Message exposure

[mils]

Hi everyone,

I would like to create the table/file as per shown below. There are 15 screeners, 4 blocks and 31 messages.
In the example below, its shown one block, 15 screeners, each one with a different set of 4 random messages. The idea behind the blocks is to try to get different sets of screeners. Due to the number of messages and screener some (messages) had to be repeated within the blocks.

Screener Block Message
1 1 16
1 1 20
1 1 25
1 1 30
2 1 8
2 1 13
2 1 16
2 1 18
3 1 1
3 1 6
3 1 12
3 1 18
4 1 3
4 1 5
4 1 21
4 1 26
5 1 13
5 1 17
5 1 28
5 1 30
6 1 14
6 1 19
6 1 24
6 1 27
7 1 8
7 1 9
7 1 11
7 1 17
8 1 2
8 1 26
8 1 29
8 1 31
9 1 1
9 1 8
9 1 22
9 1 28
10 1 6
10 1 10
10 1 20
10 1 22
11 1 11
11 1 23
11 1 30
11 1 31
12 1 4
12 1 5
12 1 18
12 1 24
13 1 2
13 1 7
13 1 15
13 1 28
14 1 5
14 1 9
14 1 15
14 1 23
15 1 1
15 1 15
15 1 29
15 1 30


My goal is to be able to replicate the above table with and SPSS syntax and then used it in another different example with different number of messages, screeners and blocks.

I was thinking of random numbers...

Thanks in advance

[David Marso]

Give it a try and if you encounter problems someone will likely have suggestions.
See INPUT PROGRAM.
See MATRIX.
--

mils wrote

Hi everyone,

I would like to create the table/file as per shown below. There are 15 screeners, 4 blocks and 31 messages.
In the example below, its shown one block, 15 screeners, each one with a different set of 4 random messages. The idea behind the blocks is to try to get different sets of screeners. Due to the number of messages and screener some (messages) had to be repeated within the blocks.

Screener Block Message
1 1 16
1 1 20
1 1 25
1 1 30
2 1 8
2 1 13
2 1 16
2 1 18
3 1 1
3 1 6
3 1 12
3 1 18
4 1 3
4 1 5
4 1 21
4 1 26
5 1 13
5 1 17
5 1 28
5 1 30
6 1 14
6 1 19
6 1 24
6 1 27
7 1 8
7 1 9
7 1 11
7 1 17
8 1 2
8 1 26
8 1 29
8 1 31
9 1 1
9 1 8
9 1 22
9 1 28
10 1 6
10 1 10
10 1 20
10 1 22
11 1 11
11 1 23
11 1 30
11 1 31
12 1 4
12 1 5
12 1 18
12 1 24
13 1 2
13 1 7
13 1 15
13 1 28
14 1 5
14 1 9
14 1 15
14 1 23
15 1 1
15 1 15
15 1 29
15 1 30


My goal is to be able to replicate the above table with and SPSS syntax and then used it in another different example with different number of messages, screeners and blocks.

I was thinking of random numbers...

Thanks in advance

[David Marso]

Here's a nudge off of the MATRIZ cliff ;-)
MATRIX.
COMPUTE msg=T({{1:31};GRADE(UNIFORM(1,31))}).
COMPUTE sorted=MAKE(31,2,0).
LOOP #=1 TO 31.
COMPUTE sorted(msg(#,2),:)=msg(#,:).
END LOOP.
PRINT msg.
PRINT sorted.
END MATRIX.

David Marso wrote

Give it a try and if you encounter problems someone will likely have suggestions.
See INPUT PROGRAM.
See MATRIX.
--

mils wrote

Hi everyone,

I would like to create the table/file as per shown below. There are 15 screeners, 4 blocks and 31 messages.
In the example below, its shown one block, 15 screeners, each one with a different set of 4 random messages. The idea behind the blocks is to try to get different sets of screeners. Due to the number of messages and screener some (messages) had to be repeated within the blocks.

Screener Block Message
1 1 16
1 1 20
1 1 25
1 1 30
2 1 8
2 1 13
2 1 16
2 1 18
3 1 1
3 1 6
3 1 12
3 1 18
4 1 3
4 1 5
4 1 21
4 1 26
5 1 13
5 1 17
5 1 28
5 1 30
6 1 14
6 1 19
6 1 24
6 1 27
7 1 8
7 1 9
7 1 11
7 1 17
8 1 2
8 1 26
8 1 29
8 1 31
9 1 1
9 1 8
9 1 22
9 1 28
10 1 6
10 1 10
10 1 20
10 1 22
11 1 11
11 1 23
11 1 30
11 1 31
12 1 4
12 1 5
12 1 18
12 1 24
13 1 2
13 1 7
13 1 15
13 1 28
14 1 5
14 1 9
14 1 15
14 1 23
15 1 1
15 1 15
15 1 29
15 1 30


My goal is to be able to replicate the above table with and SPSS syntax and then used it in another different example with different number of messages, screeners and blocks.

I was thinking of random numbers...

Thanks in advance

[mils]

I'm thinking now, that maybe what I need is some sort of combination/permutation? The random numbers was giving me duplicates...

In my example, each respondent will see a combination of 4 messages (out of 31) 15 times. Note that each respondent will be exposed only to one block (out of 4). Each block needs to have a unique set of 4 messages each "screener" (15)

In other words if a respondent is exposed to the following combination 23,3,15,31, this can not be shown again (for example 3,23,15,31) in any other block or "screener"

I've found a syntax from David, that does kind of something I would like to do (to late to think), but I was wondering if it possible to changed to 4 columns and permutation of the number 1 to 31?

SET MXLOOPS=10000000.
/* Exhaustive list */.
MATRIX.
COMPUTE x=9.
COMPUTE p=T({1:x}).

LOOP #=1 TO x-1.
+  COMPUTE p={KRONEKER(T({1:x}),MAKE(NROW(p),1,1)),KRONEKER(MAKE(x,1,1),p)}.
+  COMPUTE goodrow=1.
+  LOOP ##=1 TO NROW(p).
+    COMPUTE target=p(##,1).
+    LOOP ###=2 TO NCOL(p).
+        COMPUTE found=(p(##,###) EQ target).
+    END LOOP IF found.
+    DO IF (found EQ 0).
+      COMPUTE p(goodrow,:)=p(##,:).
+      COMPUTE goodrow=goodrow+1.
+    END IF.
+  END LOOP.
+  COMPUTE p=p(1:(goodrow-1),:).
END LOOP.
SAVE p / OUTFILE * .
END MATRIX.

I've try to change a few variables, but either didn't work or crashed :-(

*the screener is like a card that contain the messages shown to the respondent. sorry for the confusion.

[David Marso]

People sometimes pay me to think about this sort of thing.
---

mils wrote

I'm thinking now, that maybe what I need is some sort of combination/permutation? The random numbers was giving me duplicates...

In my example, each respondent will see a combination of 4 messages (out of 31) 15 times. Note that each respondent will be exposed only to one block (out of 4). Each block needs to have a unique set of 4 messages each "screener" (15)

In other words if a respondent is exposed to the following combination 23,3,15,31, this can not be shown again (for example 3,23,15,31) in any other block or "screener"

I've found a syntax from David, that does kind of something I would like to do (to late to think), but I was wondering if it possible to changed to 4 columns and permutation of the number 1 to 31?

SET MXLOOPS=10000000.
/* Exhaustive list */.
MATRIX.
COMPUTE x=9.
COMPUTE p=T({1:x}).

LOOP #=1 TO x-1.
+  COMPUTE p={KRONEKER(T({1:x}),MAKE(NROW(p),1,1)),KRONEKER(MAKE(x,1,1),p)}.
+  COMPUTE goodrow=1.
+  LOOP ##=1 TO NROW(p).
+    COMPUTE target=p(##,1).
+    LOOP ###=2 TO NCOL(p).
+        COMPUTE found=(p(##,###) EQ target).
+    END LOOP IF found.
+    DO IF (found EQ 0).
+      COMPUTE p(goodrow,:)=p(##,:).
+      COMPUTE goodrow=goodrow+1.
+    END IF.
+  END LOOP.
+  COMPUTE p=p(1:(goodrow-1),:).
END LOOP.
SAVE p / OUTFILE * .
END MATRIX.

I've try to change a few variables, but either didn't work or crashed :-(

*the screener is like a card that contain the messages shown to the respondent. sorry for the confusion.

[King Douglas]

It's true.  I've seen it happen.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
David Marso
Sent: Friday, May 01, 2015 1:33 AM
To: [hidden email]
Subject: Re: Message exposure

People sometimes pay me to think about this sort of thing.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[mils]

I can't blame them :-)

I've ended up using a different software with some random formula. This software have actually a Maxdiff design, that used in the right way, gave me what I wanted. Thanks!

However, I'm still interested in understanding how many blocks do I need and how many times do I need to show the  screeners with 4 messages to make sure messages have the same level of exposure.

Any suggestion will be really appreciate it!

Thanks.

[David Marso]

Do a web search on Youden Square Designs
You might want to read some literature on the design of experiments before venturing much further.