One question about multiple response

[Mojdeh Jamnia]

Hi there

I am a PhD student who is analyasing the survey data in SPSS.
I have decided to collapse my multiple response question which starts from
1 to 16 into 4 separate categories,I am expecting one single Count for
each category but still I have the answer from 1-16 and apparently they
haven't been collapsed.I should add that all these data are recoded.

I would be grateful if anyone could answer this question.

Regards
Mojdeh

[Sibusiso Moyo]

Dear All,

The following command applies only to numeric variables (i believe):

Select if VAr > 0.
exe.

What is the equivalent for string variables?

Sibusiso.

[Dominic Lusinchi]

Mojdeh,

Without seeing your syntax it's hard to determine what you are doing, but
perhaps you can use a DO IF-END IF statement. Maybe I am not understanding
your issue...it has been known. :-)

DO IF (oldvar LE 4).
COMPUTE newvar=1.
ELSE IF (oldvar GT 4 & oldvar LE 8).
COMPUTE newvar=2.
ELSE IF (oldvar GT 8 & oldvar LE 12).
COMPUTE newvar=3.
ELSE.
COMPUTE newvar=4.
END IF.
EXE.

If you want to do variable and value labels put the syntax before the "EXE"
statement.

Good luck.
Dominic

Dominic Lusinchi
Statistician
Far West Research
Statistical Consulting
San Francisco, California
415-664-3032
www.farwestresearch.com

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Mojdeh Jamnia
Sent: Tuesday, August 29, 2006 8:14 AM
To: [hidden email]
Subject: One question about multiple response

Hi there

I am a PhD student who is analyasing the survey data in SPSS.
I have decided to collapse my multiple response question which starts from
1 to 16 into 4 separate categories,I am expecting one single Count for
each category but still I have the answer from 1-16 and apparently they
haven't been collapsed.I should add that all these data are recoded.

I would be grateful if anyone could answer this question.

Regards
Mojdeh

[Edward Boadi]

Mojdeh,


**First recode the variables into 4 categories .
***** Here v1 to V6 was recoded into 4 categories , change classification to satisfy your needs .

RECODE V1 TO V6  (1 THRU 4 =1) (5 THRU 9 =2) (9 THRU 12 =3) (13 THRU 16 = 4) INTO NV1 NV2 NV3 NV4 NV5 NV6 .
EXECUTE.

*****This syntax creates a multiple response set for the 4 categories .
MRSETS
/MCGROUP NAME=$Responses LABEL='Responses Provided' VARIABLES=NV1 NV2 NV3 NV4 NV5 NV6
/DISPLAY NAME=[$Responses].


Regards.
Edward

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Mojdeh Jamnia
Sent: Tuesday, August 29, 2006 11:14 AM
To: [hidden email]
Subject: One question about multiple response


Hi there

I am a PhD student who is analyasing the survey data in SPSS.
I have decided to collapse my multiple response question which starts from
1 to 16 into 4 separate categories,I am expecting one single Count for
each category but still I have the answer from 1-16 and apparently they
haven't been collapsed.I should add that all these data are recoded.

I would be grateful if anyone could answer this question.

Regards
Mojdeh

[Vivek Kurup]

Sibusiso,

This should help:

select if VAR>"VALUE".

Vivek


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Sibusiso Moyo
Sent: Tuesday, August 29, 2006 11:32
To: [hidden email]
Subject: Re: Selecting non-missing cases in string variables


Dear All,

The following command applies only to numeric variables (i believe):

Select if VAr > 0.
exe.

What is the equivalent for string variables?

Sibusiso.

[Beadle, ViAnn]

1. How did you collapse the variables in your multiple response set?
2. How do you know that they weren't collapsed. Which procedure or dialog box did you use to obtain the Count of categories?

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mojdeh Jamnia
Sent: Tuesday, August 29, 2006 10:14 AM
To: [hidden email]
Subject: One question about multiple response

Hi there

I am a PhD student who is analyasing the survey data in SPSS.
I have decided to collapse my multiple response question which starts from
1 to 16 into 4 separate categories,I am expecting one single Count for
each category but still I have the answer from 1-16 and apparently they
haven't been collapsed.I should add that all these data are recoded.

I would be grateful if anyone could answer this question.

Regards
Mojdeh

[Edward Boadi]

Your subject heading contradicts what you want to do.

Supposing VAr is a string variable and you want to select non-missing cases in this variable then
use the syntax below :

SELECT IF (NOT MISSING(VAr)) .
EXECUTE.

Thanks.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Sibusiso Moyo
Sent: Tuesday, August 29, 2006 11:32 AM
To: [hidden email]
Subject: Re: Selecting non-missing cases in string variables


Dear All,

The following command applies only to numeric variables (i believe):

Select if VAr > 0.
exe.

What is the equivalent for string variables?

Sibusiso.

[Marta García-Granero]

Hi Sibusiso

Concepts like greater or lower can't be applied to string variables,
unless you are referring to text variables that contain only numbers
(but that were read as strings). If you are talking about that, then
try using NUMBER, like in this example:

DATA LIST LIST/numstring(A1).
BEGIN DATA
0
1
2
3
4
5
END DATA.
LIST.

TEMPORARY.
SELECT IF NUMBER(numstring,F1) > 0.
LIST.

HTH,
Marta

SM> Dear All,

SM> The following command applies only to numeric variables (i believe):

Select if VAr >> 0.
SM> exe.

SM> What is the equivalent for string variables?

[Marta García-Granero]

Hi Edward

You wrote:

EB> Your subject heading contradicts what you want to do.

EB> Supposing VAr is a string variable and you want to select
EB> non-missing cases in this variable then
EB> use the syntax below :

EB> SELECT IF (NOT MISSING(VAr)) .
EB> EXECUTE.

(I hadn't read the subject heading, just the body message, now I
understand a bit more Sibusiso's question). But...

String variables don't have missing values. You must try something
like this:

DATA LIST LIST/stringvar(A16) numvar(F2).
BEGIN DATA
""               1
"Not missing 1"  2
"Not missing 2"  3
"Not missing 3"  4
"Not missing 4"  5
""               6
END DATA.
LIST.

TEMPORARY.
SELECT IF LENGHT(RTRIM(stringvar)) NE 0.
LIST.




--
Regards,
Marta

[Edward Boadi]

Thanks for the correction Marta

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Marta García-Granero
Sent: Tuesday, August 29, 2006 11:54 AM
To: [hidden email]
Subject: Re: Selecting non-missing cases in string variables


Hi Edward

You wrote:

EB> Your subject heading contradicts what you want to do.

EB> Supposing VAr is a string variable and you want to select
EB> non-missing cases in this variable then
EB> use the syntax below :

EB> SELECT IF (NOT MISSING(VAr)) .
EB> EXECUTE.

(I hadn't read the subject heading, just the body message, now I
understand a bit more Sibusiso's question). But...

String variables don't have missing values. You must try something
like this:

DATA LIST LIST/stringvar(A16) numvar(F2).
BEGIN DATA
""               1
"Not missing 1"  2
"Not missing 2"  3
"Not missing 3"  4
"Not missing 4"  5
""               6
END DATA.
LIST.

TEMPORARY.
SELECT IF LENGHT(RTRIM(stringvar)) NE 0.
LIST.




--
Regards,
Marta

[Art Kendall]

In the US, ASCII character order is used.
It is often useful to convert to lower or upper case before testing strings.

There is example syntax below the sig block.

Save all your current work, then open a new instance of SPSS. Make sure
that you put warnings, etc. into the output file. <edit> <options>
<viewer>. Cut-and-paste then run the syntax.
Does this do what you want?  If not please restate your question in more
detail.
Perhaps post a small data set with variables "have" and "want".


Art

Social Research Consultants


data list list /x (a1) y (a1).
begin data
a A
a b
a B
n m
X x
a ' '
' ' 1
0 1
end data.
compute islessthan = x lt y .
list.

Sibusiso Moyo wrote:

>Dear All,
>
>The following command applies only to numeric variables (i believe):
>
>Select if VAr > 0.
>exe.
>
>What is the equivalent for string variables?
>
>Sibusiso.
>
>
>
>

[Peck, Jon]

A little clarification:
String variables do not have SYSMIS.
String variables with length <=8 do support user missing, but not ranges.
The MISSING function, unlike the other functions for missing values, may be used with either numeric or string variables in recent versions of SPSS.  (I don't recall whether that was changed in 13 or 14.)

Regards,
Jon Peck
SPSS

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Marta García-Granero
Sent: Tuesday, August 29, 2006 10:54 AM
To: [hidden email]
Subject: Re: [SPSSX-L] Selecting non-missing cases in string variables

Hi Edward

You wrote:

EB> Your subject heading contradicts what you want to do.

EB> Supposing VAr is a string variable and you want to select
EB> non-missing cases in this variable then
EB> use the syntax below :

EB> SELECT IF (NOT MISSING(VAr)) .
EB> EXECUTE.

(I hadn't read the subject heading, just the body message, now I
understand a bit more Sibusiso's question). But...

String variables don't have missing values. You must try something
like this:

[Marta García-Granero]

Hi Jon

"Thou shalt not forget thy children who dwell in darkness, far from the
light of SPSS 14..."  ;)

I was thinking "elevenish", I didn't know that newer SPSS versions
handled missing values in string variables (glad to hear that, makes
my life as a teacher easier). I have just updated to SPSS 13 after a
loooong time stuck with SPSS 11.5, and not a long time with SPSS 12
(and I should have been glad, since there still some computers with
SPSS 10 at the University, things can go very slowly in some places).

I stand corrected, thanks.

Marta

PJ> A little clarification:
PJ> String variables do not have SYSMIS.
PJ> String variables with length <=8 do support user missing, but not ranges.
PJ> The MISSING function, unlike the other functions for missing
PJ> values, may be used with either numeric or string variables in
PJ> recent versions of SPSS.  (I don't recall whether that was changed
PJ> in 13 or 14.)

[Sibusiso Moyo]

Dear All,

All I needed to do was select cases that have information. My string variable has information about types of a variable, so there is 21 of those types or so. So a case ought to have some form of variable type, and when that information is not there I needed to drop the case.

Here is the physician list!

                   1
ALI                2  ALI
CD                 3  CD
EM                 4  EM
END                5  END
GE                 6  GE
ID                 7  ID
N                  8  N
NEP                9  NEP
OBG               10  OBG
ONC               11  ONC
OPT               12  OPT
ORS               13  ORS
OTO               14  OTO
PCP               15  PCP
PED               16  PED
PS                17  PS
PSY               18  PSY
PUD               19  PUD
RHU               20  RHU
URO               21  URO

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Marta García-Granero
Sent: Tuesday, August 29, 2006 11:54 AM
To: [hidden email]
Subject: Re: Selecting non-missing cases in string variables


Hi Edward

You wrote:

EB> Your subject heading contradicts what you want to do.

EB> Supposing VAr is a string variable and you want to select
EB> non-missing cases in this variable then
EB> use the syntax below :

EB> SELECT IF (NOT MISSING(VAr)) .
EB> EXECUTE.

(I hadn't read the subject heading, just the body message, now I
understand a bit more Sibusiso's question). But...

String variables don't have missing values. You must try something
like this:

DATA LIST LIST/stringvar(A16) numvar(F2).
BEGIN DATA
""               1
"Not missing 1"  2
"Not missing 2"  3
"Not missing 3"  4
"Not missing 4"  5
""               6
END DATA.
LIST.

TEMPORARY.
SELECT IF LENGHT(RTRIM(stringvar)) NE 0.
LIST.




--
Regards,
Marta

[Marta García-Granero]

Hi Sibusiyo

Since you don't mention the SPSS version you are using (nor the name
of the string variable, I'll assume it is something like "physician"),
just to make sure that it works, use:

SELECT IF LENGHT(RTRIM(physician)) NE 0.

This will eliminate unwanted cases.

SM> All I needed to do was select cases that have information. My
SM> string variable has information about types of a variable, so
SM> there is 21 of those types or so. So a case ought to have some
SM> form of variable type, and when that information is not there I
SM> needed to drop the case.

SM> Here is the physician list!

SM>                    1
SM> ALI                2  ALI
SM> CD                 3  CD
SM> EM                 4  EM
SM> END                5  END
SM> GE                 6  GE
SM> ID                 7  ID
SM> N                  8  N
SM> NEP                9  NEP
SM> OBG               10  OBG
SM> ONC               11  ONC
SM> OPT               12  OPT
SM> ORS               13  ORS
SM> OTO               14  OTO
SM> PCP               15  PCP
SM> PED               16  PED
SM> PS                17  PS
SM> PSY               18  PSY
SM> PUD               19  PUD
SM> RHU               20  RHU
SM> URO               21  URO


Regards,
Marta

[Sibusiso Moyo]

I am using Version 14. I have actually used the following syntax, and it seemed to work:

Select if Cell ne ("").
exe.

Where Physician = Cell.

Thanks for all your input.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Marta García-Granero
Sent: Tuesday, August 29, 2006 1:54 PM
To: [hidden email]
Subject: Re: Selecting non-missing cases in string variables


Hi Sibusiyo

Since you don't mention the SPSS version you are using (nor the name
of the string variable, I'll assume it is something like "physician"),
just to make sure that it works, use:

SELECT IF LENGHT(RTRIM(physician)) NE 0.

This will eliminate unwanted cases.

SM> All I needed to do was select cases that have information. My
SM> string variable has information about types of a variable, so
SM> there is 21 of those types or so. So a case ought to have some
SM> form of variable type, and when that information is not there I
SM> needed to drop the case.

SM> Here is the physician list!

SM>                    1
SM> ALI                2  ALI
SM> CD                 3  CD
SM> EM                 4  EM
SM> END                5  END
SM> GE                 6  GE
SM> ID                 7  ID
SM> N                  8  N
SM> NEP                9  NEP
SM> OBG               10  OBG
SM> ONC               11  ONC
SM> OPT               12  OPT
SM> ORS               13  ORS
SM> OTO               14  OTO
SM> PCP               15  PCP
SM> PED               16  PED
SM> PS                17  PS
SM> PSY               18  PSY
SM> PUD               19  PUD
SM> RHU               20  RHU
SM> URO               21  URO


Regards,
Marta

[Sibusiso Moyo]

Dear All,

I have about 13000 cases in my data set. I have these two variables that I am trying to test how close they are from each other. Specifically I am trying to identify cases, where the biggest difference arises. Besides taking differences between the two, sorting the data by the resulting difference variable, plots etc, would anyone have any ideas how one would proceed with such an assignment.

Thanks,

Sibusiso.

[Richard Ristow]

Jon Peck of SPSS, Inc., had written

>>A little clarification:
>>String variables do not have SYSMIS.
>>String variables with length <=8 do support
>>user missing, but not ranges.
>>The MISSING function, unlike the other
>>functions for missing values, may be used with
>>either numeric or string variables in recent
>>versions of SPSS.  (I don't recall whether that
>>was changed in 13 or 14.)


At 01:16 PM 8/29/2006, Marta García-Granero wrote:
>Hi Jon
>
>"Thou shalt not forget thy children who dwell in
>darkness, far from the light of SPSS 14..."  ;)
>
>I was thinking "elevenish", I didn't know that
>newer SPSS versions handled missing values in string variables

I think what Jon was saying was that the MISSING
*function* started working with (short) string
variables in 13 or 14. I believe that missing
values have been supported for short string
variables from back in the days where there was
only mainframe SPSS. ('Short' is 8 or fewer
characters.) Below is draft output run with SPSS 9.

Now, second only to warnings not to use EXECUTE,
I have suggestions for SPSS.

Missing values for long strings would be VERY
useful in one important case: to mask certain values from AGGREGATE.

If you want only AGGREGATE to 'see' only values
for cases that pass a certain test, with numbers
and short strings, you can make other values
missing (using TEMPORARY and RECODE, if that
helps). There's no such recourse, for long
strings. The best I know of, is to recode the
values you don't want into something that
compares lower than anything you do want (often,
blank fits that criterion) and use the MAX function in AGGREGATE.

Probably, the reason missing values are not
supported for long strings, is that the
dictionary data structure has 8 bytes, and only 8
bytes, for the values. But you could still allow
missing values for long strings, accepting only
8-character values, and either,
. Take a string as missing if its first 8
characters match the declared missing value, and
the remainder of the string is blank
. Take a string as missing if its first 8
characters match the declared missing value,
regardless of the contents of the remaining characters.

OK, off soapbox, and here's the test code. Again,
this is SPSS 9, draft output.

NEW FILE.
DATA LIST FREE /STRING08(A08).
BEGIN DATA
Alpha Beta Gamma Delta Skip
END DATA.

STRING  String12(A12).
VARIABLE LABELS String08 "An 8-character string variable".
VARIABLE LEVEL String08 (NOMINAL).
VARIABLE ALIGNMENT String08(LEFT).
VARIABLE LABELS String12 "The 12-character string variable".
VARIABLE LEVEL String12 (NOMINAL).
VARIABLE WIDTH String12(12).
VARIABLE ALIGNMENT String12(LEFT).
COMPUTE String12 = String08.
LIST.

List
STRING08 STRING12

Alpha    Alpha
Beta     Beta
Gamma    Gamma
Delta    Delta
Skip     Skip

Number of cases read:  5    Number of cases listed:  5




*  Here's the test:                 .
Missing values STRING08 ('Skip').
VALUE LABELS String08
  "Skip    " "The missing value".
Missing values STRING12('Skip').

 >Error # 4808 in column 16.  Text: STRING12
 >Missing values are not available for long string variables.
 >This command not executed.

VALUE LABELS String12
  "Skip    " "The missing value".

 >Warning # 4484.  Command name: VALUE LABELS
 >The (ADD) VALUE LABELS command has attempted to
apply a label to a variable
 >the values of which are long character strings.  The variable will be
 >ignored.

 >The error is associated with 'STRING12'


 >Warning # 4487 in column 2.  Text: Skip
 >The (ADD) VALUE LABELS command specifies the
labeling of a value which is
 >inconsistent with the type of the variables to
be labeled.  The value and
 >the applicable label will be ignored.  This error may occur because a
 >numeric value was omitted and the system
mistook a label for a character
 >value.


FREQUENCIES String08 String12.

Frequencies
Statistics
|---------|------------------------------|--------------------------------|
|         |An 8-character string variable|The
12-character string variable|
|-|-------|------------------------------|--------------------------------|
|N|Valid  |4                             |5
                         |
|
|-------|------------------------------|--------------------------------|
|
|Missing|1                             |0
                       |
|-|-------|------------------------------|--------------------------------|
¢

Frequency Table

An 8-character string variable

|-------------------------|---------|-------|-------------|------------------|
|
|Frequency|Percent|Valid Percent|Cumulative Percent|
|-------|-----------------|---------|-------|-------------|------------------|
|Valid  |Alpha            |1        |20.0   |25.0
         |25.0              |
|
|-----------------|---------|-------|-------------|------------------|
|       |Beta             |1        |20.0   |25.0
         |50.0              |
|
|-----------------|---------|-------|-------------|------------------|
|       |Delta            |1        |20.0   |25.0
         |75.0              |
|
|-----------------|---------|-------|-------------|------------------|
|       |Gamma            |1        |20.0   |25.0
         |100.0             |
|
|-----------------|---------|-------|-------------|------------------|
|       |Total            |4        |80.0
|100.0        |                  |
|-------|-----------------|---------|-------|-------------|------------------|
|Missing|The missing
value|1        |20.0   |             |                  |
|-------|-----------------|---------|-------|-------------|------------------|
|Total                    |5        |100.0  |
         |                  |
|-------------------------|---------|-------|-------------|------------------|
¢

The 12-character string variable

|-----------|---------|-------|-------------|------------------|
|           |Frequency|Percent|Valid Percent|Cumulative Percent|
|-----|-----|---------|-------|-------------|------------------|
|Valid|Alpha|1        |20.0   |20.0         |20.0              |
|     |-----|---------|-------|-------------|------------------|
|     |Beta |1        |20.0   |20.0         |40.0              |
|     |-----|---------|-------|-------------|------------------|
|     |Delta|1        |20.0   |20.0         |60.0              |
|     |-----|---------|-------|-------------|------------------|
|     |Gamma|1        |20.0   |20.0         |80.0              |
|     |-----|---------|-------|-------------|------------------|
|     |Skip |1        |20.0   |20.0         |100.0             |
|     |-----|---------|-------|-------------|------------------|
|     |Total|5        |100.0  |100.0        |                  |
|-----|-----|---------|-------|-------------|------------------|

[Richard Ristow]

(You posted with a blank subject heading; you probably know this, but
it makes reading and replying more difficult. I've put in a guess.)

At 03:07 PM 8/29/2006, Sibusiso Moyo wrote:

>I have about 13000 cases in my data set. I have these two variables
>that I am trying to test how close they are from each other.
>Specifically I am trying to identify cases, where the biggest
>difference arises. Besides taking differences between the two, sorting
>the data by the resulting difference variable, plots etc, would anyone
>have any ideas how one would proceed with such an assignment.

So, these are numeric variables? If they are strings, a crucial new
question arises: how do you measure the 'difference' between two
values.

You list pretty much how anyone would start to do it. I'm not sure what
plots you're doing - stem-and-leaf, to see something of the
distribution of the differences?

Is there any way in which what you're doing doesn't satisfy you? If so,
say so, and maybe we'll have ideas. (I recommend new posts under the
same subject heading "Re: [Differences in variable values]".)

-Good luck,
  Richard Ristow

[Sibusiso Moyo]

Richard,

I thought I had a title in there. Apologies.

Well I have two series of numeric data which I was hoping are similar/highly correlated/the same so I could use them interchangeably in my analysis. But I know that they are not because they come from different sources. So my initial take on them was to plot both series across time, to see how they vary.

My second level of analysis was to take this data set and identify those particular cases that are giving us the most difference. With a large data set I didnt want to carry out this part manually, so I thought someone on the list might have done this before.

I might have solved the problem for now though (i hope),

Thanks for your input.

Sibusiso.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Richard Ristow
Sent: Tuesday, August 29, 2006 5:01 PM
To: [hidden email]
Subject: Re: [Differences in variable values]


(You posted with a blank subject heading; you probably know this, but
it makes reading and replying more difficult. I've put in a guess.)

At 03:07 PM 8/29/2006, Sibusiso Moyo wrote:

>I have about 13000 cases in my data set. I have these two variables
>that I am trying to test how close they are from each other.
>Specifically I am trying to identify cases, where the biggest
>difference arises. Besides taking differences between the two, sorting
>the data by the resulting difference variable, plots etc, would anyone
>have any ideas how one would proceed with such an assignment.

So, these are numeric variables? If they are strings, a crucial new
question arises: how do you measure the 'difference' between two
values.

You list pretty much how anyone would start to do it. I'm not sure what
plots you're doing - stem-and-leaf, to see something of the
distribution of the differences?

Is there any way in which what you're doing doesn't satisfy you? If so,
say so, and maybe we'll have ideas. (I recommend new posts under the
same subject heading "Re: [Differences in variable values]".)

-Good luck,
  Richard Ristow