Date:         Tue, 4 Nov 2003 17:15:25 +0100
Reply-To:     Staffan Lindberg <Staffan.Lindberg@fhi.se>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Staffan Lindberg <Staffan.Lindberg@fhi.se>
Subject:      SV: Averaging across columns with missing values
Comments: To: chris fisher <cfisher@jjay.cuny.edu>
Content-Type: text/plain; charset="iso-8859-1"

Try:

compute newvar=sum.1(var1,var2,..............varn).
exe.

It demands only 1 valid variable (not containing missing)for the whole
array. If you demand 2, just write sum.2

best

Staffan Lindberg
National Institute of Public Health
Sweden

-----Ursprungligt meddelande-----
Från: chris fisher [mailto:cfisher@jjay.cuny.edu]
Skickat: den 4 november 2003 17:12
Till: SPSSX-L@LISTSERV.UGA.EDU
Ämne: Averaging across columns with missing values


What would the syntax be for averaging across ten columns of data, where
some of the columns had system missing data???

I tried a simple (day1+day2+day3+day4....)/10

But it only works for the cases that have values in all 10 columns. I can't
just do a fill or set missing equal to zero because all subjects were not
expected to have 10 data points, it is ok if someone only has day1 and day2,
but I'd still like to know the average.

Thanks!