Hi. Let me make sure:
You are creating your own dummy-coded variables and entering them into DISCRIM. Something like this:
COMPUTE race1=0.
COMPUTE race2=0.
COMPUTE race3=0.
IF race=1 race1=1.
IF race=2 race2=1.
IF race=3 race3=1.
** VARIABLE LEVEL race1 to race3 (SCALE).
** VARIABLE LEVEL race1 to race3 (NOMINAL).
DISCRIMINANT /GROUPS=group(1 3) /VARIABLES=race1 to race3
/ANALYSIS ALL /PRIORS EQUAL /CLASSIFY=NONMISSING POOLED.
To me, it makes sense to set the variable level as NOMINAL for each, but it shouldn't matter to the results.
------------------------------
Rick Marcantonio
Quality Assurance
IBM
------------------------------
Original Message:
Sent: Mon October 31, 2022 11:20 AM
From: Clifton Hamilton
Subject: Coding Dummy Variables
I plan on conducting a discriminant function analysis using race as one predictor variable but dummy-coded. Do I identify the dummy codes as SCALE or NOMINAL data in SPSS?
#SPSSStatistics