Abstract: The focus in this study embraces the differences and similarity between deviances and the sum of squares related with linear regression, additive and partial linear additive models. In addition to these arguments, we have indicated that both non-adjusted and adjusted coefficients of determination derived using the Linear Regression Models (LRMs) can be easily generalized to additive and Partial Linear Additive Models (PLAMs). This generalization plays a similar role in evaluation of the goodness of fit of the models. To see the details in model fitting, the performances of the examining models are compared by estimating two real data sets. The results obtained from experimental evaluations show that Additive Model (AM) gives a better fit in sense of having a smaller performance measures.