btscs {DAMisc} | R Documentation |
Beck et. al. (1998) identified that binary time-series cross-section data are discrete-time duration data and time dependence can be modeled in a logistic regression by including a flexible function (e.g., cubic spline) of time since the last event as a covariate. This function creates the variable identifying time since last event.
btscs(data, event, tvar, csunit, pad.ts = FALSE)
data |
A data frame. |
event |
Character string giving the name of the dichotomous variable identifying the event (where an event is coded 1 and the absence of an event is coded 0). |
tvar |
Character string giving the name of the time variable. |
csunit |
Character string giving the name of the cross-sectional unit. |
pad.ts |
Logical indicating whether the time-series should be filled in, when panels are unbalanced. |
The original data frame with one additional variable. The
spell
variable identifies the number of observed periods since the
last event.
Dave Armstrong
Alvarez, M., J.A. Cheibub, F. Limongi and A. Przeworski. 1996. Classifying political regimes. Studies in Comparative International Development 31 (Summer): 1-37.
Beck, N.. J. Katz and R. Tucker. 1998. Beyond Ordinary Logit: Taking Time Seriously in Binary-Time-Series-Cross-Section Models. American Journal of Political Science 42(4): 1260-1288.
library(splines)
## Data from Alvarez et. al. (1996)
data(aclp)
newdat <- btscs(aclp, "democ", "year", "country")
# Estimate Model with and without spell
full.mod <- glm(democ ~ log(gdpw) + popg + bs(spell, df=4), data=newdat, family=binomial)
restricted.mod <- glm(democ ~ log(gdpw) + popg, data=newdat, family=binomial)
# Incremental F-test of time dependence
anova(restricted.mod, full.mod, test='Chisq')