The causal mediation literature has developed techniques to assess the sensitivity of an inference to pretreatment confounding, but these techniques are limited to the case of a single mediator. In this article, we extend sensitivity analysis to possible violations of pretreatment confounding in the case of multiple mediators. In particular, we develop sensitivity analyses under three alternative approaches to effect decomposition: (1) jointly considered mediators, (2) identifiable direct and indirect paths, and (3) interventional analogues effects. With reasonable assumptions, each approach reduces to a single procedure to assess sensitivity in the presence of simultaneous pre- and posttreatment confounding. We demonstrate our sensitivity analysis techniques with a framing experiment that examines whether anxiety mediates respondents’ attitudes toward immigration in response to an information prompt.
AvinC.ShpitserI.PearlJ. (2005). Identifiability of path-specific effects. In Proceedings of International Joint Conference on Artificial Intelligence (Vol. 19, pp. 357–363). Morgan Kaufmann Publishers.
3.
Bar-HaimY.LamyD.PergaminL.Bakermans-KranenburgM. J.Van IjzendoornM. H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-analytic study. Psychological Bulletin, 133(1), 1.
4.
BaronR. M.KennyD. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173.
5.
BraderT.ValentinoN. A.SuhayE. (2008). What triggers public opposition to immigration? Anxiety, group cues, and immigration threat. American Journal of Political Science, 52(4), 959–978.
6.
CinelliC.HazlettC. (2020). Making sense of sensitivity: Extending omitted variable bias. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 82(1), 39–67.
7.
CislerJ. M.KosterE. H. (2010). Mechanisms of attentional biases towards threat in anxiety disorders: An integrative review. Clinical Psychology Review, 30(2), 203–216.
8.
CoxM. G.Kisbu-SakaryaY.MiočevićM.MacKinnonD. P. (2013). Sensitivity plots for confounder bias in the single mediator model. Evaluation Review, 37(5), 405–431.
9.
DanielR. M.De StavolaB. L.CousensS. N. (2011). gformula: Estimating causal effects in the presence of time-varying confounding or mediation using the g-computation formula. The Stata Journal, 11(4), 479–517.
10.
DidelezV.DawidP.GenelettiS. (2012). Direct and indirect effects of sequential treatments. arXiv preprint arXiv:1206.6840. https://arxiv.org/abs/1206.6840
11.
FrankK. A. (2000). Impact of a confounding variable on a regression coefficient. Sociological Methods & Research, 29(2), 147–194.
12.
GadarianS. K. (2015). The politics of threat: How terrorism news shapes foreign policy attitudes. The Journal of Politics, 72(2), 469–483.
13.
GadarianS. K.AlbertsonB. (2014). Anxiety, immigration, and the search for information. Political Psychology, 35(2), 133–164.
14.
GenelettiS. (2007). Identifying direct and indirect effects in a non-counterfactual framework. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 69(2), 199–215.
15.
HongG.QinX.YangF. (2018). Weighting-based sensitivity analysis in causal mediation studies. Journal of Educational and Behavioral Statistics, 43(1), 32–56.
16.
ImaiK.KeeleL.TingleyD. (2010). A general approach to causal mediation analysis. Psychological Methods, 15, 309–334.
17.
ImaiK.YamamotoT. (2013). Identification and sensitivity analysis for multiple causal mechanisms: Revisiting evidence from framing experiments. Political Analysis, 21, 141–171.
18.
le CessieS.DebeijJ.RosendaalF. R.CannegieterS. C.VandenbrouckeaJ. P. (2012). Quantification of bias in direct effects estimates due to different types of measurement error in the mediator. Epidemiology, 23, 551–560.
19.
LokJ. J. (2019). Causal organic direct and indirect effects: Closer to Baron and Kenny. arXiv preprint arXiv:1903.04697. https://arxiv.org/abs/1903.04697
20.
MarcusG. E.MacKuenM.NeumanW. R. (2011). Parsimony and complexity: Developing and testing theories of affective intelligence. Political Psychology, 32(2), 323–336.
21.
MauroR. (1990). Understanding love (left out variables error): A method for estimating the effects of omitted variables. Psychological Bulletin, 108(2), 314.
22.
MoggK.MathewsA.BirdC.Macgregor-MorrisR. (1990). Effects of stress and anxiety on the processing of threat stimuli. Journal of Personality and Social Psychology, 59(6), 1230.
23.
NabiR. L. (1999). A cognitive-functional model for the effects of discrete negative emotions on information processing, attitude change, and recall. Communication Theory, 9(3), 292–320.
24.
ParkS.KürümE. (2018). Causal mediation analysis with multiple mediators in the presence of treatment noncompliance. Statistics in Medicine, 37(11), 1810–1829.
25.
ParkS.SteinerP.KaplanD. (2018). Identification and sensitivity analysis for average causal mediation effects with time-varying treatments and mediators: Investigating the underlying mechanisms of kindergarten retention policy. Psychometrika, 83(2), 298–320.
26.
PearlJ. (2001). Direct and indirect effects. In Proceedings of the seventeenth conference on uncertainty in artificial intelligence (pp. 411–420). Morgan Kaufmann Publishers.
27.
PearlJ. (2009). Causality: Models, reasoning, and inference (2nd ed.). Cambridge University Press.
28.
RobinsJ. M.GreenlandS. (1992). Identifiability and exchangeability for direct and indirect effects. Epidemiology, 3, 143–155.
29.
RubinD. B. (1978). Bayesian inference for causal effects: The role of randomization. The Annals of Statistics, 6, 34–58.
30.
SteenJ.LoeysT.MoerkerkeB.VansteelandtS. (2017). Flexible mediation analysis with multiple mediators. American Journal of Epidemiology, 186(2), 184–193.
31.
ValentinoN. A.HutchingsV. L.BanksA. J.DavisA. K. (2008). Is a worried citizen a good citizen? Emotions, political information seeking, and learning via the Internet. Political Psychology, 29(2), 247–273.
32.
ValeriL.LinX.VanderWeeleT. J. (2014). Mediation analysis when a continuous mediator is measured with error and the outcome follows a generalized linear model. Statistics in Medicine, 33(28), 4875–4890.
33.
VanderWeeleT. J. (2010). Bias formulas for sensitivity analysis for direct and indirect effects. Epidemiology (Cambridge, Mass.), 21(4), 540.
34.
VanderWeeleT. J. (2015). Explanation in causal inference: Methods for mediation and interaction. Oxford University Press.
35.
VanderWeeleT. J.ChibaY. (2014). Sensitivity analysis for direct and indirect effects in the presence of exposure-induced mediator-outcome confounders. Epidemiology, Biostatistics, and Public Health, 11(2), pii: e9027.
VansteelandtS.VanderWeeleT. J. (2012). Natural direct and indirect effects on the exposed: Effect decomposition under weaker assumptions. Biometrics, 68(4), 1019–1027.
38.
YiendJ. (2010). The effects of emotion on attention: A review of attentional processing of emotional information. Cognition and Emotion, 24(1), 3–47.
39.
YuQ.MedeirosK. L.WuX.JensenR. E. (2018). Nonlinear predictive models for multiple mediation analysis: With an application to explore ethnic disparities in anxiety and depression among cancer survivors. Psychometrika, 83(4), 991–1006.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
