[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Mixed Logit, error component and independence from observables
am not sure of appropriateness of posting non-stata question;
i note the inconsistency with estimating a RE probit where explanatory variables
are correlated with unobserved component,c. without knowledge of c, we need
integrate back over normal unobservables to obtain y|x only and must assume
independence of (x,c) given prob rules ?
unless one models the exact correlation...
with mixed logit we allow coefficients to have a mean and a deviation, "whose
distribution over individuals and alternatives depends in general on underlying
parameters and observed data relating to alternative i and individual q". to my
mind we are modelling this conditional distribution c|x explicitly so do not
assume independence. same with mixture of normals probit ?
any clarification appreciated,
* For searches and help try: