Missing data PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Eunice Campiran posted on Thursday, April 16, 2015 - 8:46 pm
Hi!
I'm trying to conduct multiple imputation in a multilevel dataset and I have some questions:
1) I have missing values in the dependent variable and in the covariates. Mplus can make imputations in such case?

2) I am using negative binomial regression in my multilevel model, Mplus can make imputations?

3) Can I include interactions in my multilevel model and Mplus still can do imputations?

Many thanks!
 Linda K. Muthen posted on Friday, April 17, 2015 - 6:28 am
1. Yes.
2. All variables are treated as continuous unless they are on the CATEGORICAL list. No other types of variables can be imputed.
3. I would include the two variables and create the interaction after the imputation.
 Hsien-Yuan Hsu posted on Sunday, April 26, 2015 - 8:37 pm
Hi, Dr. Muthen,
I have two questions.
1) Following Ex. 9.1, I specify a two-level random intercept model with a categorical DV, which has five categories (1-5) with missing value (.). The output showed that I have 7 categories:
Category 1 0.098 1573.000
Category 2 0.327 5235.000
Category 3 0.265 4240.000
Category 4 0.163 2619.000
Category 5 0.101 1621.000
Category 6 0.046 730.000
Category 7 0.000 1.000

I have no idea why Mplus presents 7 categories. Any thought?

2) What is the syntax to save the predicted Y values for both levels?

Thank you so much!
 Bengt O. Muthen posted on Monday, April 27, 2015 - 11:12 am
Please send input, output, and data to support along with your license number.
 Eunice Campiran posted on Wednesday, April 29, 2015 - 5:27 pm
Hi, Dr. Muthen,
I have missing values in the two level variable OCC.
In the imputed data from Mplus, I got variations for OCC within clusters, which should not be.
For instance, for cluster progidx=5,
progidx OCC
5 45.7
5 65.8
5 43.3
5 12.6
5 46.4

It should be constant within this cluster.
Thanks,

my input file:

DATA:
file = Missing.csv;
VARIABLE:


names=q98days progidx q25_lic occ female homeless
usevariables=q98days q25_lic occ female homeless;
count is q98days (nb);
categorical is q25_lic female homeless;
cluster = progidx;
within = female homeless;
between = q25_lic occ;

missing = all (-99);

ANALYSIS:
TYPE = TWOLEVEL RANDOM;
bseed = 2828;
bconvergence = 0.01;
PROCESSORS = 2;
INTEGRATION =MONTECARLO (10);
MODEL:
%within%
q98days on female homeless;
%between%
q98days on occ q25_lic;


DATA IMPUTATION:
impute=occ;
ndatasets = 20;
save =Missing*.dat;
thin = 500;
 Bengt O. Muthen posted on Wednesday, April 29, 2015 - 6:02 pm
We need the data and your input and output to help. Send to support along with your license number.
 Eunice Campiran posted on Friday, May 08, 2015 - 1:10 pm
I am using negative binomial regression in my multilevel model, this variable does not have missing values.
MPlus can make imputations in other continuous variables?

I got the following error:
*** ERROR in DATA IMPUTATION command
The DATA IMPUTATION command is not available for analysis with count, continuous-time
survival, censored or nominal variables.
 Linda K. Muthen posted on Friday, May 08, 2015 - 1:17 pm
You must be saying you want to impute for the negative binomial. Example 11.5 goes over the various options that can be used with DATA IMPUTATION.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: