Weighting regressions by propensity scores.

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Propensity score weighting is a common method to address confounding in observational studies.
• Weighting aims to create pseudo-populations where treatment assignment is independent of observed covariates.

Purpose of the Study:

• To evaluate the impact of propensity score weighting on bias and precision of causal parameter estimates.
• To compare weighting methods with direct causal modeling approaches.

Main Methods:

• The study theoretically analyzes the effects of propensity score weighting on regression estimates.
• It considers scenarios with both correctly and incorrectly specified causal models.

Main Results:

• Weighting can reduce bias but often inflates random error and biases standard error estimates downwards.
• In certain situations, propensity score weighting may exacerbate bias in causal parameters.
• Directly fitting a well-specified causal model generally yields more reliable estimates than weighting.

Conclusions:

• While propensity score weighting can be useful, its limitations regarding error inflation and potential bias must be considered.
• Investigators with a strong causal model should prioritize fitting the model directly.
• Weighting may offer some benefits when causal models are misspecified, but significant challenges remain.