A Case for Interpreting Regression Weights

Abstract

McNeil (1990) argues against interpreting estimated linear model parameters or weights, largely on the basis of the expected sample-to-sample variability in those estimates. In rebuttal it is noted that not to interpret model parameters is to ignore the strength of regression analysis. Appropriate regard may and should be given parameter uncertainty, But that is only part and parcel of parameter interpretations Examples of linear model parameter interpretation are given.