تشخيص مشاهدات موثر در معادلات همزمان

Foreword Diagnostic of Influential Observations in Simultaneous Equations

In fitting simultaneous equations models, diagnosing influential observations is so important. There have been so many considerations on the importance of diagnostics within the linear regression single-equation context. A practical approach to influence analysis involves the deletion of cases. The effect or influence of a case is measured by the difference in parameter estimates before and after removal of that case. In this paper, influence diagnostics for simultaneous equations models which have been recently studied, is reviewed. Furthermore, two approaches, one based on imputation method and the other based on the use of EM algorithm are introduced. These diagnostics methods are applied to an example in economics.