We're trying to recreate an analysis done in a book called "The Fate of
Young Democracies" which applies a Weibull model to the duration of
democracies. Two questions:
1) In the book they claim that the hazard function they are using is of the
form h(t | x_t) = pt^{p-1}e^{B0 + x_tB1}. This does not seem to be in the
same form as the hazard function described in either Zelig or Wikipedia.
That hazard function is (roughly) h(y | lambda, alpha) = alpha /
lambda^alpha * y^{alpha - 1) e^{-(y / lambda)^alpha}, where lambda =
e^{Xbeta}. Anyone more familiar with using Weibull models have any clue for
why that difference might exist?
2) The reported values are "percentage change in the baseline hazard rate in
response to a one-unit increase in the independent variable". What would
"baseline" likely constitute in this context, and is there a straightforward
way to calculate this value using Zelig? Does this likely amount to a first
difference on the hazard function after setting covariates to their means,
or something along those lines?
Thanks,
Nick
Show replies by date