in the second (optimisation) step of application of design of experiment, we tried to set a model to define most influential factors"we call too surface response methodology" , in parallel, this model give us the possibility to find the optimum of ou r objectif, but lastly in a document I found that to define this surface response we should use mean response
my question is why we shouldn't use also optimal responses and not only mean response ?
Dear Yassine,
To be able to help you, you should give more details about your experiments and results. For example, I'm confused with what you mean by "optimal responses" in " why we shouldn't use also optimal responses and not only mean response ? " Anyway, all results of all experiments in a design should be used in modeling (including rsm models), excluding only clear outliers.
Dear Veli-Matti,
thank you for answer
I am assaying four substances in ULTRA Performance Liquide Chromatography, well I used to optimize two answer which are resolution between the picks and the retention time, to make the opitmisation i work on two type of design of experiment namely: Plackett-Burman design for screening and Box-behnken design to the RSM, then I set up the model using the standard least squares method. thus, to evaluat the robustness of the method, I used to defin the design space as recomended by ICH Q8, which is capable to defin the area where our method should give a robust findings
in review articl, " Rozet, E., et al., Design Spaces for analytical methods. TrAC Trends in Analytical Chemistry, 2013. 42(Supplement C): p. 157-167. ", I found that author mention that most mean response are usefull to build RSM or what i know that like you say that all experiment must be used in design, that why i ask for this question
if only we use mean answer to set the RSM, why not optimal answer "i mean maximal " could be also used
hope that I answer all of your questions
- Badges
- Science method