[EpiData-list] Export to SPSS messed up by missing data

epidata-list at lists.umanitoba.ca epidata-list at lists.umanitoba.ca
Wed Jan 16 14:50:29 CST 2008


Hello folks.  I just noticed that when I export to SPSS, things can end up
either missing or in the wrong columns when there are empty fields in the
.rec file.  Here's a simple example I generated to test it.  I created a
.qes file that looks like this:

{string1}  String variable 1  __________
{num1}     Numeric variable 1 ##
{datevar1} Date variable 1   <yyyy/mm/dd>

{string2}  String variable 2  __________
{num2}     Numeric variable 2 ##
{datevar2} Date variable 2   <yyyy/mm/dd>

{string3}  String variable 3  __________
{num3}     Numeric variable 3 ##
{datevar3} Date variable 3   <yyyy/mm/dd>

-------------------------------

Then I entered a couple rows of data, with complete data on row 1, but
incomplete data on row 2 (num1 and datevar 1 were left blank).
When I export to Excel, I get the following (which is correct):

STRING1    NUM1   DATEVAR1 STRING2    NUM2   DATEVAR2 STRING3    NUM3
DATEVAR3
record 1.1   1  06/01/2008 record 1.2   2           . rec 1.3      3
06/01/2008
rec 2.1      .           . rec 2.2     22  01/01/2008 rec 2.3     23
02/01/2008

But if I export to SPSS, I get this, which is not correct:

STRING1    NUM1   DATEVAR1 STRING2    NUM2   DATEVAR2 STRING3    NUM3
DATEVAR3
record 1.1   1  01/06/2008 record 1.2   2  01/16/2008 rec 1.3      3
01/06/2008
rec 2.1      .           . 2201/01/20   8           . 2301/02/20
8           .

My first thought was that SPSS 16 is lousing things up.  (I've been forced
to upgrade from v15 due to an institutional decision, and I've noticed
several other problems with v16).  But the same problem occurred when I
tried on another machine running SPSS 11.

Has anyone else run into this?  I didn't find any mention of it in the
archives.

What I'm doing for now to circumvent the problem is this:

1) Exporting to SPSS to get syntax for all the variable and value labels
2) Exporting to Excel to get the data into a format I can read with SPSS
3) Creating syntax to import the Excel file, then adding the variable and
value label syntax generated by exporting to SPSS.

It's not terribly more complicated than a straight export to SPSS, but it
would be nice if I could do it in one step.

By the way, I'm running EpiData 3.1 on a ThinkPad (XP Professional, SP2, 4
gigs of memory IIRC).

Thanks,
Bruce

-- 
Bruce Weaver
bweaver at lakeheadu.ca
www.angelfire.com/wv/bwhomedir
"When all else fails, RTFM."


More information about the EpiData-list mailing list