Hostname: page-component-7dd5485656-dk7s8 Total loading time: 0 Render date: 2025-10-31T05:30:33.562Z Has data issue: false hasContentIssue false
  • English
  • Français

Causal interaction and effect modification: same model, different concepts

Published online by Cambridge University Press:  21 April 2020

Luke Keele Open the ORCID record for Luke Keele [Opens in a new window]  and
Randolph T. Stevenson
Luke Keele*
Affiliation:
University of Pennsylvania, Philadelphia, PA 19104, USA
Randolph T. Stevenson
Affiliation:
Rice University, Houston, TX 77251, USA
*
*Corresponding author. Email: luke.keele@gmail.com
Get access Rights & Permissions [Opens in a new window]

Abstract

Social scientists use the concept of interactions to study effect dependency. In the causal inference literature, interaction terms may be used in two distinct type of analysis. The first type of analysis focuses on causal interactions, where the analyst is interested in whether two treatments have differing effects when both are administered. The second type of analysis focuses on effect modification, where the analyst investigates whether the effect of a single treatment varies across levels of a baseline covariate. While both forms of interaction analysis are typically conducted using the same type of statistical model, the identification assumptions for these two types of analysis are very different. In this paper, we clarify the difference between these two types of interaction analysis. We demonstrate that this distinction is mostly ignored in the political science literature. We conclude with a review of several applications where we show that the form of the interaction is critical to proper interpretation of empirical results.

Keywords

Causal inferenceInteractioneffect modification

Information

Type
Original Article
Information
Political Science Research and Methods , Volume 9 , Issue 3 , July 2021 , pp. 641 - 649
Copyright
Copyright © The European Political Science Association 2020

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

Acharya, A, Blackwell, M and Sen, M (2016) Explaining causal findings without bias: detecting and assessing direct effects. American Political Science Review 110, 512–529.CrossRefGoogle Scholar
Berry, WD, Golder, M and Milton, D (2012) Improving tests of theories positing interaction. Journal of Politics 74, 653–671.CrossRefGoogle Scholar
Brader, T, Valentino, NA and Suhat, E (2008) What triggers public opposition to immigration? Anxiety, group cues, and immigration. American Journal of Political Science 52, 959–978.CrossRefGoogle Scholar
Brambor, T, Clark, WR and Golder, M (2006) Understanding interaction models: improving empirical analyses. Political Analysis 14, 63–82.CrossRefGoogle Scholar
Braumoeller, BF (2004) Hypothesis testing and multiplicative interaction terms. International Organization 58, 807–820.CrossRefGoogle Scholar
Clark, W, Gilligan, M and Golder, M (2006) A simple multivariate test for asymmetric hypotheses. Political Analysis 14, 311–331.CrossRefGoogle Scholar
Elwert, F. (2013) Graphical causal models. In Stephen L., Morgan (ed.). Handbook of Causal Analysis for Social Research. Amsterdam: Springer, pp. 245–273.CrossRefGoogle Scholar
Franzese, R and Kam, C (2009) Modeling and Interpreting Interactive Hypotheses in Regression Analysis. Ann Arbor, MI: University of Michigan Press.Google Scholar
Hainmueller, J, Mummolo, J and Xu, Y (2019) How much should we trust estimates from multiplicative interaction models? Simple tools to improve empirical practice. Political Analysis 27, 163–192.CrossRefGoogle Scholar
Kim, HA and LeVeck, BL (2013) Money, reputation, and incumbency in US house elections, or why marginals have become more expensive. American Political Science Review 107, 492–504.CrossRefGoogle Scholar
Montgomery, JM, Nyhan, B and Torres, M (2018) How conditioning on posttreatment variables can ruin your experiment and what to do about it. American Journal of Political Science 62, 760–775.CrossRefGoogle Scholar
Morgan, SL and Winship, C (2014) Counterfactuals and Causal Inference: Methods and Principles for Social Research, 2nd Ed. New York, NY: Cambridge University Press.CrossRefGoogle Scholar
Pearl, J (2009) Causal inference in statistics: an overview. Statistics Surveys 3, 96–146.CrossRefGoogle Scholar
Rosenbaum, PR (1984) The consequences of adjusting for a concomitant variable that has been affected by the treatment. Journal of The Royal Statistical Society Series A 147, 656–666.CrossRefGoogle Scholar
Stephens, A, Keele, LJ and Joffe, M (2016) Estimating post-treatment effect modification with generalized structural mean models. Journal of Causal Inference. in press, unpublished Manuscript.Google Scholar
Truex, R (2014) The returns to office in a “rubber stamp” parliament. American Political Science Review 108, 235–251.CrossRefGoogle Scholar
VanderWeele, T (2015) Explanation in Causal Inference: Methods for Mediation and Interaction. New York, NY: Oxford University Press.Google Scholar
VanderWeele, TJ and Robins, JM (2007) The identification of synergism in the sufficient-component-cause framework. Epidemiology 18, 329–339.CrossRefGoogle ScholarPubMed
VanderWeele, TJ and Robins, JM (2008) Empirical and counterfactual conditions for sufficient cause interactions. Biometrika 95, 49–61.CrossRefGoogle Scholar