Dear statalisters,
having tried to estimate a sequential binary choice model by seqlogit and recognizing that only pseudo-likelihood estimates are reported, because it uses cluster correction for panel data, I remembered the text I read on W. Greene's Econometric Analysis 6th edition page 517 which I copy hereunder:
"We can justify the cluster estimator based on this approximation (i.e. that the second term of the squared error outer product in the asymptotic mean squared deviation equals zero). in general, it will estimate the expected squared variation of the pseudo-MLE around its probability limit. Whether it measures the variation around the appropriate parameters of the model hangs on whether the second term (as I have quoted in the previous parethesis) equals zero...Is that likely?Certainly not if the pooled model is ignoring unobservable fixed effects. Moreover, it will be inconsistent in most cases in which the misspecification is to ignore latnent random effects as well...It is not consistent in the probit and logit models in which this approach is often used."
Based on this critique, how could one justify consistency of seqlogit routine's results and under which statistical circumstances? Is there a better alternative for sequential choices in terms of packaged routines in Stata, assuming that Nested logit would not be applicable-because of lack of choice-specific covariates-?
many thanks in advance for your comments,
Kostas Konstantaras
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
