topic Can i include a standard deviation of replicates into DOE Analyses ? in Discussions

Can i include a standard deviation of replicates into DOE Analyses ?

Viral_Jmp — Wed, 19 Feb 2025 18:26:08 GMT

Dear Community,

I have made a custom DEO with multiple factors and I am measuring 3 responses.

Now I have fitted the mean of each response and everything is fine, but is it possible to include my raw values for each response?

Or does it even make sense to include the deviation of each method?

Thanks for the answer.

Re: Can i include a standart deviation of replicates into DOE Analyses ?

statman — Tue, 18 Feb 2025 15:48:44 GMT

First, welcome to the community. I'm a little confused by your query. It is really helpful if you can attach the data table (anonymize it if necessary). Do you have replicates or repeats of each treatment? If you have replicates (e.g., independent runs of the same treatment), you have increased the DF of the experiment and therefore the model. These replicates may or may not be able to be assigned depending on how they were acquired. (e.g., If they are randomized replicates, then they are used to estimate the MSE). If you have repeats for each treatment (e.g., That is one experimental unit measured multiple times), then you can analyze (model) both the mean of the repeated measures and the variation of the repeated measures.

Re: Can i include a standart deviation of replicates into DOE Analyses ?

Victor_G — Tue, 18 Feb 2025 15:58:15 GMT

Hi @Viral_Jmp,

Welcome in the Community !

What do you mean by "I have fitted the mean of each response" ? Are you measuring each response several time and modeling the mean ?

If yes, you can use the mean and standard deviations of your raw data measurements responses and model them separately. You will then be able to understand the impact of each factors on the mean response(s) and on their variances.

A more straightforward option could be to use Loglinear Variance Models.

You can also check previous posts discussions about the same idea :

Hope this first answer may help you,