A geometric approach to assess bias due to omitted covariates in generalized linear models
Misspecifying generalized linear models by omitting covariates associated with the response can result in seriously biased estimates of the effects of the included covariates. F...