Can someone explain to me why the equation generated for the coded factors yields residuals that like 3 orders of magnitude above the actual response variable values. What am I doing wrong. The equation using the uncoded factors hit pretty closely to the actual response data.