Reading old SPSS-X files from tapes into SPSS for Windows

[Staffan Lindberg]

Dear list!

 

I have a some old SPSS-X files on tapes from the late 70’s and early 80’s. Originally they were run on an IBM (360, I think) mainframe before being stored on tape. The OS was MVS. I have them converted from EBCDIC to ASCII and tried with several different editors but they are completely unreadable.

 

If I remember correctly there were 2 ways of moving SPSS-X files on a mainframe to other platforms. The first was to save them in portable format and the second to use the command EXPORT coupled with TAPE or COMM. My problem now is that these possibilities both require that you read the tapes once more into a IBM mainframe with this old SPSS-X version and then resave them. Needless to say this is not possible any more. I have the files now both converted and unconverted on my hard drive.

 

My question is if there are other ways of getting these files into SPSS for windows? I have a vague recollection of a FILE HANDLE command, but cannot find any information on how to use it. I would be thankful for any input on this even a finger pointing in a possible direction. Hopefully there are still some data archaeologists out there?

 

best

 

Staffan Lindberg

Sweden

[Richard Ristow]

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

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

[Statistics by Greg Black]

Greg Black
Statistics by Greg Black
[hidden email]
512-491-0011



"They who can give up essential liberty to obtain a little temporary safety,
deserve neither liberty nor safety." - Benjamin Franklin
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Richard Ristow
Sent: Tuesday, May 11, 2010 7:06 AM
To: [hidden email]
Subject: Re: Reading old SPSS-X files from tapes into SPSS for Windows

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

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

[Staffan Lindberg]

Thank you Mark and Richard!

Sorry, I should have mentioned that they are stored as SPSS system files not
EBCDIC raw files. The coding schemata with variable names. positions, labels
etc have long since been lost. They now lie embedded in the system files. As
I understand you the future looks bleak. Richard mentions that there may be
some documentation on the detailed structure of an SPSS system file from
that era. Anyone knows something about this?

Richard also mentions the possibility of separating the data from the
dictionary by a "low level parse". I must confess to my ignorance as to what
a "low level parse" is. How do you do that? I know this looks bleak, but am
not ready to give up just yet.

best

Staffan Lindberg
Sweden

-----Ursprungligt meddelande-----
Från: Richard Ristow [mailto:[hidden email]]
Skickat: den 11 maj 2010 14:06
Till: Staffan Lindberg; [hidden email]
Ämne: Re: Reading old SPSS-X files from tapes into SPSS for Windows

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

=====================
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]

Are both the ebcdic and ascii files the result of export?  In other words are they portable files?



Art Kendall
Social Research Consultants

On 5/11/2010 6:36 AM, Staffan Lindberg wrote:

Dear list!

 

I have a some old SPSS-X files on tapes from the late 70’s and early 80’s. Originally they were run on an IBM (360, I think) mainframe before being stored on tape. The OS was MVS. I have them converted from EBCDIC to ASCII and tried with several different editors but they are completely unreadable.

 

If I remember correctly there were 2 ways of moving SPSS-X files on a mainframe to other platforms. The first was to save them in portable format and the second to use the command EXPORT coupled with TAPE or COMM. My problem now is that these possibilities both require that you read the tapes once more into a IBM mainframe with this old SPSS-X version and then resave them. Needless to say this is not possible any more. I have the files now both converted and unconverted on my hard drive.

 

My question is if there are other ways of getting these files into SPSS for windows? I have a vague recollection of a FILE HANDLE command, but cannot find any information on how to use it. I would be thankful for any input on this even a finger pointing in a possible direction. Hopefully there are still some data archaeologists out there?

 

best

 

Staffan Lindberg

Sweden

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

[Staffan Lindberg]

Thank you Art. The files are the result of saving them by simply transferring them onto tape as SPSS system files when the national central mainframe  in Stockholm ceased to exist. They are to my knowledge not portable files.  

 

best

 

Staffan Lindberg

Sweden

 

Från: Art Kendall [mailto:[hidden email]]
Skickat: den 11 maj 2010 16:48
Till: Staffan Lindberg
Kopia: [hidden email]
Ämne: Re: [SPSSX-L] Reading old SPSS-X files from tapes into SPSS for Windows

 

Are both the ebcdic and ascii files the result of export?  In other words are they portable files?



Art Kendall
Social Research Consultants

On 5/11/2010 6:36 AM, Staffan Lindberg wrote:

Dear list!

 

I have a some old SPSS-X files on tapes from the late 70’s and early 80’s. Originally they were run on an IBM (360, I think) mainframe before being stored on tape. The OS was MVS. I have them converted from EBCDIC to ASCII and tried with several different editors but they are completely unreadable.

 

If I remember correctly there were 2 ways of moving SPSS-X files on a mainframe to other platforms. The first was to save them in portable format and the second to use the command EXPORT coupled with TAPE or COMM. My problem now is that these possibilities both require that you read the tapes once more into a IBM mainframe with this old SPSS-X version and then resave them. Needless to say this is not possible any more. I have the files now both converted and unconverted on my hard drive.

 

My question is if there are other ways of getting these files into SPSS for windows? I have a vague recollection of a FILE HANDLE command, but cannot find any information on how to use it. I would be thankful for any input on this even a finger pointing in a possible direction. Hopefully there are still some data archaeologists out there?

 

best

 

Staffan Lindberg

Sweden

[Mark Casazza-2]

Hi Staffan,

Well, I just spent 15 minutes going through my 1990 SPSS 4.0 manual, and while it does have a section on punch cards it doesn't confirm that it can read older versions of mainframe system files (and your file is probably version 3 or earlier)--the GET command just states that it can read system files created with the SAVE and XSAVE commands.

How big is your file?  If you can get it to me I'm willing to upload it and see if our version 4.1 on OS/390 will read it.

Mark


On 2010-05-11 11:12, Staffan Lindberg wrote:

Thank you Art. The files are the result of saving them by simply transferring them onto tape as SPSS system files when the national central mainframe  in Stockholm ceased to exist. They are to my knowledge not portable files.  

 

best

 

Staffan Lindberg

Sweden

 

Från: Art Kendall [[hidden email]]
Skickat: den 11 maj 2010 16:48
Till: Staffan Lindberg
Kopia: [hidden email]
Ämne: Re: [SPSSX-L] Reading old SPSS-X files from tapes into SPSS for Windows

 

Are both the ebcdic and ascii files the result of export?  In other words are they portable files?



Art Kendall
Social Research Consultants

On 5/11/2010 6:36 AM, Staffan Lindberg wrote:

Dear list!

 

I have a some old SPSS-X files on tapes from the late 70’s and early 80’s. Originally they were run on an IBM (360, I think) mainframe before being stored on tape. The OS was MVS. I have them converted from EBCDIC to ASCII and tried with several different editors but they are completely unreadable.

 

If I remember correctly there were 2 ways of moving SPSS-X files on a mainframe to other platforms. The first was to save them in portable format and the second to use the command EXPORT coupled with TAPE or COMM. My problem now is that these possibilities both require that you read the tapes once more into a IBM mainframe with this old SPSS-X version and then resave them. Needless to say this is not possible any more. I have the files now both converted and unconverted on my hard drive.

 

My question is if there are other ways of getting these files into SPSS for windows? I have a vague recollection of a FILE HANDLE command, but cannot find any information on how to use it. I would be thankful for any input on this even a finger pointing in a possible direction. Hopefully there are still some data archaeologists out there?

 

best

 

Staffan Lindberg

Sweden

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

[Staffan Lindberg]

Thank you ever so much Mark for your effort. . I’m very grateful.  I have been in contact with  the person who in the National Archive who converted the files and told me that the files were stored  on IBM OS/ MVS release 2.1  on the mainframe.  The most important file is about 150 variables and  about 200 cases of narcotic addicts in Sweden in the 80’s, which is very important in studying the later mortality of this cohort. If you think this is meaningful I’ll of course send you the file. Ever so grateful!

 

best

 

Staffan Lindberg

Sweden

 

Från: Mark Casazza [mailto:[hidden email]]
Skickat: den 11 maj 2010 17:40
Till: Staffan Lindberg
Kopia: [hidden email]
Ämne: Re: SV: [SPSSX-L] Reading old SPSS-X files from tapes into SPSS for Windows

 

Hi Staffan,

Well, I just spent 15 minutes going through my 1990 SPSS 4.0 manual, and while it does have a section on punch cards it doesn't confirm that it can read older versions of mainframe system files (and your file is probably version 3 or earlier)--the GET command just states that it can read system files created with the SAVE and XSAVE commands.

How big is your file?  If you can get it to me I'm willing to upload it and see if our version 4.1 on OS/390 will read it.

Mark

“it’s probably worth noting that we threw off the yoke of the [British] crown for violations of human freedom and dignity that were a hell of a lot less severe than what we put up with today.”
- Radley Balko, 2010-04-19, The Agitator

Mark Casazza
Director of Academic Information
The City University of New York
555 West 57th Street, Suite 1240
New York, NY 10019

Phone: [phone]
Fax: [phone]
email: [hidden email]

On 2010-05-11 11:12, Staffan Lindberg wrote:

Thank you Art. The files are the result of saving them by simply transferring them onto tape as SPSS system files when the national central mainframe  in Stockholm ceased to exist. They are to my knowledge not portable files.  

 

best

 

Staffan Lindberg

Sweden

 

Från: Art Kendall [[hidden email]]
Skickat: den 11 maj 2010 16:48
Till: Staffan Lindberg
Kopia: [hidden email]
Ämne: Re: [SPSSX-L] Reading old SPSS-X files from tapes into SPSS for Windows

 

Are both the ebcdic and ascii files the result of export?  In other words are they portable files?



Art Kendall
Social Research Consultants

On 5/11/2010 6:36 AM, Staffan Lindberg wrote:

Dear list!

 

I have a some old SPSS-X files on tapes from the late 70’s and early 80’s. Originally they were run on an IBM (360, I think) mainframe before being stored on tape. The OS was MVS. I have them converted from EBCDIC to ASCII and tried with several different editors but they are completely unreadable.

 

If I remember correctly there were 2 ways of moving SPSS-X files on a mainframe to other platforms. The first was to save them in portable format and the second to use the command EXPORT coupled with TAPE or COMM. My problem now is that these possibilities both require that you read the tapes once more into a IBM mainframe with this old SPSS-X version and then resave them. Needless to say this is not possible any more. I have the files now both converted and unconverted on my hard drive.

 

My question is if there are other ways of getting these files into SPSS for windows? I have a vague recollection of a FILE HANDLE command, but cannot find any information on how to use it. I would be thankful for any input on this even a finger pointing in a possible direction. Hopefully there are still some data archaeologists out there?

 

best

 

Staffan Lindberg

Sweden

[John F Hall]

I had a similar problem when I retired from the Polytechnic of North London in 1992.  The computer service dumped everything of mine from the Vax cluster on to 7 magnetic tapes which I then left at the Data Archive at Essex.  They no longer had a Vax mainframe and two years later they returned the box of tapes to me with 2 CDs with everything on. 

 

As far as I recall, none of the *.sys files copied across, but some of the *.exp and *.imp did.  All raw data and SPSS setup files were there, but had extensions added to denote edition, ie test.sps would be saved as test.sps_1.  Once I worked out what was happening, by manually deleting all the _n extensions, the files became readable by SPSS for Windows.  All the SPSS *.sav files and the (very few) *.por files worked, but I had to reconstruct all the rest of my stuff from scratch, including modifying 1972 syntax!.  My documentation, including full user manuals for major surveys, was invaluable, plus I found a few errors. 

 

Everything initially deposited at Essex is still available and working: the problems are with  the smaller studies which Essex turned away after they moved towards being a data archive rather than the survey archive they were originally set up as.  They now get all government data, including surveys: whilst they still get all the big non-government surveys, they're not interested in the smaller ad-hoc surveys, however professionally conducted and regardless of their importance for theory or policy.  Some of us take great care with our data, but PNL managed to lose everything of mine from before 1986, except material already deposited at Essex.

 

There seemed to be nowhere able to read the (?9-track or 7-track?) magnetic tapes, but I eventually found an outfit in Aylesbury (UK) which could do some conversions, but only at a price beyond my means. 

 

I've checked the CDs again to see what exactly worked and what didn't.  I found a file initially exported by SPSS-X and displayed as sr501exm.por but which Windows lists as an EXP file so there must be a hidden extension on it: I probably created it with save out 'sr501exm.exp' or similar and then tried to change it.  Using right click and open with I can open it with SPSS 15 so I presume 18 as well.

 

If you can burn the files from tape to CD, does this help?

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Tuesday, May 11, 2010 3:58 PM

Subject: SV: Reading old SPSS-X files from tapes into SPSS for Windows

Thank you Mark and Richard!

Sorry, I should have mentioned that they are stored as SPSS system files not
EBCDIC raw files. The coding schemata with variable names. positions, labels
etc have long since been lost. They now lie embedded in the system files. As
I understand you the future looks bleak. Richard mentions that there may be
some documentation on the detailed structure of an SPSS system file from
that era. Anyone knows something about this?

Richard also mentions the possibility of separating the data from the
dictionary by a "low level parse". I must confess to my ignorance as to what
a "low level parse" is. How do you do that? I know this looks bleak, but am
not ready to give up just yet.

best

Staffan Lindberg
Sweden

-----Ursprungligt meddelande-----
Från: Richard Ristow [mailto:[hidden email]]
Skickat: den 11 maj 2010 14:06
Till: Staffan Lindberg; [hidden email]
Ämne: Re: Reading old SPSS-X files from tapes into SPSS for Windows

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

=====================
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]

Staffan

 

Here's the outfit I found.

 

http://www.altirium.com/

 

http://www.altirium.co.uk/Altirium/about-altirium.html

 

 

Mark Sear wrote:

 

"Sounds like a 9-track tape to me. Hopefully not a 7-track as they are not pleasant to deal with.

Restoring the files as they are stored should not be an issue so long as the condition of the tapes is good, what we will have to think about is if they need converting in any way.  The first thing we would do is to read the raw data from the tapes and then secure it by backing it up to other tapes, we’d make 2 copies. Then we could look at what work was required."

 

Being retired on a fractional pension I couldn't afford to get the conversion done, so there are still one or two important SPSS *.sys files on the tapes (eg from pilot work and main surveys of European Values) which I can't access.

 

John

 

PS  Sorry about the typeface: the computer took over.

 

 

.----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Wednesday, May 12, 2010 6:59 AM

Subject: SV: SV: Reading old SPSS-X files from tapes into SPSS for Windows

Thanks Jon!

 

You’ve given me some clues and I will test them out.

 

best

 

Staffan

 

Från: John F Hall [mailto:[hidden email]]
Skickat: den 11 maj 2010 19:46
Till: Staffan Lindberg; [hidden email]
Ämne: Re: SV: Reading old SPSS-X files from tapes into SPSS for Windows

 

I had a similar problem when I retired from the Polytechnic of North London in 1992.  The computer service dumped everything of mine from the Vax cluster on to 7 magnetic tapes which I then left at the Data Archive at Essex.  They no longer had a Vax mainframe and two years later they returned the box of tapes to me with 2 CDs with everything on. 

 

As far as I recall, none of the *.sys files copied across, but some of the *.exp and *.imp did.  All raw data and SPSS setup files were there, but had extensions added to denote edition, ie test.sps would be saved as test.sps_1.  Once I worked out what was happening, by manually deleting all the _n extensions, the files became readable by SPSS for Windows.  All the SPSS *.sav files and the (very few) *.por files worked, but I had to reconstruct all the rest of my stuff from scratch, including modifying 1972 syntax!.  My documentation, including full user manuals for major surveys, was invaluable, plus I found a few errors. 

 

Everything initially deposited at Essex is still available and working: the problems are with  the smaller studies which Essex turned away after they moved towards being a data archive rather than the survey archive they were originally set up as.  They now get all government data, including surveys: whilst they still get all the big non-government surveys, they're not interested in the smaller ad-hoc surveys, however professionally conducted and regardless of their importance for theory or policy.  Some of us take great care with our data, but PNL managed to lose everything of mine from before 1986, except material already deposited at Essex.

 

There seemed to be nowhere able to read the (?9-track or 7-track?) magnetic tapes, but I eventually found an outfit in Aylesbury (UK) which could do some conversions, but only at a price beyond my means. 

 

I've checked the CDs again to see what exactly worked and what didn't.  I found a file initially exported by SPSS-X and displayed as sr501exm.por but which Windows lists as an EXP file so there must be a hidden extension on it: I probably created it with save out 'sr501exm.exp' or similar and then tried to change it.  Using right click and open with I can open it with SPSS 15 so I presume 18 as well.

 

If you can burn the files from tape to CD, does this help?

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Tuesday, May 11, 2010 3:58 PM

Subject: SV: Reading old SPSS-X files from tapes into SPSS for Windows

 

Thank you Mark and Richard!

Sorry, I should have mentioned that they are stored as SPSS system files not
EBCDIC raw files. The coding schemata with variable names. positions, labels
etc have long since been lost. They now lie embedded in the system files. As
I understand you the future looks bleak. Richard mentions that there may be
some documentation on the detailed structure of an SPSS system file from
that era. Anyone knows something about this?

Richard also mentions the possibility of separating the data from the
dictionary by a "low level parse". I must confess to my ignorance as to what
a "low level parse" is. How do you do that? I know this looks bleak, but am
not ready to give up just yet.

best

Staffan Lindberg
Sweden

-----Ursprungligt meddelande-----
Från: Richard Ristow [mailto:[hidden email]]
Skickat: den 11 maj 2010 14:06
Till: Staffan Lindberg; [hidden email]
Ämne: Re: Reading old SPSS-X files from tapes into SPSS for Windows

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

=====================
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]

Just succeeded in reading really old SPSS files saved on DEC/Vax as *.exp (ie exported system files)  I had amountain of setup files in *.sps_n form where n = edition number of file.  Being lazy, not particularly risk averse (copies on CD anyway) and from long distant memory, I went into the directory (very laborious as you can't use CTRL+C and CTRL+V to paste the address: it had to be written in by hand over three lines) and used DOS command rename  *.sps_*  *.sps which to my great relief caused all the Windows icons to turn into syntax file icons.  Magic, and saved me a lot of time.  Encouraged by this I used rename  *.exp_*  *  *.por double clicked on one and it opened immediately (but with 15 not 18).  Nice to know some things still work!

 

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Wednesday, May 12, 2010 10:52 AM

Subject: Re: SV: Reading old SPSS-X files from tapes into SPSS for Windows

Altirium email address no longer exists:  as of 16 March it's now [hidden email] . 

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Wednesday, May 12, 2010 6:59 AM

Subject: SV: SV: Reading old SPSS-X files from tapes into SPSS for Windows

Thanks Jon!

 

You’ve given me some clues and I will test them out.

 

best

 

Staffan

 

Från: John F Hall [mailto:[hidden email]]
Skickat: den 11 maj 2010 19:46
Till: Staffan Lindberg; [hidden email]
Ämne: Re: SV: Reading old SPSS-X files from tapes into SPSS for Windows

 

I had a similar problem when I retired from the Polytechnic of North London in 1992.  The computer service dumped everything of mine from the Vax cluster on to 7 magnetic tapes which I then left at the Data Archive at Essex.  They no longer had a Vax mainframe and two years later they returned the box of tapes to me with 2 CDs with everything on. 

 

As far as I recall, none of the *.sys files copied across, but some of the *.exp and *.imp did.  All raw data and SPSS setup files were there, but had extensions added to denote edition, ie test.sps would be saved as test.sps_1.  Once I worked out what was happening, by manually deleting all the _n extensions, the files became readable by SPSS for Windows.  All the SPSS *.sav files and the (very few) *.por files worked, but I had to reconstruct all the rest of my stuff from scratch, including modifying 1972 syntax!.  My documentation, including full user manuals for major surveys, was invaluable, plus I found a few errors. 

 

Everything initially deposited at Essex is still available and working: the problems are with  the smaller studies which Essex turned away after they moved towards being a data archive rather than the survey archive they were originally set up as.  They now get all government data, including surveys: whilst they still get all the big non-government surveys, they're not interested in the smaller ad-hoc surveys, however professionally conducted and regardless of their importance for theory or policy.  Some of us take great care with our data, but PNL managed to lose everything of mine from before 1986, except material already deposited at Essex.

 

There seemed to be nowhere able to read the (?9-track or 7-track?) magnetic tapes, but I eventually found an outfit in Aylesbury (UK) which could do some conversions, but only at a price beyond my means. 

 

I've checked the CDs again to see what exactly worked and what didn't.  I found a file initially exported by SPSS-X and displayed as sr501exm.por but which Windows lists as an EXP file so there must be a hidden extension on it: I probably created it with save out 'sr501exm.exp' or similar and then tried to change it.  Using right click and open with I can open it with SPSS 15 so I presume 18 as well.

 

If you can burn the files from tape to CD, does this help?

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Tuesday, May 11, 2010 3:58 PM

Subject: SV: Reading old SPSS-X files from tapes into SPSS for Windows

 

Thank you Mark and Richard!

Sorry, I should have mentioned that they are stored as SPSS system files not
EBCDIC raw files. The coding schemata with variable names. positions, labels
etc have long since been lost. They now lie embedded in the system files. As
I understand you the future looks bleak. Richard mentions that there may be
some documentation on the detailed structure of an SPSS system file from
that era. Anyone knows something about this?

Richard also mentions the possibility of separating the data from the
dictionary by a "low level parse". I must confess to my ignorance as to what
a "low level parse" is. How do you do that? I know this looks bleak, but am
not ready to give up just yet.

best

Staffan Lindberg
Sweden

-----Ursprungligt meddelande-----
Från: Richard Ristow [mailto:[hidden email]]
Skickat: den 11 maj 2010 14:06
Till: Staffan Lindberg; [hidden email]
Ämne: Re: Reading old SPSS-X files from tapes into SPSS for Windows

At 06:36 AM 5/11/2010, Staffan Lindberg wrote:

>I have a some old SPSS-X files on tapes from the late 70's and early
>80's. Originally they were run on an IBM (360, I think) mainframe
>before being stored on tape. The OS was MVS. I have them converted
>from EBCDIC to ASCII and tried with several different editors but
>they are completely unreadable.

Whatever else will work, that probably won't. Remember, SPSS stores
numbers in the floating-point format native to the machine; on IBM
mainframes of that age, that's 32-bit hexfloat (as it's known -- the
360/370 specific format). Bytewise conversion to ASCII will scramble
those, probably beyond recovery.

>I have the files now both converted and unconverted on my hard drive.

Thank goodness for the unconverted, where you have a chance.

>My question is if there are other ways of getting these files into
>SPSS for windows? I have a vague recollection of a FILE HANDLE
>command, but cannot find any information on how to use it. I would
>be thankful for any input on this even a finger pointing in a
>possible direction. Hopefully there are still some data
>archaeologists out there?

Is there public documentation on the detailed structure of an SPSS
system file of that era? If so, at worst one could do a low-level
parse, extracting the data dictionary and data separately, noting
which were string variables and should be converted to ASCII strings,
which were numeric variable and should be converted to (probably)
text-represented numbers.

Sounds like fun, in an adventurous sort of way.

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