Unused levels appearing in Odds Ratio output after recoding a categorical variable in JMP

[#1]

After recoding a categorical variable (either with 'recode' or If) in JMP 18 pro, the old category “Either” still appears in the results — especially in the Odds Ratio output of a logistic regression — even though no rows in the new column contain that value anymore.

It seems JMP continues to recognize “Either” as an existing level of the variable, and there is no visible option to actually delete or remove it from the column properties.

How can I completely remove or drop this unused category so that it no longer appears in the analysis output?

formula ex. 

If( :Preferred drainage == "1",

"0",

If( :Preferred drainage == "2",

"1",

""

)

)

[#2]

Hello,

I am not 100% sure of what you are asking, but your nested If statement seems odd (I speculate that you may not need the second If statement). 

Is the formula below what you are trying to achieve?

If (:Preferred drainage == "1", "0", :Preferred drainage == "2", "1", "");

Assuming that all the values in the Preferred Drainage column are 1 or 2, you should not have empty (i.e., "") results from your formula.

Let us know.

Best

TS

[#3]

Hi TS — Thanks for your reply.

quick clarification:

The original column is character with Value Labels:

1="Stent", 2="Nephrostomy", 3="Either".

I created a new binary column with the nested formula and using 'recode'

• Rows with the original value 3 become "" and I exclude them from the model.
• I run logistic regression on the new column only (event = "1").
  Yet, in the Odds Ratio report I still see comparisons that reference "Either" as if a third level exists.

Could Value Labels from the original column be carried over somehow to the new column (or into the platform), causing “Either” to appear in the OR report?
What’s the correct way to ensure the OR output only reflects the two levels and does not list “Either”?