* Author: William D. MacMillan
* Prepared for ICPSR time-series and cross-section data analysis
* xtprobit, xtlogit, and interactions
* 5/22/09
* 7/9/10	update
* RJFjr revise 7/7/11

* need to install btscs first!
* net from http://www-personal.umich.edu/~wmacmill/stata/btscs


clear all

set mem 500m
use http://www-personal.umich.edu/~franzese/russetandoneilwardata.dta

local model mzcowwar1 smldem20 smldep dircont majpower lndstab lncaprat allies systsize absdemdis py _spline1 _spline2
local rhs smldem20 smldep dircont majpower lndstab lncaprat allies systsize absdemdis py _spline1 _spline2

xtset dyadid year

* Lets help xt (explore qualdeps now available), btscs, xtlogit, and xtprobit...
more
help xt
more
help btscs
more
help xtlogit
more
help xtprobit
more

* OK, let's try some
btscs mzcowwar1 year dyadid, g(py) nspline(2)


browse

* Here come the estimates, comparing logit to logit with splines...
more

logit mzcowwar1 smldem20 smldep dircont majpower lndstab lncaprat allies systsize absdemdis 

logit `model'
more

* now compare to Chamberlain's Fixed-Effects (Conditional) Logit:
more
xtlogit `model', fe 
more

* random-effects logit is now also available (forced small # iterations for time):
more
xtlogit `model', i(dyadid) re iter(2)

* and random-effects probit...
more
xtprobit `model', i(dyadid) re iter(1)
more

* THANK YOU!