Subscribe Bookmark RSS Feed

JMP - Lack of Fit - Remedial Measures?

Is anyone able to recommend methods to improve lack of fit (when RSquare is high)? Or any references? We used stepwise regression to test for all interactions and polynomials to degree 4. The resulting model has only signficant terms included. I cannot see any trends in the residual by predicted plot diagram although some design points have all residuals (for the 10 replications) above or below 0.