| Hu Chunyan,Hu Liangping,Reasonably carry out mean value comparison： Poisson distribution regression models[J].SICHUAN MENTAL HEALTH,2023,36(S1):13-17 |
| Reasonably carry out mean value comparison： Poisson distribution regression models |
DOI：10.11886/scjsws20230201003 |
| English keywords:Poisson distribution regression model Offset Standardized mortality ratio Deviation information criterion Highest posterior density interval |
| Fund projects: |
|
| Hits: |
| Download times: |
| English abstract: |
| The purpose of this paper was to introduce 6 basic concepts， calculation methods， a clinical investigation example and its SAS implementation related to the Poisson distribution regression model. The basic concepts included the Poisson distribution， Poisson distribution regression models， offsets， standardized mortality ratio （SMR）， deviation information criteria and the highest posterior density intervals. The calculation method involved the classical estimation method and the Bayesian estimation method of the Poisson distribution regression parameters. The clinical investigation example involved the data on observed and expected cases of lip cancer in 56 Scottish counties from 1975 to 1980. This article presented the whole process of using SAS software to deal with the count data in the example， including constructing five Poisson distribution regression models based on the bglimm procedure and showing the degree of agreement between the predicted SMR and the observed SMR. The output results were explained， and based on the evaluation statistics of the model fitting effect， the multiple constructed Poisson distribution regression models were compared， and finally the optimal Poisson distribution regression model suitable for the data in the paper was obtained. |
| View Full Text
View/Add Comment Download reader |
| Close |