Innovative Propensity Score Approaches for Data Integration in Survey Methodology

Researchers Beaumont, Bosa, Brennan, Charlebois, and Chu have introduced groundbreaking model selection techniques for estimating participation probabilities in non-probability sample units. Their focus revolves around the selection of likelihood and parameterization of the model to enhance the effectiveness of the proposed methods.

They explore alternative likelihood and pseudo-likelihood-based approaches for estimating participation probabilities in surveys, emphasizing the importance of choosing the right model for accurate results. Through simulations, they demonstrate the superiority of likelihood over pseudo-likelihood in scenarios where practical significance exists.

Kim and Kwon delve into the realm of data integration within the survey methodology, specifically focusing on the construction of pseudo weights in a two-phase sampling framework. They present two approaches for estimating propensity scores and introduce a novel method for constructing the propensity score function using the conditional maximum likelihood technique.

Gershunskaya and Beresovsky contribute to the ongoing conversation by proposing innovative strategies for handling data combining in surveys. Their research sheds light on the intricacies of sample likelihood and participation probabilities, highlighting the importance of proper model selection in ensuring the accuracy of survey results.

Statistics Canada plays a crucial role in fostering collaboration and research in statistical methodologies, providing a platform for researchers to share their findings and contribute to the advancement of survey methodology.