Thomas Lumley • 6/19/2025

Included-variable bias

The article discusses regression models and the bias that arises when comparing coefficients from models with different sets of variables. It argues that the common term 'omitted-variable bias' implies an incorrect directionality, and introduces the symmetric concept of 'included-variable bias'. The piece critiques the gold-standard approach to model selection and emphasizes defining the target of estimation.

0 comments

#data analysis #statistics #Bia