|
Message/Author |
|
Darren posted on Thursday, April 17, 2014 - 5:20 am
|
|
|
A "latent budget" model is described in chapter 4 of the Applied LCA book (Hagenaars and McCutcheon). It is also described here: http://statistics.ucla.edu/system/resources/BAhbBlsHOgZmSSI7MjAxMi8wNS8zMC8xNV8xMV8xOF83MzdfQV9MYXRlbnRfVGltZV9CdWRnZXRfTW9kZWwucGRmBjoGRVQ/A%20Latent%20Time-Budget%20Model.pdf It is essentially an asymetric re-organization of the latent class model. Given a matrix (i,j) with observations i and variables j, the joint probability of (i,j) is a function of k-class specific latent budgets described by Beta_j (which must sum to 1) multiplied by a set of mixing proportions Alpha_i (which must also sum to 1) - so that the observed values for any observation's variables are described as a mixture of latent budgets. Despite its close relationship to a latent class model, I can't figure out how to estimate the model parameters in Mplus, though I strongly suspect it will involve the use of KNOWNCLASS. Is this something you can shed some light on? |
|
|
I haven't studied this model. Anybody? I guess I should read up on it. |
|
Back to top |
|
|