----------------------------------------------------------------------
Date: Mon, 16 Oct 2006 08:41:30 +0200
From: "Maarten Buis" <[email protected]>
Subject: st: RE: replace missing values (fwd)
Cathy:
Mean imputation is not a good way of dealing with missing
data. Think of a scatter plot with two variables and
where your "imputed values" end up in that scatter plot.
They will look really different from the rest of the data
and will affect the estimated regression line. For an
accessible account see Paul Allison's 2002 book
"Missing Data" from Sage (a "little green sage book")
A good alternative is Patrick Royston' s -ice-
(see: -findit ice-)
HTH,
Maarten
- -----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z434
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
- -----------------------------------------
- -----Original Message-----
From: [email protected] [mailto:[email protected]]On Behalf Of Cathy L. Antonakos
Sent: maandag 16 oktober 2006 5:51
To: [email protected]
Subject: st: replace missing values (fwd)
Sending again with a formatting correction, to hopefully display the data correctly this time.
CA
- --------------------
I have a dataset with hospital and ICU data. I'd like to replace missing data
for one ICU with the average value of the other 3 icu's at that hospital. The
dataset looks like this:
Unit_ID Total_Score
11 .
12 90
13 60
14 27
I can get the average for total score for the 3 units by using "if" and
specifying the unit id's. But how can I then replace the missing unit's score
with the average from the 3 other units? I've tried several ways and searched
online but can't find an answer to this specific problem. Thanks for any help
you can provide.
Cathy
