Complex sample analysis with strata a... PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 pat berglund posted on Monday, April 14, 2008 - 12:48 pm
Hello, I have a sample with a strata (42 strata) and secu(sampling error computing unit) and want to do a simple linear regression.

My question is how to account for both variables in the Mplus code? I am using the following but want to use both stratification and cluster vars for the type=complex. Can you provide help on this issue? I don't have a 2 level model per se and only see those types of examples in the 9th chapter.


variable:
names are sampleid str secu finalp2wt hhinc sexf age ;
usevariables are str secu finalp2wt hhinc age;
cluster is str ;
weight=finalp2wt;

missing are . ;

analysis:
type=complex ;

model:
hhinc on age ;
 Linda K. Muthen posted on Monday, April 14, 2008 - 3:23 pm
You should specify the stratification variable using the STRATIFICATION option of the VARIABLE command and the CLUSTER variable using the CLUSTER option of the VARIABLE command.
 pat berglund posted on Monday, April 14, 2008 - 3:52 pm
Thanks for the help. I did try this type of setup but could not get it to run properly. Could it be the version of Mplus? I am about to install v5 and might that be the issue? Pat
 Linda K. Muthen posted on Monday, April 14, 2008 - 3:57 pm
This should work in Version 5. If not, please send your input, data, output, and license number to support@statmodel.com.
 pat berglund posted on Monday, April 14, 2008 - 4:15 pm
Thanks, I will install v5 tomorrow and hope to be on my way then. If not, I will send to your support email. Thanks very much! Pat Berglund
 Kathleen Watson posted on Thursday, April 29, 2010 - 11:11 am
Hello,
I have a few questions.
1. I am interested in performing a basic multiple regression model using complex survey data with strata, cluster, & weight. I would like in "state" so that strata & cluster would be nested in "state". Am I perceiving this correctly? If so, then it is actually a 3-level model which can't be modeled in MPlus? If so, any suggestions?

2. I am interested in running a path model in which I have strata, cluster, & weight. I also want to include households. So this again appears to be 3-levels? If so, can you treat the household members as repeated measures which Mplus may run?

Thank you,
Kathy
 Linda K. Muthen posted on Thursday, April 29, 2010 - 2:05 pm
Can you explain what in "state" means?
 Kathleen Watson posted on Friday, April 30, 2010 - 9:47 am
Sorry for not beeing clear. State as in the actual state a person lives in. So the PSUs & strata are nested within the state. I want to bring in state-level predictors
 Linda K. Muthen posted on Friday, April 30, 2010 - 3:07 pm
Are all states included in the study or a sample of states? And if a sample, how many are sampled.
 Kathleen Watson posted on Friday, April 30, 2010 - 3:49 pm
All states are included in the study-I can run a regression model in SUDAAN which will "NEST" state, strata, and PSU. If I want to fit a path model, I can't do that in SUDAAN.
From your manual, website, etc., I see that I can use the following...

weight = sample_wts;
strat = stratum;
cluster= psu;

Is it possible to add another level?
Thank you.
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: