Stats question/how do I use JMP to determine if a process factor is significant (forgot which module-last did it in 2017)

[#1]

University student.  I am analyzing data from experiments not designed using JMP, and would like to find out which parameters have significance in affecting the outcome.

The experiment stacks 2 materials, and then applies an anneal, so those are my three categorical variables.  For one material I have 2 levels, for the other I have three.  There are 3 possible anneals. 

So I guess it's a full-factorial experiment because I have 5 sample results (except one has only 4 due to a sample being dropped) for every possible combination (material A 2 levels, material B 3 levels, Anneal 3 levels = 18 combinations of 5 samples each).

The output figure of merit is a contiguous continuous variable.

There are several uncontrolled parameters where I'd like to determine if they correlate with the output variable. For example, date of process. date of measurement. position within the process chamber. etc.

I recall having a course in which we did a JMP correlation of some type that can reveal which factors are likely significant, and then you re-run it dropping out the ones that are not significant or that are mutually confounding, progressively narrowing the number of possible factors until you are left with only the parameters that actually matter. I don't recall now if it was just an ANOVA; my understanding is that ANOVA only works for linear correlations so I'm not sure if that would be the correct analysis to use on this data. 

Unfortunately the site-license for JMP at my University has expired so I'm left to finish my capstone project analysis with the demo version and no longer have access to the JMP training videos available in my class those years ago to refresh my memory how to do this (they were not free at the time I took the class). I'm trying to graduate this month so I don't think it's appropriate to buy a student license again at this point in time.

Can someone explain, or point me to the right help documents or videos so I can quickly move on with this analysis? 

Is the feature I'm thinking of in the DOE Module? Worried it might be in JMP pro and not available in the demo version.

Including a random jsl so that I can post.  This one-way analysis showed that 2 of the anneals were not statistically different from each other but the third was different from the other two.

Oneway( Y( :"SENSITIVITY*"n ), X( :Anneal ), Means( 1 ), Mean Diamonds( 1 ) )

[#2 · Accepted solution]

Sounds like you need to use:  Analyze>Fit Model

The responses go in the response box,

The inputs go into the Effects box.

Something like this: https://www.youtube.com/watch?v=b6hbOLariNE

[#3]

I have found that some of the training videos used in my class in 2017 are now free, I'm reviewing them now and will come back to view your link here when I'm done, and report on whether that module gets me where I hope to be.

Side note--another community member sent me a message saying they could do the analysis for me and send me the results, for a fee. This kind of really annoyed me on several levels, so I like to take a moment to rant in public, if that's not against the rules.

The point of pursuing the degree is to learn how to do the work, not to subcontract the work and claim it as my own.  The data also involves a corporate partner with the school who probably would not appreciate having their data tossed around the internet prior to any publication or any commercial use they might want to make of it (though doubtful, still a concern). Lastly I'm pretty sure accepting someone else's analysis and presenting it as my own would violate my school's academic code of conduct, and I don't want to add another name to the 'acknowledgements' list on the project report (equivalent of a co-author) for something I can very well (and should very well) learn to do myself.

Please consider context before offering to do someone else's work for them. Even if you were approaching a corporate worker on a deadline, their parent company would have to hire you on as a contractor with all the associated legal mumbo-jumbo and confidentiality agreements before handing over their data, and they'd probably want to shop around to get the best combo of skills and value for their money when choosing such a contractor.

[#4]

I'm sorry to hear you were annoyed by another community member offering their services (for a fee).  I have not had that experience myself.  The challenge for most of "us" community members is how can we be of assistance?  And, often, this is a challenge because we don't know exactly what help (or how much) is needed/wanted.

As you stated, you are a student...perhaps you should steer clear of getting the answers to your questions from anyone else?  There are multiple ways to analyze the data. Not to offend, I would be happy to give you a list of reference books and papers so you can do your own research?

BTW, what do you mean by the response is contiguous?  Or do you mean continuous?

[#5]

Continuous.  It was a typo, sorry.  (or possibly changed by the spellchecker, which changed my W T F below to 'why' and had to be corrected)

There's a difference between asking for guidance on how to use a tool, or under what menu to find its features, and asking someone to do the work for you.

"perhaps you should steer clear of getting the answers to your questions from anyone else?"  W T F , dude? I'm actually a master's student with several years of industrial experience but if you said that to an undergrad you might well intimidate them right off the message board.  Shame on you.

(add: why do you have no criticism of the person who thought that a student asking a question on a free community board represented an opportunity to get paid for JMP expertise?)

[#6]

@lisat , I completely apologize for my comments.  I did not mean to offend and was attempting to communicate it can be challenging "for us" to know how to respond.  Unfortunately, I am not a dude (more of a nerd), but I appreciate the compliment. I was not insinuating anything by my comment.  It can be challenging to decide what and how much advice can/should be provided.  I thank for for correcting me.  

In this case, certainly you can use quantitative analysis as suggested by Byron, or you could choose a more graphical approach.  In any case, I think the 5 (or 4) samples are repeats and should not be considered degrees of freedom.  I would look at those for consistency (within treatment) and then determine how best to summarize those data points (e.g, mean, variance) to assess the factors in your model.  I hope you are successful in the completion of your degree.

[#7]

Thanks for the apology. I hope you understand my point of view.  I agree it is hard to read intent on a message board, with the absence of voice tone etc.

I tend to use 'dude' as a non-gendered, casual word for 'human', so I hope I did not offend in that category.

[#8]

Quite the contrary...The "dude" is legendary amongst my age group (Jeff Bridges in the Big Lebowski). I was taken back by some of your acronyms, but again you are from a different generation.  All good.