Trying to Simulate Selection Rules Diffiicult

[Zachary Feinstein]

The scenario is I have 1,000 respondents in a survey.  Each person is classified as having a score between 1 and 7 for some brands they rated.  They could have rated up to 15 brands total.

 

My job is to randomly select 4 brands that they will do followup ratings with.  Problems occur because some of the rules overlap and I think I know how to fix it but I do not know the SPSS code.

 

So it starts out by saying brand1 is chosen based on who had the highest score.  Only one can fit into Brand 1.

 

Then the algorithm starts looking at filling brand2 then brand3.  Brand 2 says to pull a certain brand but if that does not occur then use another rule.  That same rule defines the first condition for brand 3.

 

So, for example, the first rule for 2 might say to pull any brand that is a 5 or 6, but if they do not exist then pull a 3 or 4 and submit that instead to brand 2.

 

But brand 3 is to first try to pull 3 or 4 but the problem is that I want to remove any responses (not cases) selected from brand 2 from the pool before I get to brand 3.

 

To visualize this I am also submitting one respondent from a stacked dataset below.  They are no. 1291.  I broke brand2 into 2 different fields.  One is called brand2_1 and the next one is called brand2_2.  No-one qualified to rate the brands based on the first Brand 2 definition but many did based on the 2nd one.  What I need to learn to program is if I pull the first response off of brand2_2 then I want to remove that same first response from brand3_1.  I am using AGGREGATE with the FIRST() functionality to pull the first instance I find for each of the listed fields.  So ideally the first for each.  If I can get the program to work it would be 3-1-5-2.  The 5 is pulled because I somehow got rid of the first 1 there because it was selected for Brand2.

 

Any and all help is highly valued and appreciated.

 

Zachary

[hidden email]

[phone]

 

RESPID	brand	brand1	brand2_1	brand2_2	brand3_1	brand3_2	brand4
1291	.	3	.	.	.	.	.
1291	1	.	.	1	1	.	.
1291	5	.	.	5	5	.	.
1291	7	.	.	7	7	.	.
1291	10	.	.	10	10	.	.
1291	12	.	.	12	12	.	.
1291	13	.	.	13	13	.	.
1291	14	.	.	14	14	.	.
1291	6	.	.	.	.	6	.
1291	8	.	.	.	.	8	.
1291	2	.	.	.	.	.	2
1291	4	.	.	.	.	.	4
1291	9	.	.	.	.	9	.
1291	11	.	.	.	.	.	.
1291	15	.	.	.	.	.	.

[ViAnn Beadle]

This sounds a bit like TURF analysis? If so, I think that there is a python-based module to do it on the SPSS DevCentral site.

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Zachary Feinstein
Sent: Wednesday, June 17, 2009 1:21 PM
To: [hidden email]
Subject: Trying to Simulate Selection Rules Diffiicult

 

The scenario is I have 1,000 respondents in a survey.  Each person is classified as having a score between 1 and 7 for some brands they rated.  They could have rated up to 15 brands total.

 

My job is to randomly select 4 brands that they will do followup ratings with.  Problems occur because some of the rules overlap and I think I know how to fix it but I do not know the SPSS code.

 

So it starts out by saying brand1 is chosen based on who had the highest score.  Only one can fit into Brand 1.

 

Then the algorithm starts looking at filling brand2 then brand3.  Brand 2 says to pull a certain brand but if that does not occur then use another rule.  That same rule defines the first condition for brand 3.

 

So, for example, the first rule for 2 might say to pull any brand that is a 5 or 6, but if they do not exist then pull a 3 or 4 and submit that instead to brand 2.

 

But brand 3 is to first try to pull 3 or 4 but the problem is that I want to remove any responses (not cases) selected from brand 2 from the pool before I get to brand 3.

 

To visualize this I am also submitting one respondent from a stacked dataset below.  They are no. 1291.  I broke brand2 into 2 different fields.  One is called brand2_1 and the next one is called brand2_2.  No-one qualified to rate the brands based on the first Brand 2 definition but many did based on the 2nd one.  What I need to learn to program is if I pull the first response off of brand2_2 then I want to remove that same first response from brand3_1.  I am using AGGREGATE with the FIRST() functionality to pull the first instance I find for each of the listed fields.  So ideally the first for each.  If I can get the program to work it would be 3-1-5-2.  The 5 is pulled because I somehow got rid of the first 1 there because it was selected for Brand2.

 

Any and all help is highly valued and appreciated.

 

Zachary

[hidden email]

[phone]

 

RESPID	brand	brand1	brand2_1	brand2_2	brand3_1	brand3_2	brand4
1291	.	3	.	.	.	.	.
1291	1	.	.	1	1	.	.
1291	5	.	.	5	5	.	.
1291	7	.	.	7	7	.	.
1291	10	.	.	10	10	.	.
1291	12	.	.	12	12	.	.
1291	13	.	.	13	13	.	.
1291	14	.	.	14	14	.	.
1291	6	.	.	.	.	6	.
1291	8	.	.	.	.	8	.
1291	2	.	.	.	.	.	2
1291	4	.	.	.	.	.	4
1291	9	.	.	.	.	9	.
1291	11	.	.	.	.	.	.
1291	15	.	.	.	.	.	.