![IconGems-Print](images/IconGems-Print.png)
449 – SPEED: Topics in Genetics and Biopharmaceutical Applications, Part 2
Model Identification in Linear Fixed Effects Models
Ondrej Blaha
LSU Health Sciences Center
Julia Volaufova
LSU Health Sciences Center
Lynn Roy LaMotte
LSU Health Sciences Center
Our study focuses on numerical investigation of performances of current existing variable selection techniques incorporating statistics like adjusted R^2, AIC, BIC, or SBC for linear models. Specifically, we focus on the ability of these statistics to detect a true model among all possible sub-models. Furthermore, we explore the dependence of the successful true model detection on the parameter setting. Simulation studies were designed to investigate properties of detection of the true model among all possible models. Results provide a new perspective on the current, commonly used techniques. The consequences of the results are discussed as well.