Answer: In general, R-Square Adjusted is a bit lower than R-square, and the smaller the gap between the two, the better the model. Also, under Summary of Fit, look at Root Mean Square Error. If it goes down as you remove terms, the resulting model is likely a bit better. Also look at Analysis of Variance Whole Model Fit (Probability>F) and if you have multiple models, look at Lack of Fit to get and idea of how well the model is explaining the variation in your data.

See Summary of Fit and Analysis of Variance Report.

This video segment is also included in Bill Worley's

billw@jmp **Specifying Fitting Models, Modeling Continuous Data** webcast video.