Fitted Response - The fitted response is the response predicted by a model with a specific set of main factors, interactions and responses. The fitted response may be based on raw or transformed factor-levels. It may use raw or calculated responses, including subgroup means or variances and combinations of responses.
In order to calculate confidence limits around the mean response, a measure of error is required. In some designs, where the effect of higher-order terms (above quadratic main factors and above three factor interactions) is assumed to be insignificantly small, the degrees of freedom made available may be used for estimating the error. In other designs, the higher order terms are used to represent other factors or interactions or are convoluted with other terms. In that case, additional runs are needed, usually as repeats or duplicates of other runs, to obtain an error estimate.
Learn more about the Regression tools in Six Sigma Demystified (2011, McGraw-Hill) by Paul Keller, in his online Regression short course (only $99), or his online Black Belt certification training course ($875).