse count data with excess zeroes. However, results generated by different algorithms, by various statistical packages or procedures in R and SAS,… Click to show full abstract
ABSTRACT Zero-inflated Poisson models are frequently used to analyse count data with excess zeroes. However, results generated by different algorithms, by various statistical packages or procedures in R and SAS, are often inconsistent, especially for small sample sizes or when the proportion of zero inflation is not large. In this study, we compared the underlying nonlinear optimization approaches and the statistical theories on which common packages and procedures are based. Then, multiple sets of simulated data of small, medium, and large sample sizes were fitted to test the performance of algorithms in available R packages and SAS procedures. They were also compared by using a real-data example. The zeroinfl function with methods CD type 1, CD type 2, and CD type 3 in the PSCL package in R and the GENMOD procedure in SAS generally outperformed in the simulation studies and produced consistent results for the real-data example.
               
Click one of the above tabs to view related content.