We begin by opening the cabinet dataset.
#-------Dataset
cabinet <- read.dta("~/Dropbox/event_history/dta/cabinet.dta")
We create the survival object for our dependent variable, using the syntax Surv(time, event).
#--------Data prepation
dv <- Surv(cabinet$durat, cabinet$'_d')
We run the Generalized Gamma and Weibull models.
# Generalized Gamma model
gg_mod <- flexsurvreg(dv ~invest + polar + numst + format + postelec + caretakr,
data = cabinet, dist = "gengamma")
# Weibull model
weib_mod <- survreg(dv ~invest + polar + numst + format + postelec + caretakr,
data = cabinet)
Let’s compare the results of our models.
| Generalized Gamma | Weibull | ||
|---|---|---|---|
| Constant | 2.96 (0.14) | 2.99 (0.13) | |
| Investiture | -0.30 (0.11) | -0.30 (0.11) | |
| Polarization | -0.02 (0.01) | -0.02 (0.01) | |
| Majority | 0.47 (0.10) | 0.46 (0.10) | |
| Formation | -0.10 (0.03) | -0.10 (0.03) | |
| Post-Election | 0.68 (0.11) | 0.68 (0.10) | |
| Caretaker | -1.33 (0.21) | -1.33 (0.20) | |
| Shape Parameter | 0.79 | 0.77 | |
| Scale Parameter | 0.92 | ||
| Log Likelihood | -1014.55 | -1014.62 | |
| N | 314 | 314 | |