*You will need to install the spost_ado.pkg STATA categorical commands developed by Long et al.
* Remove the * in front of the next two lines 
*net describe spost9_ado, from(http://www.indiana.edu/~jslsoc/stata)
*net install spost9_ado.pkg
*Get the data, either from EUP or my own webpage
*use "http://www.uni-konstanz.de/eup/Data/Hoyland.dta", clear
use "http://folk.uio.no/bjornkho/EUPdata.dta", clear
*The next command reproduces the descriptive table
by government, sort : tabulate epgroup, summarize(reports)
*The next few lines estimates the alternative models 
zinb reports government  nominate partysize, /// 
inflate(government  nominate partysize) vuong zip
listcoef, help
estimates store theory
zinb reports government interaction nominate partysize  /// 
position salience dissent ep epp green pes eldr, /// 
inflate(government interaction  nominate partysize ///
position salience dissent ep eldr epp green pes) vuong zip
estimates store full
*A formal likelihood-ratio test of the models
lrtest ( full) ( theory ), force
*Testing the papameters of the models
testparm government nominate interaction
testparm government nominate
testparm epp eldr green pes, equal
testparm position salience dissent ep
testparm epp pes, equal
testparm eldr pes, equal
testparm green pes, equal
*Predicting number of reports
// prcounts prerep, max(5) plot
label var prerepobeq "Observed Proportion"
label var prereppreq "Predicted Proportion"
label var prerepval "Number of Reports"
list prerepval prerepobeq prereppreq in 1/6
*Predicting the standard deviation and unit change
listcoef, help
prchange nominate, rest(mean)
*Predited values 
prvalue, x(government=0) rest(mean)save
prvalue, x(government=1) rest(mean)dif
prvalue, x(government=0 pes=1 epp=0 eldr=0 green=0) rest(mean)save
prvalue, x(government=1 pes=1 epp=0 eldr=0 green=0) rest(mean)dif
prvalue, x(government=0 epp=1 pes=0 eldr=0 green=0) rest(mean)save
prvalue, x(government=1 epp=1 pes=0 eldr=0 green=0) rest(mean)dif