Coding missing values
Jonathan Fry
David Waller wrote:
>
> I have a fairly large data set (64 cases, 488 variables) in text format
> with missing values coded as '999'. I would like to read this file into
> SPSS and assign '999' as a missing value across all variables. I can see
> doing one of two things in order to do this:
>
> 1. Use an ascii character in the text file which SPSS will recognize as a
> system missing value. Is there such a thing?
>
> 2. Re-define the default variable template to include '999' as a system
> missing value. I've tried this method, but haven't gotten it to work.
> Any ideas why?
>
> Any thoughts would be appreciated. Thanks,
>
> Dave
--
After you can see the data in the data editor, run the following command
from a syntax window.
MISSING VALUES ALL(999).
--------------------
Jonathan Fry
SPSS Inc.
j...@spss.com
David Waller
Richard F Ulrich
David Waller (dwa...@u.washington.edu) wrote:
: I have a fairly large data set (64 cases, 488 variables) in text format
: with missing values coded as '999'. I would like to read this file into
: SPSS and assign '999' as a missing value across all variables. I can see
: doing one of two things in order to do this:
: 1. Use an ascii character in the text file which SPSS will recognize as a
: system missing value. Is there such a thing?
-- To have SPSS read in data as system-missing, you want to have it
BLANK (or, less favorably, something illegal). If ALL the 9's or 999's
in your data are sys-miss, you might do that with an editor; but I can
say from experience that there is an element of risk. Example:
a "global-replace" might seem fine to change all "data" from 999 to
blanks; but it will ALSO change ID# 999 to blanks, if you are not
careful. OR: there is more serious trouble if you change <999> to
<>, no columns, instead of to < >, 3 columns, so that the spaces
to the REST of your data in the line are not read correctly.
: 2. Re-define the default variable template to include '999' as a system
: missing value. I've tried this method, but haven't gotten it to work.
: Any ideas why?
-- You can say MISSING VALUES, but 999 can only be User-missing,
because (I think) sys-miss is a special, non-used value. You can
to a RECODE to SYSMIS, and then save the file again....
Rich Ulrich, biostatistician wpi...@pitt.edu
Western Psychiatric Inst. and Clinic Univ. of Pittsburgh
Mark A. Serafin
David Waller (dwa...@u.washington.edu) wrote:
: I have a fairly large data set (64 cases, 488 variables) in text format
: with missing values coded as '999'. I would like to read this file into
: SPSS and assign '999' as a missing value across all variables. I can see
: doing one of two things in order to do this:
: 1. Use an ascii character in the text file which SPSS will recognize as a
: system missing value. Is there such a thing?
: 2. Re-define the default variable template to include '999' as a system
: missing value. I've tried this method, but haven't gotten it to work.
: Any ideas why?
I think the following command, executed after you have read the
dat in, should work (depending on which version you are using):
missing values all (999).
This will assign '999' as a missing value to all variables in the
data set.
--
Mark Serafin | "Reality must take precedence over public
Snohomish Health District | relations. Nature cannot be fooled."
I speak only for myself | - Richard Feynman
Frank Klatil
I created a sample file:
1.00 999 5.00
3.00 5.00 999
999 23.00 45.00
and in Excel I changed all "999" to letter "a". I also added variable
names to the first row. Then I named the range and using ODBC read this
excel file into SPSS 7.
VAR00001 VAR00002 VAR00003
1.00 a 5.00
3.00 5.00 a
a 23.00 45.00
The resulting SPSS file looked like this:
1.00 . 5.00
3.00 5.00 .
. 23.00 45.00
If you could live with 999 defined as user missing, you could also use
Variable template feature. (Look under Data->Templates...)
franK
David Waller wrote:
>
> I have a fairly large data set (64 cases, 488 variables) in text format
> with missing values coded as '999'. I would like to read this file into
> SPSS and assign '999' as a missing value across all variables. I can see
> doing one of two things in order to do this:
>
> 1. Use an ascii character in the text file which SPSS will recognize as a
> system missing value. Is there such a thing?
>
> 2. Re-define the default variable template to include '999' as a system
> missing value. I've tried this method, but haven't gotten it to work.
> Any ideas why?
>
> Any thoughts would be appreciated. Thanks,
>
> Dave
--
*****************fr...@welch.jhu.edu*****************
All human actions are equivalent ... and ... all are on principle doomed
...
Jean-Paul Sartre (Being and Nothingness, Conclusion,
sct. 2)
lillianzh...@gmail.com
> David Waller wrote:
> >
> > I have a fairly large data set (64 cases, 488 variables) in text format
> > with missing values coded as '999'. I would like to read this file into
> > SPSS and assign '999' as a missing value across all variables. I can see
> > doing one of two things in order to do this:
> >
> > 1. Use an ascii character in the text file which SPSS will recognize as a
> > system missing value. Is there such a thing?
> >
> > 2. Re-define the default variable template to include '999' as a system
> > missing value. I've tried this method, but haven't gotten it to work.
> > Any ideas why?
> >
> > Any thoughts would be appreciated. Thanks,
> >
> > Dave
>
> --
> from a syntax window.
>
> MISSING VALUES ALL(999).
> --------------------
> Jonathan Fry
> SPSS Inc.
> j...@spss.com
Thank you all!