Extract confidence interval values from a graph to data table - help needed please

[#1]

Hi All

I'm looking to extract the CI data that is computed from JMP17 and shown in the graph. When fitting y by x model and getting CI from there, it seems like the values do not match what is shown in the graph (but I appreciate I'm eyeballing the graph data hence may be mistaken here..).

Could someone provide an advice please? I don't seem to find applicable answer here. 

Cheers

Agnieszka

Note: I'm a newbie, no scripting experience. 

[#2]

I'm not sure, but I believe your graph builder interval is treating your X variable as nominal while your Fit Y by X is built using X as continuous.  If I am correct, then the slight differences are due to the continuous treatment of X as using more information than the nominal treatment (the fit line uses all of the data while the nominal treatment uses only each discrete X value to estimate the means and confidence intervals).  I'll defer to a better statistician to confirm or deny my speculation.

[#3]

Thank you for the insight.

[#4]

Hi @AgnieszkaL : To extend @dlehman1  response, the graph builder is not using the same model as the Fit Y  by X model; i.e., it is treating each set of data (the X's in your Graph Builder plot) independently, with each interval based on the respective sample means and sample SD's from each X so there is no overall estimate of error (based on the total sample) like you'd find in your fit Y by X model; so, you are comparing apples and oranges.     

[#5]

 @MRB3855 Thanks! So, I need to CI to be from the individual set of values corresponding to each level of relative potency. Is there a way to get the CI from the graph into a data table? Or alternatively, calculate the CI from these data sets?

[#6]

Hi @AgnieszkaL : I'm not sure how to get those intervals from the graph into the table (probably doable via JSL, but I'm no JSL expert); but, why do you not want to use the model driven intervals given by Fit Y by X Model?

[#7]

Think I finally got it! the CI for each data set one can get from the distribution platform..

Why I do not want to use the model driven intervals given by Fit Y by X Model - is to align with historical data that were computed this way using Excel and following regulatory guidelines. 

Thanks a lot for getting involved