If I issue a SPSSINC CREATE DUMMIES statement immediately after calling OPEN DATASET, all expected dummy variables are created, but for the first k dummies every case gets a 0 value. k varies from 1 to about 5, rarely 10 or so. Typical number of dummies created per SPSSINC CREATE DUMMIES call is between 12 and 50 in my current use case.

If I immediately call SPSSINC CREATE DUMMIES another time (with identical or different arguments) on the same dataset, the created dummies have the expected values for all cases.
The same is true for all subsequent calls on the same dataset.

I am currently helping myself by duplicating the first call. However, I would be more reassured if I did not have to do this. While I do carefully validate the dummy values, a degree of uncertainty remains as to whether all the values are correct.

Does this sound familiar in any way? If not, I can supply a data file with corresponding syntax.

(SPSS 28.0.1.1. (14), macOS 11.6.4.)

I have never seen this problem.  Please send me data and syntax to try to reproduce (jkpeck@gmail.com).

------------------------------
Jon Peck
------------------------------

Original Message:
Sent: Fri February 24, 2023 09:41 AM
From: Frank Watzl
Subject: SPSSINC CREATE DUMMIES: First call per dataset yields unreliable dummy values

If I issue a SPSSINC CREATE DUMMIES statement immediately after calling OPEN DATASET, all expected dummy variables are created, but for the first k dummies every case gets a 0 value. k varies from 1 to about 5, rarely 10 or so. Typical number of dummies created per SPSSINC CREATE DUMMIES call is between 12 and 50 in my current use case.

If I immediately call SPSSINC CREATE DUMMIES another time (with identical or different arguments) on the same dataset, the created dummies have the expected values for all cases.
The same is true for all subsequent calls on the same dataset.

I am currently helping myself by duplicating the first call. However, I would be more reassured if I did not have to do this. While I do carefully validate the dummy values, a degree of uncertainty remains as to whether all the values are correct.

Does this sound familiar in any way? If not, I can supply a data file with corresponding syntax.

(SPSS 28.0.1.1. (14), macOS 11.6.4.)