Question about factor coding conventions in Fit Model Platform

[#1]

Hello Community,

I'm running Fit Model to analyze contributions of both nominal and continuous factors in a system.  I'm getting different parameter estimates (and significance) between JMP and Python.  I understand that his might be related to how the nominal factors are encoded (Effects coding (default in JMP) vs. Dummy coding (default in Python).  I've been trying to find a path for selecting how I can switch between the two conventions but am coming up blank.  Some online guidance mentions the option either in the Column Info or in the Model Effects box when I launch the Fit Model platform: I can't find it anywhere.  The results from Python and JMP are similar but different.  I'm trying to understand why and how I can 1) choose the best convention to adopt and 2) be able to execute it in JMP.  The dataset here is a reduced example of what I'm trying to analyze...trying to understand impacts of different parameters on Efficiency with an emphasis on understanding impacts of Fill Volume (two levels) and concentration (range).  

  I'm using JMP19.

Thanks in advance !

Chris

[#2]

Hi @ClusterFerret68,

I think you may find some explanations about nominal factors encoding in the following discussions:

Differences in parameter estimates using same multiple regression analysis. 
How DOE analysis handle categorical factor in regression analysis 
Interpretation of Dummy Variables in Stepwise Regression wtih {0-1} Next to Variable Name 
In JMP, how do you adjust the parameterization of categorical variables in a regression model? 

Hope you'll find the answers in these discussions,

[#3]

You did not include a script or description of the model you are trying to fit.  

If Python is using "dummy" variable coding, it is likely that the "Indicator Parameterization" estimates in the Fit Lease Squares report is what you are looking for to compare with.  see https://www.jmp.com/support/help/en/19.0/index.shtml#page/jmp/indicator-parameterization-estimates.s...

[#4]

First of all, be sure to read the Statistical Details section of the online documentation for an explanation of the parameterization of nominal and continuous predictors.

There are different parameterizations. They are all legitimate. They represent different ways of thinking about effects  and testing parameter estimates. JMP uses effects coding. As such, the parameters represent the difference between the least squares mean for a given level and the overall mean. So an estimate of -3.5 means that the response is 3.5 units lower with this level than the mean response.