I've been reading up on endogeneity and I see many articles where authors use 2SLS or instrumental variables without checking whether or not endogeneity exists.
My data set is a balanced panel data set and I'm using a change HLM model: DeltaY_ij=〖β0〗_j+〖β1〗_j X_ij+Ɛ_ij.
Can't I just save the residuals from my model and run a correlation test each of my predictors? If they are correlated then endogeneity exists and then I would use an instrument, if not then endogeneity is not an issue.