Date:         Mon, 22 Feb 1999 18:49:47 -0300
Reply-To:     "Hector E. Maletta" <hmaletta@OVERNET.COM.AR>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         "Hector E. Maletta" <hmaletta@OVERNET.COM.AR>
Subject:      Re: filter out multiple observations
Comments: To: Kerri-Ann Kiniorski <kerriann@ERISS.COM>
Content-Type: text/plain; charset=us-ascii

You don't need any filtering. Rather, you may use AGGREGATE, if you're
sure the employer-specific information is the same for all ocuppations
reported by the same employer. The syntax may be something like the
following:
 
 
AGGREGATE /outfile=* /BREAK=employer_id
 /employer1 employer2 employer3 = first (employer1 employer2 employer3).
 
where employer1 employer2 employer3 is the list of employer-specific
variables, and EMPLOYER_ID is whatever variable serves to identify
employers.
The new file will have a unique record per employer. You may also have
in that file some aggregate figures from the occupations each employer
reported (e.g. average salary paid, number of occupations reported,
etc), as you can easily learn from the syntax of the AGGREGATE command.
 
Hector Maletta
Universidad del Salvador
Buenos Aires, Argentina
 
> Kerri-Ann Kiniorski wrote:
>
> Hello all,
>
> I've got an employer data set full of occupational-specific and
> employer-specific info. Some employers gave info. on more than one
> occupation so there are multiple observations for those employers.
> Right now, i'm interested in analyzing some of the employer-specific
> information which will be the same for each of the employer's multiple
> observations. How do I filter my data set so that I'm only seeing one
> observation for each employer, regardless of how many observations
> they have?
>
> Thank you kindly!
> Kerri-Ann Kiniorski