topic Re: One DOE for two contradictory research questions? in Discussions

One DOE for two contradictory research questions?

Thommy7571 — Tue, 26 Mar 2024 12:59:17 GMT

Hello,

I want to study a chemical reaction, which occurs in the frame of a measurement as a side reaction. I want to know, 1) which factors and which values give an optimal result and 2) which values give a minimal result.

It seems that I have two possibilities:

Two DoE Plans for each question

One DoE Plan for both.

However, to get both results, all kinds of parameters need to be studied thoroughly and

this gives a lot of experiments in view of the eight parameters of the measurement.

Also, in the latter case scanning experiments appear to be difficult, since two reactions need to be targeted and

not only one. What would you recommend?

Sincerely

Thommy7571

Re: One DOE for two contradictory research questions?

statman — Tue, 26 Mar 2024 14:29:34 GMT

Not enough information about the situation to provide any specific advice. Question: is the optimal result of the reaction independent from the minimum result of the reaction? Is there one response variable that measures the "quality" of the reaction? Is it a continuous variable? Are you trying to pick a winner or are you interested in understanding causal structure?

Most situations require sequential investigation to understand. Start with a large design space (e.g., lots of factors at bold levels) and iterate. How many experiments you will need depends on your current subject matter knowledge.

Re: One DOE for two contradictory research questions?

Thommy7571 — Sat, 30 Mar 2024 08:14:55 GMT

Hallo,

sorry for the delay, It belongs to the work and I am on hoildays:

Question: is the optimal result of the reaction independent from the minimum result of the reaction?

Of course there i sa chemical rate law which connects both results but as long as it cannot be determined by th JMP I guess I need to answer that such results are independent, even if a lhigh value for the maximum automatically means a lower value for the minimum value. However it is not clear if there are no parameters which give a Gaussian distribution of the result.

Is there one response variable that measures the "quality" of the reaction?

Yes there is one parameter, which will be simply the ratio of the two signal intensities of the two compounds to maximize:

either compund I is maximized - and compound 2 minimized - or the other way round.

Is it a continuous variable? Are you trying to pick a winner or are you interested in understanding causal structure?

Variables are continuous. I don't know much about the possibilities of JMP. If I have the winner is fine. If there is a way to get chemical parameters - kinetics - from JMP that wood be even better. I would automatically have both results.

The problem is that I have 9 parameters but I only know from two more or less which influence they should have. I have already some measurements but without any special design behind. I neither know if I can use them to prepare my DOE experiment nor how (if yes).