There remain many questions to be asked about the reliability of this model. Are the predictors a good subset of all potential predictors of the response? Are the coefficient estimates accurate? Is the relationship between predictors and response, indeed, linear? Are model forecasts dependable? In short, is the model well-specified and does OLS do a good job fitting it to the data?Another method of the LinearModel class, anova, returns additional fit statistics in the form of a tabular array, useful for comparing nested models in a more extended specification analysis: