More thann two levels? PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 John posted on Saturday, March 19, 2005 - 2:06 pm
Hello,
for cross-sectional data, is it possible to handle more than two levels with Mplus? I have cross-sectional data and want to model individuals (level 1) within counties (level 2) within states (level 3)?

Thanks,
John
 Linda K. Muthen posted on Saturday, March 19, 2005 - 2:15 pm
No, this is not currently possible in Mplus.
 Veronique Van ACker posted on Friday, November 24, 2006 - 1:46 pm
Has it now become possible by TYPE = COMPLEX TWOLEVEL ? If yes, is there any example available on this ?

Thanks in advance,
Veronique
 Bengt O. Muthen posted on Friday, November 24, 2006 - 3:32 pm
Type = Complex Twolevel is a rather recent additions. I believe it was introduced in the May release of Mplus Versionoption and your study sounds like it will benefit from it.
 Yasamin Kusunoki posted on Tuesday, March 03, 2009 - 5:30 pm
So, just to confirm the above conversation, a three-level model can be run by combining complex and twolevel in the Type statement but can only be done in later versions of Mplus, correct? I have version 3. Thank you.
 Linda K. Muthen posted on Tuesday, March 03, 2009 - 7:32 pm
Three cross-sectional levels can be handled by COMPLEX TWOLEVEL which became available in Version 4. Three-level models can also be handled with TWOLEVEL for a repeated measures model.
 Sam Vuchinich posted on Monday, September 14, 2009 - 8:51 pm
Are there any plans for Mplus to have the capability of estimating models with a four-level data structure, either with level-1 being time or not (i.e., four cross-sectional levels)?

Sam Vuchinich
 Bengt O. Muthen posted on Tuesday, September 15, 2009 - 3:12 pm
The next version will accomodate modeling with 3-level data, so that one can have a cross-sectional model of say students within classrooms within schools. This means 4-level modeling if the first level is time given that Mplus uses a multivariate, wide format growth approach.

Mplus also does not add a level for multiple indicators.

Note also the Type = Complex Twolevel option where you can handle 3-level data with the nesting at the top level dealt with by correcting the SEs.
 Sam Vuchinich posted on Tuesday, September 15, 2009 - 4:59 pm
Thank you for this useful information. I realize that it is impossible to know exactly when future software upgrades will be available. But could you give a "ballpark" estimate? For example, would it be within the next year, next 2 years, or next 5 years? We are dealing with grant proposal cycles and are interested in when we will be able to utilize Mplus for 4-level analysis?

Sam Vuchinich
 Linda K. Muthen posted on Tuesday, September 15, 2009 - 5:51 pm
It will be within the next year.
 Amir Sariaslan posted on Wednesday, January 06, 2010 - 6:27 am
Dr. Muthén,

Does that estimated time frame also include non-nested random effects models, i.e. "cross-classified" and "multiple membership" models?

Sincerely,
Amir
 Linda K. Muthen posted on Wednesday, January 06, 2010 - 4:14 pm
These will not be in Version 6. They may be in Version 6.1.
 Amir Sariaslan posted on Wednesday, January 06, 2010 - 5:53 pm
I can't wait! :-) Thank you for the update, Dr. Muthén.
Amir
 Haug Leuschner posted on Sunday, April 25, 2010 - 9:03 pm
Dr. Muthen,

I would like to conduct a "cross-classified 2-level model multigroup CFA" (for testing measurement invariance), this is EXAMPLE 9.11 of the user's guide but for cross-classified data (like pupils are nested within schools, and also within neighborhoods).


Will this be possible in version 6.1 and when will version 6.1 be available?

Sincerely,
Haug
 Haug Leuschner posted on Sunday, April 25, 2010 - 9:28 pm
Dr. Muthen,

I have ordered Mplus 6 to conduct a "cross-classified 2-level model multigroup CFA". I have "ratings for several stimuli by each observer" (level 1) nested within "country of observer" crossed with "country of stimuli" (level 2). Level 2 is a 3x3 design (3 countries each) for explaining level 1. Is there any workaround available in version 6 (may be a latent variable trick or declaring fixed effects)?

Sincerely,
Haug
 Linda K. Muthen posted on Monday, April 26, 2010 - 2:11 pm
Cross-classification is not in Version 6. It is on the list for future development. There has been no date established for Version 6.1.
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: