Multiple membership model PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Scott C. Roesch posted on Tuesday, June 30, 2015 - 3:43 am
I have a large data set where employees (level-1) are clustered within agencies (level-2). A majority of the employees have worked in multiple agencies (i.e., employees are nested within more than 1 agency). Two questions...can MPlus handle a multiple membership model such as this. If not, is there a procedure in MPlus that would allow me to create a variable to sum the number of agencies that an employee works in so that I can treat this as a fixed effect. For example, if a participant has data that looks like the following:

agencyid employeeid
1 1
1 1
2 1
2 1

Because this employee worked in 2 agencies I would like to be able to create a variable that would return a value of 2, since this employee worked in 2 agencies. Thank you!
 Tihomir Asparouhov posted on Wednesday, July 01, 2015 - 6:19 pm
Consider using TYPE=CROSSCLASSIFIED; see for example User's Guide example 9.25 where the observations are nested within the two clustering variable: agencyid employeeid.

See also these slides
https://www.statmodel.com/download/handouts/Beijing2012-Day3.pdf
starting at page 110
 Pia Kreijkes posted on Wednesday, March 27, 2019 - 6:54 am
Hi, I looked at the TYPE = CROSSCLASSIFIED syntax in the handout and could not see any examples in which weights are applied to reflect the relative contribution of different groups in a multiple membership model. I have students nested in teachers with some students having multiple teachers or having changed teachers across the considered time period. I don't want that all teachers are given equal weight but that those who taught students for longer receive more weight. Can I do this in Mplus? Thanks.
 Tihomir Asparouhov posted on Friday, March 29, 2019 - 9:44 am
We don't use weights like that. You might find useful the example on page 183

https://www.statmodel.com/download/handouts/Beijing2012-Day3.pdf
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: