cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP is taking Discovery online, April 16 and 18. Register today and join us for interactive sessions featuring popular presentation topics, networking, and discussions with the experts.
Choose Language Hide Translation Bar
Danijel_V
Level III

Method for determining significant effects in experiments

I am trying to analyze around 60 experiments. Each experiment has a lot (around 30 or more) of different continuous and categorical factors (process variables) and for every experiment, various responses (around 10) were measured. Unfortunately, experiments were not made with the design of experiments approach, there is usually a change in just one variable between two experiments (but in the long run, there could be a lot of changes if you compare experiment no. 1 and no. 60).

 

Is there any useful analysis method for determining the most significant effects on each factor and if possible to also determine the effect size. It would really help me a lot since at the moment I am trying to figure out the most important effects with comparing two experiments at a time. Also, recognition of any patterns in the data would be helpful.

Thank you in advance, Daniel

1 ACCEPTED SOLUTION

Accepted Solutions
julian
Community Manager Community Manager

Re: Method for determining significant effects in experiments

Hi @Danijel_V,

You can use the platform in either way, in fact. If you add a large number of potential predictors, and just one response, you will receive the same kind of output. When you include multiple predictors as well as multiple responses, you get all possible combinations.

@julian

View solution in original post

3 REPLIES 3
julian
Community Manager Community Manager

Re: Method for determining significant effects in experiments

Hi @Danijel_V,

Have you looked into the Response Screening platform in JMP (Analyze > Screening > Response Screening)? At the very least it can help guide your process of detecting the active effects among a large number of comparisons (and the false discovery rate correction will be useful to you, I'm sure). I've included a webinar recording below that also demonstrates the functionality, but as you dig into the platform I suggest you read the documentation I linked above. 

@julian

Danijel_V
Level III

Re: Method for determining significant effects in experiments

Thank you for your suggestion, I will certainly look into the response screening more deeply. However, as I see it, response screening is meant mainly for looking on which responses has a certain factor (just 1) effect. I am more interested in the opposite analysis, so which factors have the most influence on a certain response. Is there any similar method than for responses?

julian
Community Manager Community Manager

Re: Method for determining significant effects in experiments

Hi @Danijel_V,

You can use the platform in either way, in fact. If you add a large number of potential predictors, and just one response, you will receive the same kind of output. When you include multiple predictors as well as multiple responses, you get all possible combinations.

@julian