Concepts (S)
Propensity Score Analysis
This section provides a comprehensive exploration into various facets of propensity score (PS) methods and their application in observational studies and surveys. Beginning with an in-depth look into key concepts and calculations related to ATE and ATT, the content navigates through the practical application and diagnostic checks of covariate balance using the SMD. It further elucidates the methodology and application of PS, particularly focusing on matching and weighting to mitigate bias and create comparable groups for causal inference. The intricacies of employing PS methods within surveys are explored, highlighting different approaches and the incorporation of design variables in PS and outcome models. Fundamental assumptions for causal inference, namely Conditional Exchangeability, Positivity, and Causal Consistency, are dissected to form a foundational understanding for conducting robust causal analyses. Additionally, the content optionally delves into the nuances of implementing IPW in surveys. Lastly, additional optional content features an insightful workshop, offering more explanations of PS method implementations in research contexts.
Reading list
Key reference: (Peter C. Austin 2011)
Optional reading:
- Propensity score introduction (Karim 2021) External link
- Extensions of Propensity score approaches External link: prepared for Guest Lecture in SPPH 500/007 (Analytical Methods in Epidemiological Research)
- Propensity score for complex surveys External link: Uses the same lectures here, with some added text descriptions. This also includes a a structured framework for reporting analyses using PS methods in research manuscripts.
- Reporting guideline (Stuart 2018; Simoneau et al. 2022)
- Assumptions (Hernán and Robins 2020)
Theoretical references for propensity score analyses in complex surveys:
(Peter C. Austin, Jembere, and Chiu 2018; DuGoff, Schuler, and Stuart 2014; Zanutto 2006; Leite, Stapleton, and Bettini 2018; Lenis, Ackerman, and Stuart 2018; Lenis et al. 2017; Ridgeway et al. 2015)
Video Lessons
Average Treatment Effect (ATE) vs. Average Treatment effect on the Treated (ATT)
Balance and standardized mean difference (SMD) in observational studies
Post Conference Workshop for 2021 Conference - Canadian Society for Epidemiology and Biostatistics (CSEB)
Video Lesson Slides
Target parameters
Balance
Propensity score matching
Propensity score matching in complex survey
Propensity score weighting in complex survey
Causal Assumptions
FAQ
Links
Target parameters
Balance
Propensity score matching
Propensity score matching in complex survey
Propensity score weighting in complex survey
Causal Assumptions