Tanja,
I assume you are at at university with access to the PsychInfo database.
Please do a search on the name 'John Graham' at Penn State University. John
has written a number of articles on missing data imputation and comparisons
of multiple imputation and full information methods. Others have written
excellent articles/websites on missing data; I am more familiar John's work
and his name comes first to my mind.
Gene Maguin
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
SPSSX-L@...] On Behalf Of Dr.
Tanja Legenbauer
Sent: Monday, November 02, 2009 9:00 AM
To:
SPSSX-L@...
Subject: number of missing values being allowed to be replaced
Dear all,
I have to deal with a drop out rate of 35 to 38% over a period of four
years follow up with three assessment time points (baseline, one year,
two years and four years follow-up). I know that it is recommended to
reassess about 75% of participants, but we just have not been able to.
Now I want to perform an intention to treat analyses with the missing
data being replaced by a multiple imputation procedure. As far as I know
this is the best way to deal with missing values. However, is there a
limit as to how much missing values are allowed to be replaced? I would
be very happy if there is anyone who knows some articles that I can
refer to.
Thanks,
Tanja
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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
=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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
