[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: interpretation of latent class model results

From   "Mentzakis, Emmanouil" <[email protected]>
To   <[email protected]>
Subject   st: RE: interpretation of latent class model results
Date   Fri, 30 May 2008 11:57:57 +0100



Each class of the LCM can be interpreted as a separate conditional logit. You cannot directly compare coefficients across classes. However, you can take wpt values for each class and then compare them across classes. 


The reason for not being able to compute marginal effects for each class is because you don't know what the classes are, they are "latent". The marginal effects are based on averaging the estimates across the class estimates probabilistically.





From: [email protected] on behalf of Simona Rasciute
Sent: Thu 5/29/2008 8:08 PM
To: [email protected]
Subject: st: interpretation of latent class model results


I would like to ask you a question regarding the
interpretation of the latent class model results. Usually
in discrete choice models the magnitude of the estimated
parameters is not informative and elasticities have to be
estimated for continuous variables and marginal effects
for dummy variables. Does that mean that in the latent
class model the coefficients for different classes cannot
be compared (the willingness to pay is not applicable in
my model)? For example,  if one of the classes has the
highest coefficent for a particular variables I can
conclude that the group is more responsive to that
variable than other groups. The latent class model does
not estimate elasticities and marginal effects for each
latent class separatelly and aggregate results are not of
much help. Could you please tell me what is the best way
to analyse the results of the latent class model.

I appreciate your help.

Kind Regards,
*   For searches and help try:

The University of Aberdeen is a charity registered in Scotland, No SC013683.


© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index