SPSSX Discussion

Dummy variables to provide section labels in tables

Classic

List

Threaded

3 messages Options

Dennis Deck

Dummy variables to provide section labels in tables

Occasionly I produce tables to summarize the item results from a test or survey using MEANS (or SUMMARIZE). To make the labeling more informative in a long list of items, I create dummy variables with labels but missing values. These provide labels to mark the beginning of the next subscale (eg, "Scale 1: School Policies") in the table. However, MEANS (and SUMMARIZE) now drop any fields with no values.

* Example of approach that no longer works .
COMPUTE Scale1 = $sysmis .
COMPUTE Scale2 = $sysmis .
VARIABLE LABELS
Scale1 'Scale 1 Items'
Scale2 'Scale 2 Items' .
MEANS
Tables= Scale1 Item01 to Item20
Scale2 Item21 to Item40 by Group
/Cells= mean .

I tried specifying /Missing=Include but only got the warning message that those fields had no data and no table.
I believe that this indicates an error in the behavior of /Missing=Include .

Dennis Deck, PhD
RMC Research Corporation
111 SW Columbia Street, Suite 1200
Portland, Oregon 97201-5843
voice: 503-223-8248 x715
voice: 800-788-1887 x715
fax: 503-223-8248
[hidden email]

Dennis Deck, PhD
RMC Research Corporation
111 SW Columbia Street, Suite 1200
Portland, Oregon 97201-5843
voice: 503-223-8248 x715
voice: 800-788-1887 x715
fax: 503-223-8248
[hidden email]

====================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

Richard Ristow

Re: Dummy variables to provide section labels in tables

At 07:07 PM 3/18/2008, Dennis Deck wrote:

>Occasionally I produce tables to summarize the item results from a
>test or survey using MEANS (or SUMMARIZE). To make the labeling
>more informative in a long list of items, I create dummy variables
>with labels but missing values. These provide labels to mark the
>beginning of the next subscale (eg, "Scale 1: School Policies") in
>the table. However, MEANS (and SUMMARIZE) now drop any fields with
>no values.
>
>* Example of approach that no longer works .
>COMPUTE Scale1 = $sysmis .
>COMPUTE Scale2 = $sysmis .
>VARIABLE LABELS
> Scale1 'Scale 1 Items'
> Scale2 'Scale 2 Items' .
>MEANS
> Tables= Scale1 Item01 to Item20
> Scale2 Item21 to Item40 by Group
> /Cells= mean .

How about this work-around? It's draft output (WRR:not saved
separately), which I've edited some, to be shorter and clearer.

COMPUTE Scale1 = $sysmis .
COMPUTE Scale2 = $sysmis .
IF $CASENUM EQ 1 Scale1=0 /* Added as work-around */.
IF $CASENUM EQ 1 Scale2=0 /* Added as work-around */.
VARIABLE LABELS
Scale1 'Scale 1 Items'
Scale2 'Scale 2 Items' .
MEANS
Tables= Scale1 Itm1 to Itm4
Scale2 Itm5 to Itm8 by Group
/Cells= mean .

Means
|-----------------------------|---------------------------|
|Output Created |18-MAR-2008 23:44:25 |
|-----------------------------|---------------------------|
Case Processing Summary
|---------------|--------------------------------------------|
| |Cases |
| |---------------|---------------|------------|
| |Included |Excluded |Total |
| |-------|-------|-------|-------|----|-------|
| |N |Percent|N |Percent|N |Percent|
|---------------|-------|-------|-------|-------|----|-------|
|Scale1 Scale 1|1 |3.3% |29 |96.7% |30 |100.0% |
|Items * Group | | | | | | |
|---------------|-------|-------|-------|-------|----|-------|
|Itm1 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm2 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm3 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm4 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|---------------|-------|-------|-------|-------|----|-------|
|Scale2 Scale 2|1 |3.3% |29 |96.7% |30 |100.0% |
|Items * Group | | | | | | |
|---------------|-------|-------|-------|-------|----|-------|
|Itm5 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm6 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm7 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|Itm8 * Group |30 |100.0% |0 |.0% |30 |100.0% |
|---------------|-------|-------|-------|-------|----|-------|

Report
Mean
|-----|-------|----|----|----|----|-------|----|----|----|----|
|Group|Scale1 |Itm1|Itm2|Itm3|Itm4|Scale2 |Itm5|Itm6|Itm7|Itm8|
| |Scale 1| | | | |Scale 2| | | | |
| |Items | | | | |Items | | | | |
|-----|-------|----|----|----|----|-------|----|----|----|----|
|01 |.0000 |3.00|3.25|2.58|3.67|.0000 |3.00|2.75|3.33|3.00|
|-----|-------|----|----|----|----|-------|----|----|----|----|
|02 | |2.63|2.63|3.13|2.13| |3.25|2.63|2.38|3.50|
|-----|-------|----|----|----|----|-------|----|----|----|----|
|03 | |3.30|3.00|2.80|3.10| |3.10|3.50|2.20|3.10|
|-----|-------|----|----|----|----|-------|----|----|----|----|
|Total|.0000 |3.00|3.00|2.80|3.07|.0000 |3.10|2.97|2.70|3.17|
|-----|-------|----|----|----|----|-------|----|----|----|----|

===================
APPENDIX: Test data
===================
* ................................................................. .
* ................. Test data ..................... .
SET RNG = MT /* 'Mersenne twister' random number generator */ .
SET MTINDEX = 6920 /* Providence, RI telephone book */ .

NEW FILE.
INPUT PROGRAM.
. NUMERIC CaseID (N3).
. LEAVE CaseID.
. NUMERIC Group (N2).
. NUMERIC Itm1 TO Itm8 (F2).
. VECTOR ITEM
=Itm1 TO Itm8.
. LOOP CaseID = 1 TO 30.
. COMPUTE Group = TRUNC(RV.UNIFORM(1,4)).
. LOOP #ItemNum = 1 TO 8.
. COMPUTE ITEM(#ItemNum) = TRUNC(RV.UNIFORM(1,6)).
. END LOOP.
. END CASE.
. END LOOP.
END FILE.
END INPUT PROGRAM.

LIST.

List
|-----------------------------|---------------------------|
|Output Created |18-MAR-2008 23:44:24 |
|-----------------------------|---------------------------|
CaseID Group Itm1 Itm2 Itm3 Itm4 Itm5 Itm6 Itm7 Itm8

001 01 3 2 5 4 1 1 2 1
002 02 3 5 2 2 1 5 2 5
003 01 1 2 2 3 1 4 5 4
004 02 1 2 5 1 5 5 1 3
005 02 2 2 3 4 2 2 4 4
006 03 3 3 3 1 3 2 1 5
007 01 2 3 2 5 3 5 1 3
008 01 4 4 3 4 1 1 1 3
009 02 4 2 2 1 3 3 1 3
010 02 4 1 5 4 5 2 4 5
011 02 2 1 2 2 5 1 4 4
012 03 3 3 1 4 4 3 4 3
013 01 3 5 2 3 5 5 5 5
014 02 2 4 4 1 2 2 1 2
015 03 2 4 1 2 3 5 2 5
016 03 2 4 3 4 3 2 2 2
017 03 5 3 3 4 4 5 2 4
018 01 2 4 1 3 5 1 4 5
019 01 3 4 1 5 3 1 4 3
020 01 4 1 3 5 1 3 4 3
021 01 2 2 2 4 5 3 4 1
022 03 4 3 5 2 1 4 1 1
023 03 5 3 3 5 4 5 2 5
024 01 4 5 3 4 4 1 2 5
025 03 3 4 2 3 3 5 2 1
026 02 3 4 2 2 3 1 2 2
027 03 5 2 5 1 4 3 2 3
028 01 5 2 3 1 4 3 3 2
029 01 3 5 4 3 3 5 5 1
030 03 1 1 2 5 2 1 4 2

Number of cases read: 30 Number of cases listed: 30

=====================
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

Dennis Deck

Re: Dummy variables to provide section labels in tables

Thanks Richard.

That creates the dummy variables all right.
Unfortunately there is no way to hide the Scale1/Scale2 means of zero,
something I was trying to avoid.

These would, however, serve as placeholders that could be fixed with a
manual edit.
I may be able to a) hide these values using FORMATS or b) search and
replace in Word during final editing.

Dennis

-----Original Message-----
From: Richard Ristow [mailto:[hidden email]]
Sent: Tuesday, March 18, 2008 8:51 PM
To: Dennis Deck; [hidden email]
Subject: Re: Dummy variables to provide section labels in tables

At 07:07 PM 3/18/2008, Dennis Deck wrote: