Scratch variable causing error in SPSS 21.0.0.2 (64-bit)

[Bruce Weaver]

I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7 Professional (SP1).  The following lines are from a syntax file to test hypotheses about and generate confidence intervals for single regression coefficients:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.

This syntax file worked perfectly under whatever version of SPSS I had when I wrote it (probably v19 or 20).  But now, I get the following error messages in the output window:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg  0 - abs(t).
 
>Error # 4382 in column 16.  Text: 0
>An equals sign was not found when expected after a target variable in a
>COMPUTE command.
>Execution of this command stops.

COMPUTE p = cdf.t(#tneg,df)*2.
 
>Error # 4285 in column 19.  Text: #tneg
>Incorrect variable name: either the name is more than 64 characters, or it is
>not defined by a previous command.
>Execution of this command stops.

Notice that in the syntax that is echoed to the output viewer, the 'equals' sign has vanished from the line that computes #tneg.

COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've checked)
COMPUTE #tneg  0 - abs(t). <-- in the output viewer

If I change #tneg to tneg in both lines where it appears, there are no errors.  

AND--this one is really bizarre--if I add a second 'equals' sign to that line (as shown below), again, it runs with no errors!

COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!

This makes no sense.  Any thoughts on what is going on here?  

If anyone wants to try the complete syntax file, it's file number 2 here:

https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch

Thanks,
Bruce

[Art Kendall]

there is no assignment operator in that line

On 4/29/2014 11:21 AM, Bruce Weaver [via SPSSX Discussion] wrote:

COMPUTE #tneg  0 - abs(t).
 

[Jon K Peck]

Bruce,

Seems like your computer is having a bad day.  I cannot produce this misbehavior in V21.0.0.1 or V22.0.0.1, and I cannot find anything that sounds like this in our bug database.  My first thought was that you had some nonprinting characters in the syntax file, but dumping your original file in hex, it looks perfectly normal.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Bruce Weaver <[hidden email]>
To:        [hidden email],
Date:        04/29/2014 09:22 AM
Subject:        [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2 (64-bit)
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7
Professional (SP1).  The following lines are from a syntax file to test
hypotheses about and generate confidence intervals for single regression
coefficients:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.

This syntax file worked perfectly under whatever version of SPSS I had when
I wrote it (probably v19 or 20).  But now, I get the following error
messages in the output window:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg  0 - abs(t).

>Error # 4382 in column 16.  Text: 0
>An equals sign was not found when expected after a target variable in a
>COMPUTE command.
>Execution of this command stops.

COMPUTE p = cdf.t(#tneg,df)*2.

>Error # 4285 in column 19.  Text: #tneg
>Incorrect variable name: either the name is more than 64 characters, or it
is
>not defined by a previous command.
>Execution of this command stops.

Notice that in the syntax that is echoed to the output viewer, the 'equals'
sign has vanished from the line that computes #tneg.

COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've
checked)
COMPUTE #tneg  0 - abs(t). <-- in the output viewer

If I change #tneg to tneg in both lines where it appears, there are no
errors.

AND--this one is really bizarre--if I add a second 'equals' sign to that
line (as shown below), again, it runs with no errors!

COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!

This makes no sense.  Any thoughts on what is going on here?

If anyone wants to try the complete syntax file, it's file number 2 here:

https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch

Thanks,
Bruce





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Bruce Weaver]

Hi Art.  There IS an assignment operator in the syntax file, but it is not showing up in the syntax that is echoed in the output viewer.  I.e., this line appears in the syntax file:

COMPUTE #tneg = 0 - abs(t).

But THIS line is echoed in the output viewer (and followed by the error message):

COMPUTE #tneg  0 - abs(t).

If I don't use a scratch variable (i.e., change both occurrences of #tneg to tneg), the 'equals' sign is echoed in the output, and there is no error.

Cheers,
Bruce

Art Kendall wrote

there is no assignment
        operator in that line
       
     
      Art Kendall
Social Research Consultants
      On 4/29/2014 11:21 AM, Bruce Weaver [via SPSSX Discussion] wrote:
   
    COMPUTE #tneg  0 - abs(t).
       

[David Marso]

No can repro ver 22.0.0.1 doze 64.

[Art Kendall]

Bruce,
did you try the magic-fix-many-things-in-windows trick?

restart
On 4/29/2014 2:27 PM, David Marso [via SPSSX Discussion] wrote:

No can repro ver 22.0.0.2 doze 64.

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706p5725715.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

[Alan Howard]

Then Bruce isn't the only one having a bad computer day.

In 22.0.0.0 running on 64-bit Win 7 Pro (SP1), I can reproduce the error.  But in 21.0.0.0 on the same machine, the syntax runs without error.

In 22.0.0.0 running on Mac OS X 10.9.2, I can reproduce the error.   But in 21.0.0.0 on the same machine, the syntax runs without error.

I am about to apply fix packs to both 21 and 22 and rerun.

Alan....

On 4/29/2014 1:57 PM, Jon K Peck wrote:

Bruce,

Seems like your computer is having a bad day.  I cannot produce this misbehavior in V21.0.0.1 or V22.0.0.1, and I cannot find anything that sounds like this in our bug database.  My first thought was that you had some nonprinting characters in the syntax file, but dumping your original file in hex, it looks perfectly normal.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Bruce Weaver [hidden email]
To:        [hidden email],
Date:        04/29/2014 09:22 AM
Subject:        [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2 (64-bit)
Sent by:        "SPSSX(r) Discussion" [hidden email]

---

I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7
Professional (SP1).  The following lines are from a syntax file to test
hypotheses about and generate confidence intervals for single regression
coefficients:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.

This syntax file worked perfectly under whatever version of SPSS I had when
I wrote it (probably v19 or 20).  But now, I get the following error
messages in the output window:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg  0 - abs(t).

>Error # 4382 in column 16.  Text: 0
>An equals sign was not found when expected after a target variable in a
>COMPUTE command.
>Execution of this command stops.

COMPUTE p = cdf.t(#tneg,df)*2.

>Error # 4285 in column 19.  Text: #tneg
>Incorrect variable name: either the name is more than 64 characters, or it
is
>not defined by a previous command.
>Execution of this command stops.

Notice that in the syntax that is echoed to the output viewer, the 'equals'
sign has vanished from the line that computes #tneg.

COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've
checked)
COMPUTE #tneg  0 - abs(t). <-- in the output viewer

If I change #tneg to tneg in both lines where it appears, there are no
errors.

AND--this one is really bizarre--if I add a second 'equals' sign to that
line (as shown below), again, it runs with no errors!

COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!

This makes no sense.  Any thoughts on what is going on here?

If anyone wants to try the complete syntax file, it's file number 2 here:

https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch

Thanks,
Bruce





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Alan Howard]

In both Win 7 and OS X 10.9.2:

Scratch variable causes error in 22.0.0.0, but not in 22.0.0.1.
Scratch variable does not cause error in 21.0.0.0, but does in 21.0.0.2.

On 4/29/2014 3:00 PM, Alan Howard wrote:

Then Bruce isn't the only one having a bad computer day.

In 22.0.0.0 running on 64-bit Win 7 Pro (SP1), I can reproduce the error.  But in 21.0.0.0 on the same machine, the syntax runs without error.

In 22.0.0.0 running on Mac OS X 10.9.2, I can reproduce the error.   But in 21.0.0.0 on the same machine, the syntax runs without error.

I am about to apply fix packs to both 21 and 22 and rerun.

Alan....

On 4/29/2014 1:57 PM, Jon K Peck wrote:

Bruce,

Seems like your computer is having a bad day.  I cannot produce this misbehavior in V21.0.0.1 or V22.0.0.1, and I cannot find anything that sounds like this in our bug database.  My first thought was that you had some nonprinting characters in the syntax file, but dumping your original file in hex, it looks perfectly normal.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Bruce Weaver [hidden email]
To:        [hidden email],
Date:        04/29/2014 09:22 AM
Subject:        [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2 (64-bit)
Sent by:        "SPSSX(r) Discussion" [hidden email]

---

I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7
Professional (SP1).  The following lines are from a syntax file to test
hypotheses about and generate confidence intervals for single regression
coefficients:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.

This syntax file worked perfectly under whatever version of SPSS I had when
I wrote it (probably v19 or 20).  But now, I get the following error
messages in the output window:

COMPUTE t = (b - bstar) / se.
COMPUTE #tneg  0 - abs(t).

>Error # 4382 in column 16.  Text: 0
>An equals sign was not found when expected after a target variable in a
>COMPUTE command.
>Execution of this command stops.

COMPUTE p = cdf.t(#tneg,df)*2.

>Error # 4285 in column 19.  Text: #tneg
>Incorrect variable name: either the name is more than 64 characters, or it
is
>not defined by a previous command.
>Execution of this command stops.

Notice that in the syntax that is echoed to the output viewer, the 'equals'
sign has vanished from the line that computes #tneg.

COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've
checked)
COMPUTE #tneg  0 - abs(t). <-- in the output viewer

If I change #tneg to tneg in both lines where it appears, there are no
errors.

AND--this one is really bizarre--if I add a second 'equals' sign to that
line (as shown below), again, it runs with no errors!

COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!

This makes no sense.  Any thoughts on what is going on here?

If anyone wants to try the complete syntax file, it's file number 2 here:

https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch

Thanks,
Bruce





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Bruce Weaver]

Very interesting, Alan!  Thanks to all who tried to reproduce the error.

Jon, what do you make of this?  

Alan Howard wrote

In both Win 7 and OS X 10.9.2:

Scratch variable causes error in 22.0.0.0, but not in 22.0.0.1.
Scratch variable does not cause error in 21.0.0.0, but does in 21.0.0.2.



On 4/29/2014 3:00 PM, Alan Howard wrote:
> Then Bruce isn't the only one having a bad computer day.
>
> In 22.0.0.0 running on 64-bit Win 7 Pro (SP1), I can reproduce the
> error.  But in 21.0.0.0 on the same machine, the syntax runs without
> error.
>
> In 22.0.0.0 running on Mac OS X 10.9.2, I can reproduce the error.
> But in 21.0.0.0 on the same machine, the syntax runs without error.
>
> I am about to apply fix packs to both 21 and 22 and rerun.
>
> Alan....
>
>
>
> --
> Alan Howard
> Statistical Software Support & Consulting Services
> 115 Bailey/Howe Library
> University of Vermont
> Burlington, VT  05405
> 802-656-2009
>
> On 4/29/2014 1:57 PM, Jon K Peck wrote:
>> Bruce,
>>
>> Seems like your computer is having a bad day.  I cannot produce this
>> misbehavior in V21.0.0.1 or V22.0.0.1, and I cannot find anything
>> that sounds like this in our bug database.  My first thought was that
>> you had some nonprinting characters in the syntax file, but dumping
>> your original file in hex, it looks perfectly normal.
>>
>>
>> Jon Peck (no "h") aka Kim
>> Senior Software Engineer, IBM
>> [hidden email]
>> phone: [phone]
>>
>>
>>
>>
>> From: Bruce Weaver <[hidden email]>
>> To: [hidden email],
>> Date: 04/29/2014 09:22 AM
>> Subject: [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2
>> (64-bit)
>> Sent by: "SPSSX(r) Discussion" <[hidden email]>
>> ------------------------------------------------------------------------
>>
>>
>>
>> I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7
>> Professional (SP1).  The following lines are from a syntax file to test
>> hypotheses about and generate confidence intervals for single regression
>> coefficients:
>>
>> COMPUTE t = (b - bstar) / se.
>> COMPUTE #tneg = 0 - abs(t).
>> COMPUTE p = cdf.t(#tneg,df)*2.
>>
>> This syntax file worked perfectly under whatever version of SPSS I
>> had when
>> I wrote it (probably v19 or 20).  But now, I get the following error
>> messages in the output window:
>>
>> COMPUTE t = (b - bstar) / se.
>> COMPUTE #tneg  0 - abs(t).
>>
>> >Error # 4382 in column 16.  Text: 0
>> >An equals sign was not found when expected after a target variable in a
>> >COMPUTE command.
>> >Execution of this command stops.
>>
>> COMPUTE p = cdf.t(#tneg,df)*2.
>>
>> >Error # 4285 in column 19.  Text: #tneg
>> >Incorrect variable name: either the name is more than 64 characters,
>> or it
>> is
>> >not defined by a previous command.
>> >Execution of this command stops.
>>
>> Notice that in the syntax that is echoed to the output viewer, the
>> 'equals'
>> sign has vanished from the line that computes #tneg.
>>
>> COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've
>> checked)
>> COMPUTE #tneg  0 - abs(t). <-- in the output viewer
>>
>> If I change #tneg to tneg in both lines where it appears, there are no
>> errors.
>>
>> AND--this one is really bizarre--if I add a second 'equals' sign to that
>> line (as shown below), again, it runs with no errors!
>>
>> COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!
>>
>> This makes no sense.  Any thoughts on what is going on here?
>>
>> If anyone wants to try the complete syntax file, it's file number 2 here:
>>
>> https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch
>>
>> Thanks,
>> Bruce
>>
>>
>>
>>
>>
>> -----
>> --
>> Bruce Weaver
>> [hidden email]
>> http://sites.google.com/a/lakeheadu.ca/bweaver/
>>
>> "When all else fails, RTFM."
>>
>> NOTE: My Hotmail account is not monitored regularly.
>> To send me an e-mail, please use the address shown above.
>>
>> --
>> View this message in context:
>> http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706.html
>> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>
>> =====================
>> 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
>>
>>
>

[Jon K Peck]

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Bruce Weaver <[hidden email]>
To:        [hidden email],
Date:        04/29/2014 02:07 PM
Subject:        Re: [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2              (64-bit)
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Very interesting, Alan!  Thanks to all who tried to reproduce the error.

Jon, what do you make of this?

>>>Two computers having a bad day??  Use of the notorious Canadian script??  Vermont is pretty close to Canada, after all.  I can't find any trace of such misbehavior.  I have a question in to an engineer to see if this rings any bells.  Out of curiosity, what happens if you change the order of the terms so that 0 is not first on the right hand side?




Alan Howard wrote
> In both Win 7 and OS X 10.9.2:
>
> Scratch variable causes error in 22.0.0.0, but not in 22.0.0.1.
> Scratch variable does not cause error in 21.0.0.0, but does in 21.0.0.2.
>
>
>
> On 4/29/2014 3:00 PM, Alan Howard wrote:
>> Then Bruce isn't the only one having a bad computer day.
>>
>> In 22.0.0.0 running on 64-bit Win 7 Pro (SP1), I can reproduce the
>> error.  But in 21.0.0.0 on the same machine, the syntax runs without
>> error.
>>
>> In 22.0.0.0 running on Mac OS X 10.9.2, I can reproduce the error.
>> But in 21.0.0.0 on the same machine, the syntax runs without error.
>>
>> I am about to apply fix packs to both 21 and 22 and rerun.
>>
>> Alan....
>>
>>
>>
>> --
>> Alan Howard
>> Statistical Software Support & Consulting Services
>> 115 Bailey/Howe Library
>> University of Vermont
>> Burlington, VT  05405
>> 802-656-2009
>>
>> On 4/29/2014 1:57 PM, Jon K Peck wrote:
>>> Bruce,
>>>
>>> Seems like your computer is having a bad day.  I cannot produce this
>>> misbehavior in V21.0.0.1 or V22.0.0.1, and I cannot find anything
>>> that sounds like this in our bug database.  My first thought was that
>>> you had some nonprinting characters in the syntax file, but dumping
>>> your original file in hex, it looks perfectly normal.
>>>
>>>
>>> Jon Peck (no "h") aka Kim
>>> Senior Software Engineer, IBM
>>>

> peck@.ibm

>>> phone: [phone]
>>>
>>>
>>>
>>>
>>> From: Bruce Weaver &lt;

> bruce.weaver@

> &gt;
>>> To:

> SPSSX-L@.uga

> ,
>>> Date: 04/29/2014 09:22 AM
>>> Subject: [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2
>>> (64-bit)
>>> Sent by: "SPSSX(r) Discussion" &lt;

> SPSSX-L@.uga

> &gt;
>>> ------------------------------------------------------------------------
>>>
>>>
>>>
>>> I am running release 21.0.0.2 (64-bit) for Windoze under Windoze 7
>>> Professional (SP1).  The following lines are from a syntax file to test
>>> hypotheses about and generate confidence intervals for single regression
>>> coefficients:
>>>
>>> COMPUTE t = (b - bstar) / se.
>>> COMPUTE #tneg = 0 - abs(t).
>>> COMPUTE p = cdf.t(#tneg,df)*2.
>>>
>>> This syntax file worked perfectly under whatever version of SPSS I
>>> had when
>>> I wrote it (probably v19 or 20).  But now, I get the following error
>>> messages in the output window:
>>>
>>> COMPUTE t = (b - bstar) / se.
>>> COMPUTE #tneg  0 - abs(t).
>>>
>>> >Error # 4382 in column 16.  Text: 0
>>> >An equals sign was not found when expected after a target variable in a
>>> >COMPUTE command.
>>> >Execution of this command stops.
>>>
>>> COMPUTE p = cdf.t(#tneg,df)*2.
>>>
>>> >Error # 4285 in column 19.  Text: #tneg
>>> >Incorrect variable name: either the name is more than 64 characters,
>>> or it
>>> is
>>> >not defined by a previous command.
>>> >Execution of this command stops.
>>>
>>> Notice that in the syntax that is echoed to the output viewer, the
>>> 'equals'
>>> sign has vanished from the line that computes #tneg.
>>>
>>> COMPUTE #tneg = 0 - abs(t). <-- in the syntax editor (it is there--I've
>>> checked)
>>> COMPUTE #tneg  0 - abs(t). <-- in the output viewer
>>>
>>> If I change #tneg to tneg in both lines where it appears, there are no
>>> errors.
>>>
>>> AND--this one is really bizarre--if I add a second 'equals' sign to that
>>> line (as shown below), again, it runs with no errors!
>>>
>>> COMPUTE #tneg == 0 - abs(t).  <-- runs with no errors!
>>>
>>> This makes no sense.  Any thoughts on what is going on here?
>>>
>>> If anyone wants to try the complete syntax file, it's file number 2
>>> here:
>>>
>>> https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/weaver_wuensch
>>>
>>> Thanks,
>>> Bruce
>>>
>>>
>>>
>>>
>>>
>>> -----
>>> --
>>> Bruce Weaver
>>>

> bweaver@

>>> http://sites.google.com/a/lakeheadu.ca/bweaver/
>>>
>>> "When all else fails, RTFM."
>>>
>>> NOTE: My Hotmail account is not monitored regularly.
>>> To send me an e-mail, please use the address shown above.
>>>
>>> --
>>> View this message in context:
>>> http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706.html
>>> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>>
>>> =====================
>>> To manage your subscription to SPSSX-L, send a message to
>>>

> LISTSERV@.UGA

>  (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
>>>
>>>
>>





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706p5725721.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Art Kendall]

The syntax below runs okay on windows 8.1 64 bit SPSS 21.0.0.1


data list list /id (f2) b (f7.3) bstar(f7.3) se (f7.3)df (f4).
begin data
01 4 3 .2 10
02 5 4 .5 20
03 4.4 4 .2 200
end data.
COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.
formats p (f7.5).
list.

On 4/29/2014 9:28 PM, Jon K Peck [via SPSSX Discussion] wrote:

COMPUTE t = (b - bstar) / se.
>>> COMPUTE #tneg = 0 - abs(t).
>>> COMPUTE p = cdf.t(#tneg,df)*2.

[Alan Howard]

So it seems like a fix pack in v21 may have introduced this error, which carried over to v22, then was magically fixed again with another fix pack. 

21.0.0.0 : no error
21.0.0.1 : no error
21.0.0.2 : error
22.0.0.0 : error
22.0.0.1 : no error

On 4/30/2014 8:07 AM, Art Kendall wrote:

The syntax below runs okay on windows 8.1 64 bit SPSS 21.0.0.1


data list list /id (f2) b (f7.3) bstar(f7.3) se (f7.3)df (f4).
begin data
01 4 3 .2 10
02 5 4 .5 20
03 4.4 4 .2 200
end data.
COMPUTE t = (b - bstar) / se.
COMPUTE #tneg = 0 - abs(t).
COMPUTE p = cdf.t(#tneg,df)*2.
formats p (f7.5).
list.

On 4/29/2014 9:28 PM, Jon K Peck [via SPSSX Discussion] wrote:

COMPUTE t = (b - bstar) / se.
>>> COMPUTE #tneg = 0 - abs(t).
>>> COMPUTE p = cdf.t(#tneg,df)*2.

---

View this message in context: Re: Scratch variable causing error in SPSS 21.0.0.2 (64-bit)
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

[Bruce Weaver]

This keeps getting more mysterious.  Art's syntax runs error-free for me too.  That made me wonder if the PRESERVE-SET DECIMAL-RESTORE lines in my syntax file are causing a problem.  I.e., my syntax file looks like this:

PRESERVE.
SET DECIMAL = dot.

* DATA LIST & syntax for computations inserted here .

RESTORE.

I commented out the PRESERVE-SET-RESTORE lines in my syntax file, and it ran without errors.  

The next step was to try Art's syntax inside of that same PRESERVE-RESTORE business.  It ran without errors too.  

But if I go back to the original version of my syntax file, the #tneg scratch variable still causes an error.  (Even after restarting the computer, which I did just now.)  

Turning to a question Jon raised in another post, if I change the first line in which #tneg appears from

   COMPUTE #tneg = 0 - abs(t).

to

   COMPUTE #tneg = abs(t)*-1.

the syntax runs error-free.  Care to explain what you were thinking there, Jon?  

Art Kendall wrote

The syntax below runs
        okay on windows 8.1 64 bit SPSS 21.0.0.1
       
       
        data list list /id (f2) b (f7.3) bstar(f7.3) se (f7.3)df (f4).
        begin data
        01 4 3 .2 10
        02 5 4 .5 20
        03 4.4 4 .2 200
        end data.
        COMPUTE t = (b - bstar) / se.
        COMPUTE #tneg = 0 - abs(t).
        COMPUTE p = cdf.t(#tneg,df)*2.
        formats p (f7.5).
        list.
     
      Art Kendall
Social Research Consultants
      On 4/29/2014 9:28 PM, Jon K Peck [via SPSSX Discussion] wrote:
   
    COMPUTE t = (b - bstar) / se.
          >>> COMPUTE #tneg = 0 - abs(t).
          >>> COMPUTE p = cdf.t(#tneg,df)*2.

[Jon K Peck]

We located a resolved bug in the bug database that might cover this situation, although it was about a vanishing = in a RECODE command.  It was resolved in fixpacks for V21 and V22.  I have no information on the exact circumstances when the equals sign would get discarded.  Try this syntax to see if you get the same failure.

data list free /dependent independent constant split.
begin data.
3 2 7 3
5 1 7 3
8 3 7 3
end data.
recode all (0=sysmis).

Bruce, My thought about reordering the terms in the expression came from the idea that numeric tokens are handled differently from alphabetic ones.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        Bruce Weaver <[hidden email]>
To:        [hidden email],
Date:        04/30/2014 06:54 AM
Subject:        Re: [SPSSX-L] Scratch variable causing error in SPSS 21.0.0.2              (64-bit)
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

This keeps getting more mysterious.  Art's syntax runs error-free for me too.
That made me wonder if the PRESERVE-SET DECIMAL-RESTORE lines in my syntax
file are causing a problem.  I.e., my syntax file looks like this:

PRESERVE.
SET DECIMAL = dot.

* DATA LIST & syntax for computations inserted here .

RESTORE.

I commented out the PRESERVE-SET-RESTORE lines in my syntax file, and it ran
without errors.

The next step was to try Art's syntax inside of that same PRESERVE-RESTORE
business.  It ran without errors too.

But if I go back to the original version of my syntax file, the #tneg
scratch variable still causes an error.  (Even after restarting the
computer, which I did just now.)

Turning to a question Jon raised in another post, if I change the first line
in which #tneg appears from

  COMPUTE #tneg = 0 - abs(t).

to

  COMPUTE #tneg = abs(t)*-1.

the syntax runs error-free.  Care to explain what you were thinking there,
Jon?





Art Kendall wrote
> The syntax below runs
>         okay on windows 8.1 64 bit SPSS 21.0.0.1
>
>
>         data list list /id (f2) b (f7.3) bstar(f7.3) se (f7.3)df (f4).
>         begin data
>         01 4 3 .2 10
>         02 5 4 .5 20
>         03 4.4 4 .2 200
>         end data.
>         COMPUTE t = (b - bstar) / se.
>         COMPUTE #tneg = 0 - abs(t).
>         COMPUTE p = cdf.t(#tneg,df)*2.
>         formats p (f7.5).
>         list.
>
>       Art Kendall
> Social Research Consultants
>       On 4/29/2014 9:28 PM, Jon K Peck [via SPSSX Discussion] wrote:
>
>     COMPUTE t = (b - bstar) / se.
>           &gt;&gt;&gt; COMPUTE #tneg = 0 - abs(t).
>           &gt;&gt;&gt; COMPUTE p = cdf.t(#tneg,df)*2.





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706p5725727.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Bruce Weaver]

Thanks Jon.  That syntax runs error-free on my system.  

Jon K Peck wrote

We located a resolved bug in the bug database that might cover this
situation, although it was about a vanishing = in a RECODE command.  It
was resolved in fixpacks for V21 and V22.  I have no information on the
exact circumstances when the equals sign would get discarded.  Try this
syntax to see if you get the same failure.

data list free /dependent independent constant split.
begin data.
3 2 7 3
5 1 7 3
8 3 7 3
end data.
recode all (0=sysmis).

Bruce, My thought about reordering the terms in the expression came from
the idea that numeric tokens are handled differently from alphabetic ones.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:   Bruce Weaver <[hidden email]>
To:     [hidden email],
Date:   04/30/2014 06:54 AM
Subject:        Re: [SPSSX-L] Scratch variable causing error in SPSS
21.0.0.2              (64-bit)
Sent by:        "SPSSX(r) Discussion" <[hidden email]>



This keeps getting more mysterious.  Art's syntax runs error-free for me
too.
That made me wonder if the PRESERVE-SET DECIMAL-RESTORE lines in my syntax
file are causing a problem.  I.e., my syntax file looks like this:

PRESERVE.
SET DECIMAL = dot.

* DATA LIST & syntax for computations inserted here .

RESTORE.

I commented out the PRESERVE-SET-RESTORE lines in my syntax file, and it
ran
without errors.

The next step was to try Art's syntax inside of that same PRESERVE-RESTORE
business.  It ran without errors too.

But if I go back to the original version of my syntax file, the #tneg
scratch variable still causes an error.  (Even after restarting the
computer, which I did just now.)

Turning to a question Jon raised in another post, if I change the first
line
in which #tneg appears from

   COMPUTE #tneg = 0 - abs(t).

to

   COMPUTE #tneg = abs(t)*-1.

the syntax runs error-free.  Care to explain what you were thinking there,
Jon?





Art Kendall wrote
> The syntax below runs
>         okay on windows 8.1 64 bit SPSS 21.0.0.1
>
>
>         data list list /id (f2) b (f7.3) bstar(f7.3) se (f7.3)df (f4).
>         begin data
>         01 4 3 .2 10
>         02 5 4 .5 20
>         03 4.4 4 .2 200
>         end data.
>         COMPUTE t = (b - bstar) / se.
>         COMPUTE #tneg = 0 - abs(t).
>         COMPUTE p = cdf.t(#tneg,df)*2.
>         formats p (f7.5).
>         list.
>
>       Art Kendall
> Social Research Consultants
>       On 4/29/2014 9:28 PM, Jon K Peck [via SPSSX Discussion] wrote:
>
>     COMPUTE t = (b - bstar) / se.
>           >>> COMPUTE #tneg = 0 - abs(t).
>           >>> COMPUTE p = cdf.t(#tneg,df)*2.





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Scratch-variable-causing-error-in-SPSS-21-0-0-2-64-bit-tp5725706p5725727.html

Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Art Kendall]

The snippet does contain something that I hope users would avoid.  Assigning a sysmis value by user command.

I modified the syntax to make it have something to recode, but still had no problem
Windows 8.1 64 bit SPSS 21.0.0.1

data list free /dependent independent constant split.
begin data.
3 2 7 3
5 1 7 3
8 3 7 3
1 1 1 0
0 1 1 1

end data.
recode all (0=sysmis).
list.

On 4/30/2014 3:20 PM, Jon K Peck [via SPSSX Discussion] wrote:

data list free /dependent independent constant split.
begin data.
3 2 7 3
5 1 7 3
8 3 7 3
end data.
recode all (0=sysmis).