188 – Estimation,Testing, and Diagnostics
A Diagnostic of Influential Cases in Generalized Linear Mixed Models
Sponsor: Quality and Productivity Section
Keywords: Modeling diagnostic, generalized linear mixed model, informational complexity criterion, logistic regression, case-deleted data set, random effects
Junfeng Shang
Bowling Green State University
In the generalized linear mixed modeling setting, we develop a diagnostic for detecting influential cases based on the informational complexity criterion. The computational formula of the informational complexity criterion is evaluated using the Fisher information matrix. The diagnostic compares the informational complexity criteria between the full data set and a case-deleted data set. To demonstrate its effectiveness, the proposed diagnostic is applied in a data set where the mortality of cancer cells under radiation is modeled by the logistic regression with random effects.
