Design of Experimental is one of the efficient technique to optimize analytical method development. Does anyone could be sharing with me, which one better either full factorial design or fraction factorial design in case we didnt have many previous information on particular compounds.
Thank you
I would suggest for example 2^(5-1) which is a fractional factorial design with high resolution.
The choice of specific design is matter of your objective and the complexity of the situation that you try to understand as Derek says is important that you explain more about the current situation and your particular objective in order to give you a better advice. For example when you talk abou compound do it's mean that you have a mixture?
Hi Derek and Manuel Jose
Yes.. i was given task to study multi element of organometallic compound (mercury and tin, in total 7 ) by extraction using Solid Micro Extraction. I m not worry about cost of analysis but want to know if we increase number of factor, sure the complexity of interaction will increase. Variables of study i will separate between optimization of extraction and optimization instrument condition. yes quantitative study
I have found that an optimal factorial design is usually best for screening and an optimal response surface is good for optimization of a system.
A reason to use a fractional factorial design over a full factorial design, when possible is about the reduction of samples without reducing the useful information. If you look at Mattias's suggestion, the fraction factorial design is half the size of the full design. You lose some ability to see 3-way and higher interactions. However, these types of interactions don't occur very often. So you can't see something that's probably not there. The suggestion is to use a fractional factorial design because you see the most likely interactions.
If you use a full factorial design, you end up using more samples which just builds the error term in the model. This allows you to see smaller effect sizes. If that is what you want to do, then I would suggest using a fractional factorial design and replicating it.
The complexity of the interactions is not associated to the experimental design but to the reality of the situation itsefl. The clear advantage of full factorial design is just that you can learn more about the actual situation itself. I agree with the excellent answer from Andrew, and I also suggest full factorial design. The multiple interactions are not common but because your situation is on organic chemistry context who is very complex in nature phenomena I won't be surprised if that happens. I strongly recommend you to think in advance if your phenomenoma is linear or non-linear respect with each variable because as your phenomena is quantitative it shold be studied as nonlinear then implying using mixed level design (3 in factors that you expect to find curvature and 2 in factors that you think in advance that are linear). In any case when you are working in the optimization stage is better to use response surface as your right tool
Obviously, full factorial design can provide an opportunity to test for higher order interaction effects and will include all possible treatment combinations. On the other hand, fractional factorial design is a subset of full design and thus lose information about the response behavior. However, fractional designs are practically plausible and statistically efficient.
- Badges
- Science topic