A mixture model for duration data: analysis of second births in China

Area of Science:

• Demography
• Biostatistics
• Epidemiology

Background:

• Analyzing duration data, such as time to second birth, requires sophisticated statistical models.
• Traditional models may not adequately capture the dual influence of factors on both event occurrence and timing.

Purpose of the Study:

• To introduce a novel mixture model integrating logistic regression and piecewise proportional hazards models.
• To simultaneously estimate covariate effects on the probability and timing of events.
• To apply this model to understand factors influencing second births in China.

Main Methods:

• Development of a mixture model combining logistic regression and piecewise proportional hazards.
• Application of the model to analyze data on second births in China.
• Estimation of covariate effects on both the probability and duration of the second birth interval.

Main Results:

• Both women's characteristics and one-child certificate acceptance affected the probability of having a second child.
• Acceptance of a one-child certificate demonstrated a significant and strong effect on the duration of the second-birth interval.
• Covariate effects on event probability and event timing were estimated simultaneously.

Conclusions:

• The proposed mixture model effectively analyzes duration data by separating effects on event probability and timing.
• Policy interventions, like the one-child certificate, can have distinct impacts on fertility behavior and its timing.
• This modeling approach offers valuable insights for demographic and epidemiological research.