Listing of cases used

[Matthew Reeder]

Hey all,

  Quick question regarding the FACTOR proc (though this could apply to others - RELIABILITY, CORR, etc). I'm working with a sizeable dataset where some groups of cases systematically miss data (such that, for instance, we admistered eight or nine forms of a survey to several groups; some items overlap across forms, others do not).

  Obviously, when running the FACTOR proc, it excludes cases on a listwise basis. Given the size of the dataset, is there a clean-cut way to ascertain which cases are included in the analysis? Better yet, is there a way to ascertain which groups based on a given variable are being included (saying that we have the var FORM taking the values of 01, 02, 03, ..., 08, 09)? Assuming that 987 cases were used for the analysis, I imagine a table included within the factor analysis output that breaks these cases down by FORM. I suspect I'm asking too much on this, but was simply curious, for curiosity's sake.

  Thanks in advance for any suggestions.

  - Matt

 __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.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

[Richard Ristow]

At 07:14 PM 10/26/2007, Matthew Reeder wrote:

>   Quick question regarding the FACTOR proc, [etc.] It excludes cases
> on a listwise basis. Is there a clean-cut way to ascertain which
> cases are included in the analysis?

FACTOR syntax begins with

.  FACTOR VARIABLES=<varlist>.

The test for no missing values in a variable list is

NMISS(<varlist>) EQ 0.

So,

.  SELECT IF NMISS(<varlist>) EQ 0.

gives you the cases used for the analysis. (This is a good candidate to
run with a TEMPORARY command, to get your statistics on the cases used
without actually discarding the others.)

>Better yet, is there a way to ascertain which groups based on a given
>variable are being included (saying that we have the var FORM taking
>the values of 01, 02, 03, ..., 08, 09)?

Is this what you want?

.  TEMPORARY.
.  SELECT IF NMISS(<varlist>) EQ 0.
.  FREQUENCIES FORM.

That gives you the number of cases included in the analysis for each
value of FORM.

-Cheers and good wishes,
  Richard

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

[Matthew Reeder]

Hi Richard,

  Works perfectly. Thanks for the pointer!


  Regards,

  - Matt

Richard Ristow <[hidden email]> wrote:
  At 07:14 PM 10/26/2007, Matthew Reeder wrote:

> Quick question regarding the FACTOR proc, [etc.] It excludes cases
> on a listwise basis. Is there a clean-cut way to ascertain which
> cases are included in the analysis?

FACTOR syntax begins with

. FACTOR VARIABLES=.

The test for no missing values in a variable list is

NMISS() EQ 0.

So,

. SELECT IF NMISS() EQ 0.

gives you the cases used for the analysis. (This is a good candidate to
run with a TEMPORARY command, to get your statistics on the cases used
without actually discarding the others.)

>Better yet, is there a way to ascertain which groups based on a given
>variable are being included (saying that we have the var FORM taking
>the values of 01, 02, 03, ..., 08, 09)?

Is this what you want?

. TEMPORARY.
. SELECT IF NMISS() EQ 0.
. FREQUENCIES FORM.

That gives you the number of cases included in the analysis for each
value of FORM.

-Cheers and good wishes,
Richard



 __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.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