Module 5 — Modern Causal Analysis

Sensitivity and Partial Identification

Quantify how conclusions move when exchangeability, trends, exclusion or missingness assumptions are relaxed

Sensitivity and Partial Identification

Robustness is not a list of specifications

Adding controls, deleting outliers and changing standard errors asks whether results are numerically stable. Sensitivity analysis asks a sharper question:

How large must a specified violation of the identifying assumption be to change the conclusion?

The answer is conditional on the sensitivity model. It does not certify that the violation is smaller.

Match the sensitivity analysis to the design

Design threatSensitivity parameter or boundUseful output
unmeasured confoundingstrength of omitted treatment/outcome associationadjusted effect or robustness value
DID trend violationpermitted post-period deviation from pre-trendinterval over deviation sizes
IV exclusion/ignorabilityinstrument side effect or instrument confoundingadjusted IV estimate/interval
attritionmissing potential-outcome range or monotonicityidentified effect interval
hidden bias after matchingodds-of-treatment departureconclusion across Γ\Gamma

Use the design’s weak point. An IV exclusion analysis does not address weak instruments; a DID trend analysis does not address anticipation.

Benchmark omitted confounding

Suppose adjustment gives an eight-point completion effect. Rather than “results survive many controls,” compare a hypothetical omitted variable with observed adviser support:

  • how strongly would it need to explain scholarship receipt, conditional on controls?
  • how strongly would it need to explain completion residuals?
  • would that joint strength reduce the estimate to zero or only below a policy threshold?

Cinelli and Hazlett (2020) express omitted-variable sensitivity using partial R2R^2 measures and observed-variable benchmarks. For IV, Cinelli and Hazlett (2025) separately quantify possible side effects of the instrument and confounding of the instrument. Zhang and Zhao (2026) bound average, rather than only worst-case, confounding strength in a newer marginal-sensitivity model.

Always label new methods by publication status and implementation maturity in the reading map.

Bounds expose what missing outcomes permit

Completion is binary. Response is 80% in the offer arm with observed mean 0.60, and 90% in control with mean 0.50. Without assumptions about missing outcomes:

E[Y(1)][0.8(0.60),  0.8(0.60)+0.2]=[0.48,0.68],E[Y(1)]\in[0.8(0.60),\;0.8(0.60)+0.2]=[0.48,0.68],E[Y(0)][0.9(0.50),  0.9(0.50)+0.1]=[0.45,0.55].E[Y(0)]\in[0.9(0.50),\;0.9(0.50)+0.1]=[0.45,0.55].

Therefore the effect lies in

[0.480.55,  0.680.45]=[0.07,0.23].[0.48-0.55,\;0.68-0.45]=[-0.07,0.23].

The data alone permit harm or substantial benefit. A credible monotonicity, response or administrative-linkage assumption can narrow the set—but must be stated.

Sensitivity curves beat a single threshold

Show the effect estimate and interval over a substantively interpretable range. Mark:

  • zero;
  • a policy-relevant minimum effect;
  • benchmarks from observed covariates or pre-trends;
  • the range experts regard as plausible;
  • where support becomes too weak to calculate reliably.

For staggered DID, Rambachan and Roth (2023) make conclusions conditional on restrictions relating post-treatment trend violations to pre-treatment deviations.

Quick check

The effect remains statistically non-zero until an omitted confounder is twice as strong as any observed covariate. Is the study now proven causal?

Answer
No. The benchmark may omit the strongest plausible confounder, and the sensitivity model may not capture interference, measurement error or selection. Report the calibrated claim: the conclusion survives violations of the specified form up to the stated strength.

Next: External Validity

Copyright © 2026