Multilist population estimation with incomplete and partial stratification.

Area of Science:

• Ecology
• Population Biology
• Statistical Modeling

Background:

• Multilist capture-recapture methods are standard for estimating elusive population sizes.
• Stratification by features like age or sex aims to reduce bias from heterogeneous list membership.
• A common challenge arises when data collection lists are not active across all strata.

Purpose of the Study:

• To develop a general statistical method for estimating population size when not all capture-recapture lists are active in every stratum.
• To address biases arising from incomplete list participation across population strata.

Main Methods:

• Development of a novel expectation maximization (EM) algorithm.
• Application of a flexible log-linear modeling framework.
• Accommodating list dependencies and differential ascertainment probabilities.

Main Results:

• The proposed method effectively estimates population size in the presence of incomplete list activity across strata.
• The log-linear model framework provides flexibility in accounting for complex data structures.
• Demonstrated applicability through two real-world examples.

Conclusions:

• The developed expectation maximization (EM) algorithm offers a robust solution for population size estimation with incomplete multilist capture-recapture data.
• This method enhances the reliability of ecological and biological population estimates when data strata are unevenly covered.