**************************************************************************************
**************************************************************************************
**																					**
**		Do-File for the Working Paper How to explain ministerial turnover: 			**
**				macro-political factors or biographic determinants –				**
**	an event history analysis of the East German Länder ministers (1990-2011)		**
**																					**
**	The corresponding dataset can be downloaded at www.sebastianjaeckle.de			**
**	Before using it, please contact the author										**
**																					**
**************************************************************************************

**Boxplots of duration by Bundesland
			graph hbox duration  , over(bundesland_code)

**Kaplan Meier estimators**
			* duration = time variable, censored = death variable
			stset duration, failure(censored==0) scale(1)
			
** Graph 1 - Kaplan-Meier estimates by "after_1995"
			sts	graph,	by(after_1995) name(km_pooled, replace)ci
			sts	graph,	hazard	by(after_1995) name(hazard_pooled, replace)ci
			
** Graph Annex1 - Kaplan-Meier estimates by "sex"		
			sts	graph,	by(sex) name(km_pooled, replace) ci
			sts	graph,	hazard	by(sex) name(hazard_pooled, replace)ci
			
** Graph Annex2 - Kaplan-Meier estimates by "party_code"		
			sts	graph,	by(party_code) name(km_pooled, replace) ci
			sts	graph,	hazard	by(party_code) name(hazard_pooled, replace) --> Fehlermeldung: left and right boundary regions overlap; specify smaller bandwidth(s) in width()
			
** Graph Annex3 - Kaplan-Meier estimates by "bundesland_code"			
			sts	graph,	by(bundesland_code) name(km_pooled, replace)
			sts	graph,	hazard	by(bundesland_code) name(hazard_pooled, replace)ci
			
**Multivariate analysis - Cox-models

**Model 1
			stset duration, failure(censored==0) scale(1)
			stcox   spg coalition_gov_prime parl_strength ideol_dist important_ministry  , basehc(h) bases(s) basech(c)  exactm 
			stcurve, hazard
			stcurve, surv
			stcurve, cumh
			drop h s c

**multicollinearity 
regr duration spg coalition_gov_prime parl_strength ideol_dist important_ministry 
vif

**Model 2
			stset duration, failure(censored==0) scale(1)
			stcox    age_end phd sex  regional_rooted party_member_duration      , basehc(h) bases(s) basech(c)  exactm 
			stcurve, hazard
			stcurve, surv
			stcurve, cumh
			drop h s c

**Model 3
			stset duration, failure(censored==0) scale(1)
			stcox    bloc west career np_min premier spell portfolio      , basehc(h) bases(s) basech(c)  exactm 
			stcurve, hazard
			stcurve, surv
			stcurve, cumh
			drop h s c
			
**Grambsch-Therneau-Test for proportional hazards
			stset duration, failure(censored==0) scale(1)
			drop sca* sch*
			stcox bloc west career np_min premier spell portfolio   , scaledsch(sca*) schoenfeld(sch*)  efron
			estat phtest, detail
			estat phtest, rank detail			
			

**Model 4
			stset duration, failure(censored==0) scale(1)
			stcox    spg coalition_gov_prime parl_strength  important_ministry age_end phd sex  regional_rooted party_member_duration  west career np_min premier spell portfolio      , basehc(h) bases(s) basech(c)  exactm 
			stcurve, hazard
			stcurve, surv
			stcurve, cumh
			drop h s c

**Grambsch-Therneau-Test for proportional hazards
			stset duration, failure(censored==0) scale(1)
			drop sca* sch*
			stcox spg coalition_gov_prime parl_strength  important_ministry age_end phd sex  regional_rooted party_member_duration bloc west career  premier spell portfolio  , scaledsch(sca*) schoenfeld(sch*)  efron
			estat phtest, detail
			estat phtest, rank detail