I have used Design expert software to design a set of experiments, and I want to model concentration against some parameters. In the procedure of my experiment I should ultrasonic the solution, but this causes the evaporation of some solution (water). Besides, after the ultrasonication a part of my sample sediment. Now I have new concentration in my solution; of course I can determine this new concentration, but I don’t know what to do with these concentrations that are difference from the ones from my DOE. I would be grateful If anyone suggest me a solution.
I've had a similar issue before.
Do you know how much the concentrations changes? If it changes by the same amount each time, there probably is not an issue.
I think it is better to enter the actual concentrations than the theoretical. Usually, the orthogonality is still acceptable, and it can be handle. Look at the condition number. If the condition number is not too high, the design can still be evaluated. Otherwise, any projection method like PLS instead of MLR can be used to handle the colinearities.
Thanks for your answer dear Andrew; no, they are not the same. Can you explain me how did you solve your problem...
Thank you Mattias for your suggestion, I will investigate the concentration differences and will replace the actual concentration in the case of little difference,
I would follow what Mattias said. Enter in the new concentrations and let Design Expert work it's magic. DX software will alert you if there are any collinearity issues.
Stat Ease has several consultants that can help you if you are not sure about the analysis. Their services are free too. They usually get back to me within a few hours.
Right, you can introduce the real concentrations into the experiment matrix and DX will analyse the results taking in account these changes.