Counterfactual graphical models for longitudinal mediation analysis with unobserved confounding

Ilya Shpitser

Research output: Contribution to journalArticle

Abstract

Questions concerning mediated causal effects are of great interest in psychology, cognitive science, medicine, social science, public health, and many other disciplines. For instance, about 60% of recent papers published in leading journals in social psychology contain at least one mediation test (Rucker, Preacher, Tormala, & Petty, 2011). Standard parametric approaches to mediation analysis employ regression models, and either the "difference method" (Judd & Kenny, 1981), more common in epidemiology, or the "product method" (Baron & Kenny, 1986), more common in the social sciences. In this article, we first discuss a known, but perhaps often unappreciated, fact that these parametric approaches are a special case of a general counterfactual framework for reasoning about causality first described by Neyman (1923) and Rubin (1924) and linked to causal graphical models by Robins (1986) and Pearl (2006). We then show a number of advantages of this framework. First, it makes the strong assumptions underlying mediation analysis explicit. Second, it avoids a number of problems present in the product and difference methods, such as biased estimates of effects in certain cases. Finally, we show the generality of this framework by proving a novel result which allows mediation analysis to be applied to longitudinal settings with unobserved confounders.

Original languageEnglish (US)
Pages (from-to)1011-1035
Number of pages25
JournalCognitive Science
Volume37
Issue number6
DOIs
StatePublished - Jan 1 2013
Externally publishedYes

Fingerprint

Social sciences
Social Sciences
Epidemiology
Public health
Regression analysis
Cognitive Science
Medicine
Social Psychology
Songbirds
Causality
Public Health
Regression Analysis
Psychology
Mediation
Causal

Keywords

  • Causal inference
  • Counterfactuals
  • Direct and indirect effects
  • Graphical models
  • Longitudinal studies
  • Mediation analysis
  • Path-specific effects

ASJC Scopus subject areas

  • Language and Linguistics
  • Experimental and Cognitive Psychology
  • Cognitive Neuroscience
  • Artificial Intelligence

Cite this

Counterfactual graphical models for longitudinal mediation analysis with unobserved confounding. / Shpitser, Ilya.

In: Cognitive Science, Vol. 37, No. 6, 01.01.2013, p. 1011-1035.

Research output: Contribution to journalArticle

@article{8c73a70cfec446b4a523eb688b0edcaf,
title = "Counterfactual graphical models for longitudinal mediation analysis with unobserved confounding",
abstract = "Questions concerning mediated causal effects are of great interest in psychology, cognitive science, medicine, social science, public health, and many other disciplines. For instance, about 60{\%} of recent papers published in leading journals in social psychology contain at least one mediation test (Rucker, Preacher, Tormala, & Petty, 2011). Standard parametric approaches to mediation analysis employ regression models, and either the {"}difference method{"} (Judd & Kenny, 1981), more common in epidemiology, or the {"}product method{"} (Baron & Kenny, 1986), more common in the social sciences. In this article, we first discuss a known, but perhaps often unappreciated, fact that these parametric approaches are a special case of a general counterfactual framework for reasoning about causality first described by Neyman (1923) and Rubin (1924) and linked to causal graphical models by Robins (1986) and Pearl (2006). We then show a number of advantages of this framework. First, it makes the strong assumptions underlying mediation analysis explicit. Second, it avoids a number of problems present in the product and difference methods, such as biased estimates of effects in certain cases. Finally, we show the generality of this framework by proving a novel result which allows mediation analysis to be applied to longitudinal settings with unobserved confounders.",
keywords = "Causal inference, Counterfactuals, Direct and indirect effects, Graphical models, Longitudinal studies, Mediation analysis, Path-specific effects",
author = "Ilya Shpitser",
year = "2013",
month = "1",
day = "1",
doi = "10.1111/cogs.12058",
language = "English (US)",
volume = "37",
pages = "1011--1035",
journal = "Cognitive Science",
issn = "0364-0213",
publisher = "Wiley-Blackwell",
number = "6",

}

TY - JOUR

T1 - Counterfactual graphical models for longitudinal mediation analysis with unobserved confounding

AU - Shpitser, Ilya

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Questions concerning mediated causal effects are of great interest in psychology, cognitive science, medicine, social science, public health, and many other disciplines. For instance, about 60% of recent papers published in leading journals in social psychology contain at least one mediation test (Rucker, Preacher, Tormala, & Petty, 2011). Standard parametric approaches to mediation analysis employ regression models, and either the "difference method" (Judd & Kenny, 1981), more common in epidemiology, or the "product method" (Baron & Kenny, 1986), more common in the social sciences. In this article, we first discuss a known, but perhaps often unappreciated, fact that these parametric approaches are a special case of a general counterfactual framework for reasoning about causality first described by Neyman (1923) and Rubin (1924) and linked to causal graphical models by Robins (1986) and Pearl (2006). We then show a number of advantages of this framework. First, it makes the strong assumptions underlying mediation analysis explicit. Second, it avoids a number of problems present in the product and difference methods, such as biased estimates of effects in certain cases. Finally, we show the generality of this framework by proving a novel result which allows mediation analysis to be applied to longitudinal settings with unobserved confounders.

AB - Questions concerning mediated causal effects are of great interest in psychology, cognitive science, medicine, social science, public health, and many other disciplines. For instance, about 60% of recent papers published in leading journals in social psychology contain at least one mediation test (Rucker, Preacher, Tormala, & Petty, 2011). Standard parametric approaches to mediation analysis employ regression models, and either the "difference method" (Judd & Kenny, 1981), more common in epidemiology, or the "product method" (Baron & Kenny, 1986), more common in the social sciences. In this article, we first discuss a known, but perhaps often unappreciated, fact that these parametric approaches are a special case of a general counterfactual framework for reasoning about causality first described by Neyman (1923) and Rubin (1924) and linked to causal graphical models by Robins (1986) and Pearl (2006). We then show a number of advantages of this framework. First, it makes the strong assumptions underlying mediation analysis explicit. Second, it avoids a number of problems present in the product and difference methods, such as biased estimates of effects in certain cases. Finally, we show the generality of this framework by proving a novel result which allows mediation analysis to be applied to longitudinal settings with unobserved confounders.

KW - Causal inference

KW - Counterfactuals

KW - Direct and indirect effects

KW - Graphical models

KW - Longitudinal studies

KW - Mediation analysis

KW - Path-specific effects

UR - http://www.scopus.com/inward/record.url?scp=84881548526&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84881548526&partnerID=8YFLogxK

U2 - 10.1111/cogs.12058

DO - 10.1111/cogs.12058

M3 - Article

VL - 37

SP - 1011

EP - 1035

JO - Cognitive Science

JF - Cognitive Science

SN - 0364-0213

IS - 6

ER -