Exploring of Statisical Method in Meta-Analysis Including Parameter Estimation and Hypothesis Testing

According the developing technology and data science, the Evidence Based Medicine (EBM) receives much attention and is very important right now. Meta-analysis method is one step of systematic review, and combines the point and standard error estimators from different articles by weighted average method. The traditional meta-analysis is a two stages testing method which would be get the pseudo association relationship because of the type I error are larger than significant level. But application statistics methods in medicine, a lot of people do not care about the bias results due to statistics methods. In our study, we would propose the one stage combined method in binomial response variable to apply in the 2x2 contingency table. We could consider the heterogeneity and homogeneity between the studies, and simulation the data is analogous to several studies. This method would be including (1) a combine statistics method and a test statistics based on likelihood function, (2) calculated the traditional method in pooling point estimator and standard error, (3) compare the results in (1) and (2). Then we could find out the shortcoming of traditional method and have the mistake interpretations.