Bayesian latent class models for capture-recapture in the presence of missing data

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Modeling

Background:

• Multiple record systems are crucial for population size estimation.
• Missing data presents a significant challenge in these systems.
• Accurate population size estimation is vital for public health surveillance.

Purpose of the Study:

• To propose a novel statistical method for population size estimation in multiple record systems.
• To address the challenge of missing data within these systems.
• To validate the proposed method using a known dataset.

Main Methods:

• A latent class model was employed for parameter estimation.
• A Gibbs sampler was utilized to estimate the latent structure.
• The method was applied to a dataset of neural tube defects registrations.

Main Results:

• The proposed latent class model effectively estimates population size.
• The Gibbs sampler successfully handles missing data in multiple record systems.
• The method's performance was demonstrated on a real-world dataset.

Conclusions:

• The developed method provides a robust approach to population size estimation.
• This technique is particularly valuable when dealing with incomplete records.
• The findings have implications for epidemiological studies and public health.