Can you provide more information?
What software do you use?
Perhaps it is a result of including in the model not statistically significant independent variables? As I understand properly, you included C and C2 while their p-values are 0.065 and 0.649 respectively? Similarly, you included 2-way interactions while p-values of all of them are way beyond 0.05.
thank you for reply , i have use minitab 17 software. as i remove terms then model shows 35% error. data is of MMR for Box-Behnken design with 6 center points.
According to MiniTab support, asterisk at lack-of-fit indicates that software could not calculate values due to the lack of degrees of freedom, which is not a case in your ANOVA. Another reason might be an oversaturated model, so insignificant terms.
While you said, after the deletion of the insignificant terms, the error skyrockets to 35%, perhaps there is an issue with the measurments or the design itself.
If you wish, write a private message here in ResearchGate or an email (find in my profile), so perhaps I can help you and we will menage to solve the problem :)
- Badges
- Science topic