Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Maarten Buis <maartenlbuis@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: ZOIB procedure |
Date | Tue, 20 Sep 2011 12:38:58 +0200 |
On Tue, Sep 20, 2011 at 12:19 PM, Prerna S <maruiprerna@gmail.com> wrote: > What i really should have asked is how does the marginal effects > option in zoib combine the effects of all 2-3 equations in the model? > Heckman (I understand it's a different procedure and not really > comparable) uses a mills ratio to do so. Does zoib do something > similar? No, this is also why I warned you not to think of -zoib- as a selection model. In principle you can estimate a zoib by just estimating a beta-model for the part between 0 and 1, and a logit for the exact 0s and exact 1s. -zoib- assumes that these two processes are completely separate. However, by estimating them in one model you can compute a expected proportion that includes the inflation parts, and that is what the default prediction is, and through that becomes the default for marginal effects. However, for you this is irrelevant as you said earlier that you do not to care about this expected proportion but only about the conditional proportion and the probability of having an exact 0. Below is an example that shows that the equations in -zoib- are completely separate. *--------------------- begin example ---------------------- // get data use http://fmwww.bc.edu/repec/bocode/k/k401.dta, clear replace totemp = totemp/100 // estimate zoib as one model zoib prate mrate totemp age sole, /// oneinflate( mrate totemp age sole) // Indication 1: // the covariances of the proportion and oneinflate equation // and oneinlate and ln_phi equation are effectively 0 matlist e(V) // Indication 2: // estimate zoib in two separate parts leads to exactly // the same estimates betafit prate , mu(mrate totemp age sole) gen one = prate == 1 if prate < . logit one mrate totemp age sole *---------------------- end example ----------------------- -- Maarten -------------------------- Maarten L. Buis Institut fuer Soziologie Universitaet Tuebingen Wilhelmstrasse 36 72074 Tuebingen Germany http://www.maartenbuis.nl -------------------------- * * 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/