Related Experiment Video
Updated: Jun 23, 2025

Problem-Solving Before Instruction PS-I: A Protocol for Assessment and Intervention in Students with Different Abilities
Published on: September 11, 2021
Sensitivity analysis for principal ignorability violation in estimating complier and noncomplier average causal
Trang Quynh Nguyen1, Elizabeth A Stuart1,2,3, Daniel O Scharfstein4
1Department of Mental Health, Johns Hopkins School of Public Health, Baltimore, Maryland.
This study introduces sensitivity analysis for principal ignorability (PI) assumption violations in causal effect estimation with noncompliance. It assesses how effect estimates change when compliers and noncompliers differ under control conditions.
Area of Science:
- Causal Inference
- Statistical Sensitivity Analysis
- Econometrics
Background:
- Principal ignorability (PI) is crucial for identifying causal effects with noncompliance but is untestable.
- Assessing the sensitivity of effect estimates to PI violations is essential for robust causal inference.
- Noncompliance creates principal strata (compliers, noncompliers) that may differ in unobserved ways.
Purpose of the Study:
- To develop and apply sensitivity analysis methods for the principal ignorability assumption in one-sided noncompliance settings.
- To quantify the impact of potential PI violations on causal effect estimates.
- To provide practical tools for researchers to assess the robustness of their findings.
Main Methods:
- Sensitivity analysis is performed by allowing differences in outcome-mean-given-covariates functions between compliers and noncompliers under the control condition.
- Various sensitivity parameters are used, including odds ratio, generalized odds ratio, mean ratio, and standardized mean difference.
- Techniques are tailored to principal ignorability-based analysis methods like outcome regression, influence function (IF)-based, and weighting methods.
Main Results:
- The study demonstrates how to tailor sensitivity analysis techniques to different PI-based main analysis methods.
- It provides guidance on selecting appropriate ranges for sensitivity parameters.
- Illustrations using the JOBS II study showcase the application of these methods for various outcome types.
Conclusions:
- Sensitivity analysis is a vital tool for evaluating the robustness of causal effect estimates derived under the principal ignorability assumption.
- The proposed methods offer practical approaches for researchers to gauge the potential impact of PI violations.
- Understanding the sensitivity of results enhances the credibility and interpretability of causal findings in the presence of noncompliance.
Related Concept Videos
Mechanistic Models: Compartment Models in Individual and Population Analysis
Strategies for Assessing and Addressing Confounding
Confounding can be addressed at both the design phase of a study and through analytical methods after data...
Confounding in Epidemiological Studies
One-Way ANOVA
Censoring Survival Data
Variability: Analysis
The range is a simple measure of variability, indicating the difference between the highest and...

