Correlation of composition/treatment

[#1]

Hello all!
I have results from an experiment that I made, with 5 different temperature treatments on a plant and then an analysis of the Essential Oil composition (percentage).
I want to apply a statistical test to show correlation between different temperatures and specific substances from the composition (i.e., which compound is more related to what temperature).
Could anyone advise me on a correct test for that? Also, if possible, I would like it to be displayed in a visual way (some sort of cluster).

Sorry of this seems vague and thanks in advance!

Gilado

[#2]

I have three rules for successful analysis of data:

1. plot the data.

2. Plot The Data.

3. PLOT THE DATA.

Worry about 'tests' after plotting the data. You're also mixing two different analytical approaches. Correlation analysis is not a test per se. Simple correlation analysis of continuous data (I presume your response is continuous) is a analytical technique that evaluates the degree of linear associativity between two variables. So the Multivariate platform would be the best path in JMP to start with if that's what you are after...

But you also used the word, '...test...', which implies to me some sort of hypothesis test of population parameters for the response as a function of temperature. If this is the case, one now should proceed down a different analytical path...and that path depends almost entirely on the specific hypotheses you are evaluating and how you treat temperature and the responses within the context of the experiment and hypotheses. You may end up with the Fit Model platform...or Fit Y by X, or some other platform?

But I return to my first recommendation...plot the data...you may be able to answer your practical question from a simple drag and drop in Graph Builder of Fit Y by X?

[#3]

As always, @P_Bartell  gives excellent advice.  

I would note that you imply on one plant and it also seems you have just one factor, temperature?  This is a very narrow inference space.  I take it there are multiple essential oils?  How "good" is your measurement system?  

Scatter Plots (Multivariate Methods>Multivariate) and other relational graphics that can be displayed using Graph Builder is a good start.  Make sure the variation in the essential oils is of practical value before determining statistical significance.