Multiple Imputation model and questions

[jchaudhu]

Hello - a couple questions about MI

1) When creating a database to impute missing values,  is it redundant to include subscales as well as summary score ( of the same measure) , if these are just auxiliary variables to inform the imputation?  I am worried about multicollinearity, but am not sure if that is really an issue with imputation.

2) What is the best way to determine how many datasets to impute?  (People in our lab have used Rubin's table of relative efficiencies, others have used the general rule to impute as many datasets as there is percentage of missing values (25% missing values, impute 25 data sets).

[Maguin, Eugene]

1) Remember that MI involves a regression equation estimation step. So,
multicollinearity can occur. Whether MI programs/routines will test for and
issue warning messages, I don't know.

2) I haven't ever seen the general rule you refer to. Since you are at a
university do a search on John Graham or, maybe, Todd Little or Scott
Maxwell. One of them, I think it was John did a study of number of
imputations and power (and the answer was not 5) that appeared in either
Psych Methods or SEM.

Gene Maguin


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
jchaudhu
Sent: Thursday, October 13, 2011 2:10 PM
To: [hidden email]
Subject: Multiple Imputation model and questions

Hello - a couple questions about MI

1) When creating a database to impute missing values,  is it redundant to
include subscales as well as summary score ( of the same measure) , if these
are just auxiliary variables to inform the imputation?  I am worried about
multicollinearity, but am not sure if that is really an issue with
imputation.

2) What is the best way to determine how many datasets to impute?  (People
in our lab have used Rubin's table of relative efficiencies, others have
used the general rule to impute as many datasets as there is percentage of
missing values (25% missing values, impute 25 data sets).

-----
Jana H. Chaudhuri, Ph.D.
Lecturer, Child Development
Data Manager, Massachusetts Healthy Families Evaluation
Tufts University
177 College Avenue
Medford, MA 02155
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Multiple-Imputation-model-and-
questions-tp4900070p4900070.html
Sent from the SPSSX Discussion mailing list archive at Nabble.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

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

[jchaudhu]

Thank you for your response.  Another person I asked has referred me to this article regarding number of datasets:
Graham, Olchowski, and Gilreath 2007

On Thu, Oct 13, 2011 at 2:40 PM, Gene Maguin [via SPSSX Discussion] <[hidden email]> wrote:

1) Remember that MI involves a regression equation estimation step. So,
multicollinearity can occur. Whether MI programs/routines will test for and
issue warning messages, I don't know.

2) I haven't ever seen the general rule you refer to. Since you are at a
university do a search on John Graham or, maybe, Todd Little or Scott
Maxwell. One of them, I think it was John did a study of number of
imputations and power (and the answer was not 5) that appeared in either
Psych Methods or SEM.

Gene Maguin


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
jchaudhu
Sent: Thursday, October 13, 2011 2:10 PM
To: [hidden email]
Subject: Multiple Imputation model and questions

Hello - a couple questions about MI

1) When creating a database to impute missing values,  is it redundant to
include subscales as well as summary score ( of the same measure) , if these
are just auxiliary variables to inform the imputation?  I am worried about
multicollinearity, but am not sure if that is really an issue with
imputation.

2) What is the best way to determine how many datasets to impute?  (People
in our lab have used Rubin's table of relative efficiencies, others have
used the general rule to impute as many datasets as there is percentage of
missing values (25% missing values, impute 25 data sets).

-----
Jana H. Chaudhuri, Ph.D.
Lecturer, Child Development
Data Manager, Massachusetts Healthy Families Evaluation
Tufts University
177 College Avenue
Medford, MA 02155
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Multiple-Imputation-model-and-
questions-tp4900070p4900070.html
Sent from the SPSSX Discussion mailing list archive at Nabble.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

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

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Multiple-Imputation-model-and-questions-tp4900070p4900191.html

To unsubscribe from Multiple Imputation model and questions, click here.