Abstract
To estimate a survival rate curve by the Utstein Style data, though a number of studies have used a Logit model and a Probit model generally, these methods are not necessarily optimal and desirable. By comparing with their curves estimated by a Scobit model based on Burr II distribution and a semi-nonparametric model with the Hermite polynomial expansion from the predictive point of view, this study aims to propose more desirable methods to estimate a survival rate curve. In the public-works project assessment, the estimation of a survival rate curve showing the relation between an emergency transport time and a lifesaving rate can make a significant contribution to measuring a mortality risk reduction benefit produced by reducing any temporal loss by road improvement and maintenance.
