Invited Discussion of "A Unified Framework for De-Duplication and Population Size Estimation"

TL;DR

This paper discusses the advantages and challenges of joint models for record linkage and population size estimation, emphasizing the importance of nuanced cost-benefit analysis and performance measurement.

Contribution

It provides a critical discussion on the trade-offs of joint modeling approaches and proposes improved methods for evaluating their performance.

Findings

Joint models offer benefits but have complex cost structures

Performance metrics need to be more nuanced and task-specific

Better evaluation methods can improve model selection and application

Abstract

Invited Discussion of "A Unified Framework for De-Duplication and Population Size Estimation", published in Bayesian Analysis. My discussion focuses on two main themes: Providing a more nuanced picture of the costs and benefits of joint models for record linkage and the "downstream task" (i.e. whatever we might want to do with the linked and de-duplicated files), and how we should measure performance.