TITLE:
Survival Analysis in Modeling the Birth Interval of the First Child in Indonesia
AUTHORS:
Rahmat Hidayat, Hadi Sumarno, Endar H. Nugrahani
KEYWORDS:
Analysis Survival; Censored; Covariate; Cox Proportional Hazard; Cox Extended
JOURNAL NAME:
Open Journal of Statistics,
Vol.4 No.3,
April
14,
2014
ABSTRACT: First birth interval is one of the examples
of survival data. One of the characteristics of survival data is its
observation period that is fully unobservable or censored. Analyzing the
censored data using ordinary methods will lead to bias, so that reducing such
bias required a certain method called survival analysis. There are two methods
used in survival analysis that are parametric and non-parametric method. The
objective of this paper is to determine the appropriate method for modeling the
birth of the first child. The exponential model with the inclusion of
covariates is used as parametric method, considering that the newly married
couples tend to have desire for having baby as soon as possible and such desire
will be weakened by increasing age of marriage. The data that will be analyzed
were taken from the Indonesia Demographic and Health Survey (IDHS) 2012. The
result of data analysis shows that the birth of the first child data is not
exponentially distributed thus the Cox proportional
hazard method is used. Because of the suspicion that disproportional
covariate exists, then the proportional hazard test is conducted to show that the covariate of age is not proportional, the
generalized Cox proportional method
is used, namely Cox extended that
allows the inclusion of disproportional covariates. The result of analysis
using Cox extended model indicates
that the factors affecting the birth of the first child in Indonesia are the
area of residence, educational history and its age.