topic Re: Linear Contrast in Fit Model in Discussions

Linear Contrast in Fit Model

crofut — Thu, 24 Aug 2017 19:33:59 GMT

I've noticed that when using Fit Model (even for a simple one-way ANOVA), that the Linear Contrast is not correctly saved when I save the script to the data file to re-execute at a later time. Also, since I'm creating (attempting) several contrasts, is there an easy way to label/title each constrast?

Re: Linear Contrast in Fit Model

Dan_Obermiller — Thu, 24 Aug 2017 19:49:42 GMT

I'm not seeing a problem. Which version of JMP are you using? Can you show the script that is saved along with the output that the script SHOULD be producing?

Re: Linear Contrast in Fit Model

crofut — Thu, 24 Aug 2017 19:58:44 GMT

Hi Dan, I'm using JMP PRO 13.0 (64-bit). I can't email JMP files (security blockage at company). SO I've attached a PPT file that shows a simple data table, with the script. It also shows the original output, and the output when the script is re-executed.

Thanks,

Steve

Re: Linear Contrast in Fit Model

Dan_Obermiller — Thu, 24 Aug 2017 20:13:06 GMT

Hi Steve!

This is a bug in version 13 of JMP. Updating to 13.2 will fix the problem.

From the 13.2 release notes on fixes made:

Standard Least Squares • When you create an LSMeans Contrast and then save the script for the analysis, the LSMeans Contrast is saved correctly in the script. • The Contrast Specification boxes are aligned correctly

As for the labeling of the contrasts, I don't know of any way to easily do that.

Re: Linear Contrast in Fit Model

jules — Mon, 28 Aug 2017 03:10:29 GMT

Hi @crofut,

It looks like Dan addressed your main issue, but with regard to labeling, when I need to keep track of several contrasts I find it's helpful to simply rename the outline boxes (i.e. double click on the outline box title). This renaming is preserved in the script you save for the report.

I hope this helps!

Julian

Re: Linear Contrast in Fit Model

ghartel — Tue, 11 Sep 2018 04:27:32 GMT

I'm on version 14.1 and it seems there is a similar problem. I'm using Generalized Linear Model to fit a logistic and I'm using contrast to compare levels of a categorical variable. It works fine point and click but the script only saves "Run(Contrast( 1 ))" which requires the contrasts to be specified manually again. I don't see anything in the documentation about how to specify the contrasts via script. Is this just not implemented?