Compute command and System User missing

[J McClure]

Hi,
I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
trying to do something quite simple but am running into a problem with
system missing and user missing values.
Old variables are H_now (no missing values) and H_12mo which has 3 system
missing and 7 user missing (99).
I want to create a new variable (hosp). I set the new variable to 8 so that
any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
compute hosp=8.
if H_now EQ 1  hosp = 1.
if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
If H_12mo EQ 99  hosp = 98.
Variable labels hosp 'outcome'.
Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
Freq
/var=hosp
/missing=include.
Thanks for any help,
Jan

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

[Rich Ulrich]

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 8.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich

> Date: Thu, 3 Nov 2011 12:26:12 -0400

> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1 hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99 hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.

[ ... ]

[J McClure]

Hi,
Sorry, I see that I wasn't specific.
The results of the code I posted are 10 cases that fall into the category of 8.  I am using 8 only to see where my code does not work. In other words, if there are any '8' s in my new variable I did something wrong.
I traced the source of my 10 cases that were incorrectly coded as an '8' to the variable h_12mo which has 7 cases that are user missing (coded as 99) and 3 cases that are system missing. 
In the new variable I want the user missing to be a 98 and the system missing to still be system missing.
Thanks,
Jan

On 11/3/2011 10:37 AM, Rich Ulrich wrote:

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 8.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich

> Date: Thu, 3 Nov 2011 12:26:12 -0400
> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1 hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99 hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.
[ ... ]

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

[John F Hall]

Jan

Try taking the slash out of your freq command:

freq hosp /mis inc.     [abbreviated syntax]

Not sure exactly what you've done with missing values or what they are, but
SPSS only allows 3.  However you can cram a range of values of yours into

missing values <varlist> (<val1> thru <val2>) which only uses two and leaves
you a spare.

Hope this helps.  There are 400+ pages of syntax-based tutorials on my
website, all centred on data from questionnaire surveys, if that's of
interest.  I can possibly be of more help if you send me (off-list) an
extract of your *.sav file and your *.sps file as attachments (no charge).

John F Hall

[hidden email]
www.surveyresearch.weebly.com




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Jan
McClure
Sent: 03 November 2011 17:26
To: [hidden email]
Subject: Compute command and System User missing

Hi,
I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
trying to do something quite simple but am running into a problem with
system missing and user missing values.
Old variables are H_now (no missing values) and H_12mo which has 3 system
missing and 7 user missing (99).
I want to create a new variable (hosp). I set the new variable to 8 so that
any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
and are listed as an "8". These are the 3 sysmis and the 7 user missing
(99).
compute hosp=8.
if H_now EQ 1  hosp = 1.
if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
If H_12mo EQ 99  hosp = 98.
Variable labels hosp 'outcome'.
Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
Freq
/var=hosp
/missing=include.
Thanks for any help,
Jan

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

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

[Rich Ulrich]

>Date: Thu, 3 Nov 2011 10:46:06 -0700

>From: [hidden email]
>Subject: Re: Compute command and System User missing
>To: [hidden email]
>
> Hi,
>Sorry, I see that I wasn't specific.
>The results of the code I posted are 10 cases that fall into the category of 8.  I am using 8 only to see where my code does not work. In other words, if there are any '8' s in my new variable I did something wrong.
>I traced the source of my 10 cases that were incorrectly coded as an '8' to the variable h_12mo which has 7 cases that are user missing (coded as 99) and 3 cases that are system missing. 
>In the new variable I want the user missing to be a 98 and the system missing to still be system missing.
>Thanks,


The Keyword is $sysmis.  You still start with COMPUTE hosp=98.

The final IF becomes

IF   SYSMIS(H_12mo)  hosp= $sysmis.


--
Rich Ulrich

**************previous
On 11/3/2011 10:37 AM, Rich Ulrich wrote:

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 8.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich

> Date: Thu, 3 Nov 2011 12:26:12 -0400
> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1 hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99 hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.
[ ... ]

[Robin Darton]

Dear Jan,

 

The VALUE function can be used to ignore user missing-value definitions and return the actual value. In your example, you could use:

 

If VALUE(H_12mo) EQ 99 hosp = 98.

See p.121 in the PASW Statistics 18 Command Syntax Reference.

 

Hope this helps.

 

Robin Darton.

 

 

--

Robin A. Darton,

Senior Research Fellow,

Personal Social Services Research Unit,

University of Kent,

Canterbury,

Kent CT2 7NF,

U.K.

Tel: [phone]

Fax: [phone]

E-Mail: [hidden email]

Website: http://www.PSSRU.ac.uk/

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rich Ulrich
Sent: 03 November 2011 18:02
To: [hidden email]
Subject: Re: Compute command and System User missing

 

>Date: Thu, 3 Nov 2011 10:46:06 -0700

>From: [hidden email]
>Subject: Re: Compute command and System User missing
>To: [hidden email]
>
> Hi,
>Sorry, I see that I wasn't specific.
>The results of the code I posted are 10 cases that fall into the category of 8.  I am using 8 only to see where my code does not work. In other words, if there are any '8' s in my new variable I did something wrong.
>I traced the source of my 10 cases that were incorrectly coded as an '8' to the variable h_12mo which has 7 cases that are user missing (coded as 99) and 3 cases that are system missing. 
>In the new variable I want the user missing to be a 98 and the system missing to still be system missing.
>Thanks,


The Keyword is $sysmis.  You still start with COMPUTE hosp=98.

The final IF becomes

IF   SYSMIS(H_12mo)  hosp= $sysmis.


--
Rich Ulrich

**************previous
On 11/3/2011 10:37 AM, Rich Ulrich wrote:

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 8.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich

> Date: Thu, 3 Nov 2011 12:26:12 -0400

> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1 hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99 hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.

[ ... ]

 

[David Marso]

"
 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me. "

IF (VALUE( H_12mo) EQ  99) hosp=8.

;-)))

Rich Ulrich-2 wrote

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 8.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich
> Date: Thu, 3 Nov 2011 12:26:12 -0400
> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1  hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99  hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.
[ ... ]

[David Marso]

"
 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me. "

IF (VALUE( H_12mo) EQ  99) hosp=8.

;-)))

Rich Ulrich-2 wrote

You are not totally explicit about what the problem is.
Presumably, "eq 99" does not work because 99 is missing.

One obvious correction is to un-MISS  99 for H_12mo  before the
computation (using MISSING VALUES h_12mo ( ) ); and re-instate
it at the end.  The more direct correction uses the SYSMIS( )
function.

Apparently, you want the value to be 98 if H_12mo was 99,
and 8 if it was SYSMIS.  Thus -- make the initializing of Hosp
set it to 98; and change the test at the end to

IF   SYSMIS(H_12mo)  hosp= 98.

 - If there is a function that tests for MISSING, but not SYSMIS,
you could use that one in place of the present test on 99.
But I don't remember that name, and quick Googling doesn't
show it to me.

--
Rich Ulrich
> Date: Thu, 3 Nov 2011 12:26:12 -0400
> From: [hidden email]
> Subject: Compute command and System User missing
> To: [hidden email]
>
> Hi,
> I'm a doctoral student (epidemiology)and have not used SPSS in awhile. I'm
> trying to do something quite simple but am running into a problem with
> system missing and user missing values.
> Old variables are H_now (no missing values) and H_12mo which has 3 system
> missing and 7 user missing (99).
> I want to create a new variable (hosp). I set the new variable to 8 so that
> any cases that "fall thru" will be evident to me. Indeed, 10 cases fall thru
> and are listed as an "8". These are the 3 sysmis and the 7 user missing (99).
> compute hosp=8.
> if H_now EQ 1  hosp = 1.
> if (H_now EQ 0 and H_12mo EQ 1) hosp = 2.
> if (H_now EQ 0 and H_12mo EQ 0) hosp = 3.
> If H_12mo EQ 99  hosp = 98.
> Variable labels hosp 'outcome'.
> Value labels hosp 3 'no event' 2'12 mo event' 1'immediate event'.
> Freq
> /var=hosp
> /missing=include.
[ ... ]

[J McClure]

Thanks! Worked great and I learned from it.
Jan
(John, I used your tutorials a great deal in past and they are
wonderful. Thanks for the count reminder, I forgot about it. I have used
the list function often and I think it was you who suggested that)

On 11/3/2011 11:01 AM, Rich Ulrich wrote:
> IF   SYSMIS(H_12mo)  hosp= $sysmis.

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