Date:         Fri, 11 Jan 2008 16:43:24 +0100
Reply-To:     Andre Wielki <wielki@INED.FR>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Andre Wielki <wielki@INED.FR>
Subject:      Re: SPSS data into SAS
Comments: To: "Coleman, Mark" <Mark.Coleman@LIBERTYMUTUAL.COM>
In-Reply-To:  <6D81DF01A2FD1D43BFD0D52D4BF7510AF079AD@LM-EXMSG-05.lm.lmig.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Many times indeed

if you have sas access to pc file format rights:

  i can make a reference to my recent messages on sas-l



235765  07/05/14        10:31   57      import spss table + values: quality
Here my opinion about the multiples ways to import spss(13 or more) tables
into sas 9.1.3

i was in the state of a traditional misfunctionning of Dbmscopy (v7)
with a table where the length (variables name +label) were exceeding the memory allowed for
working.  (I don't know if Dbmscopy-Sas v8 has solved this problem)

The proc import of SAS in 9.1.3 sp4 was creating a sas table with complaint about
the lack of a format catalog : but in fack it was creating a formats.sas7bcat in the work
and was putting a SYSTEMATIC format name in the sas table.
(Note be carefull when using proc import in those cases because if you are
importing more than one table sas is adding the new formats in the existing catalog).

With nofmterr   and fmtsearch,  you can work with  but in my two tables
the creation of inexistent formats (in the catalog) was a trouble.
With confrontation of cntlout and contents results tableds i was able to detect
the formats name to suppress in the sas tables.
Another effect is that the sas table are a lot of bigger  (6m to 75m)
as the spss users are creating numeric variable (of short length 2, 3, 4 ...)
which are becoming in sas each 8 long. Compression can avoid this a little.

But when i was going curious to SPSS (v13) i discover that it was doing a better work
producing a sas table (with the problem of 8 bytes i agree) but with also producing
a program (if asked) with the proc format code and the proc datasets code to integrate
those. But the sas table was obliged to be a v6 one as the total length in memory
of variables name + label was exceeding like in dbmscopy the memory allowed.

Finally, with Stattransfer v8 or +,  you have  the best work
a sas table of v9 level + the code for the formats (pro format and proc datasets)
and in addition the reducing of the numeric length to 3,4 bytes if possible after
internal analysis.

and the previous one upon v9.1.3 sp3 or more.


234080  07/04/26        10:44   53      Re: Can I convert SPSS .sav file(NOT .por file) to SAS dataset.

Hello import users,

i have checkup carefully what happen upon a spss table WITH value labels when imported directly in sas through
the last possibility
133 filename sps "\\Pc_00761\SPSS\employes.sav";
134 proc import dbms=sav datafile=sps out=test replace; run;
NOTE: .TEST was successfully created. NOTE: PROCEDURE IMPORT used (Total process time): real time 0.06 seconds
cpu time 0.03 seconds

in fact this new possibility is of quality

BECAUSE -you got in the sas table the personnal format names -you got ALSO into the work.formats catalog the
formats entries

Be carefull : in case of multiple import on many spss tables all the formats entries are going into the SAME
temporary catalog.

HTH
Andre

Coleman, Mark a écrit :
> Hi,
>
> I presume this problem has surfaced more than once on SAS-L. How can one
> import a native (binary) SPSS file into SAS? Or more precisely, is it
> possible to import a binary SPSS file into SAS without having SPSS?
>
> Thanks,
>
> Mark
>
>


--
    André WIELKI
    INED (Institut National d'Etudes Démographiques)
    Service Informatique
    133 Boulevard Davout      75980 Paris Cedex 20
    mél : wielki@ined.fr      tél :  33 (0) 1 56 06 21 54