Thomas Lumley • 1/13/2016

Another view of the ‘nearly true’ model

This technical article examines statistical model comparison, analyzing the efficiency trade-offs between a large model estimator and a more efficient small model estimator. It discusses the implications of slight model misspecification, the performance of a difference-based test (D), and the asymptotic behavior of estimators under correct and misspecified models, referencing concepts from semiparametric efficiency and hypothesis testing.

0 comments

#Statistical Inference #Asymptotic Theory #Regression Analysis