2020
DOI: 10.1111/rssa.12621
|View full text |Cite
|
Sign up to set email alerts
|

Did you Conduct a Sensitivity Analysis? A New Weighting-Based Approach for Evaluations of the Average Treatment Effect for the Treated

Abstract: In non‐experimental research, a sensitivity analysis helps determine whether a causal conclusion could be easily reversed in the presence of hidden bias. A new approach to sensitivity analysis on the basis of weighting extends and supplements propensity score weighting methods for identifying the average treatment effect for the treated (ATT). In its essence, the discrepancy between a new weight that adjusts for the omitted confounders and an initial weight that omits them captures the role of the confounders.… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
10
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
5
2

Relationship

1
6

Authors

Journals

citations
Cited by 14 publications
(10 citation statements)
references
References 60 publications
0
10
0
Order By: Relevance
“…Rosenbaum (1987) proposed a sensitivity analysis for matched estimators that computes the largest odds ratio between the probability of treatment, conditional on X i , and the probability of treatment, conditional on both X i and U i , before the estimated treatment effect becomes zero. While the original framework inherently requires the assumption of a constant treatment effect, recent literature has allowed for treatment effect heterogeneity and other extensions (i.e., Tan (2006), Shen et al (2011), Zhao et al (2019, Hong et al (2021), Dorn and Guo (2021), Nie et al (2021)). Alternative simulation-based approaches have also been developed, in which researchers must invoke a distributional assumption on the omitted confounder U i (i.e., Ichino et al (2008);Tozzi et al (2019); Burgette et al (2020)).…”
Section: Related Literaturementioning
confidence: 99%
See 2 more Smart Citations
“…Rosenbaum (1987) proposed a sensitivity analysis for matched estimators that computes the largest odds ratio between the probability of treatment, conditional on X i , and the probability of treatment, conditional on both X i and U i , before the estimated treatment effect becomes zero. While the original framework inherently requires the assumption of a constant treatment effect, recent literature has allowed for treatment effect heterogeneity and other extensions (i.e., Tan (2006), Shen et al (2011), Zhao et al (2019, Hong et al (2021), Dorn and Guo (2021), Nie et al (2021)). Alternative simulation-based approaches have also been developed, in which researchers must invoke a distributional assumption on the omitted confounder U i (i.e., Ichino et al (2008);Tozzi et al (2019); Burgette et al (2020)).…”
Section: Related Literaturementioning
confidence: 99%
“…The contributions of this paper are two-fold: (1) develop a sensitivity analysis framework for the generalization or transportation of experimental results without requiring distributional or functional form assumptions on the individual-level treatment effect or confounder, and (2) provide a set of tools for researchers to transparently justify plausible ranges of values for the different sensitivity parameters. The proposed sensitivity analysis extends the frameworks developed by Hong et al (2021) and Shen et al (2011), and allows researchers to discuss potential substantive changes to a point estimate due to unobserved confounding; however, it does not directly address the effect of omitted confounders on uncertainty. Researchers can equivalently think of this as an asymptotic analysis, as the uncertainty associated with a point estimate will disappear as the sample size gets larger (i.e., n → ∞).…”
Section: Related Literaturementioning
confidence: 99%
See 1 more Smart Citation
“…With such a specific unmeasured confounder in mind, an analyst may determine the sensitivity parameter values based on existing data or previous empirical findings. Alternatively, one may compare its unique confounding role to those of the observed confounders based on existing data, previous empirical findings, or theoretical reasoning, so that the coefficients of the observed covariates in the mediator and outcome models could be used as referent values for determining specific values or a plausible range of the sensitivity parameters (e.g., Carnegie et al, 2016; Hong et al, 2021; Imbens, 2003). For example, one may argue that the conditional associations of motivation with the mediator and the outcome do not exceed those of the baseline depression level.…”
Section: Applicationmentioning
confidence: 99%
“…More recent examples of sensitivity analysis include Rosenbaum and Rubin (1983a), Rosenbaum (2002), VanderWeele and Ding (2017), Franks et al (2019), andCinelli andHazlett (2020). See Hong et al (2020) for a recent discussion of weighting-based sensitivity methods.…”
Section: Setup and Sensitivity Modelmentioning
confidence: 99%