Skip Navigation LinksHome > September 2011 - Volume 22 - Issue 5 > Transportability and Causal Generalization
Epidemiology:
doi: 10.1097/EDE.0b013e3182254b8f
Letters

Transportability and Causal Generalization

Schwartz, Sharon; Gatto, Nicolle M.; Campbell, Ulka B.

Free Access
Article Outline
Collapse Box

Author Information

Department of Epidemiology, Columbia University, Mailman School of Public Health, New York, NY, sbs5@columbi.edu (Schwartz)

Department of Epidemiology, Columbia University, Mailman School of Public Health, New York, NY, Epidemiology, Worldwide Safety Strategy, Pfizer Inc., New York, NY (Gatto, Campbell)

Back to Top | Article Outline

To the Editor:

We read with interest the paper by Hernán and Vanderweele “On compound treatments and transportability of causal inference.”1 They note 3 factors that influence the transportability (ie, generalizability) of the causal effect identified in a study—compound treatments, interference, and effect modification.

We think that consideration of transportability is long overdue. One limitation of the potential outcomes perspective is, in our view, a tendency to reify the causal effect estimated in a study as an actual intervention effect.2 The relationship between the 2 needs explicit consideration, but the lack of direct correspondence does not necessarily imply an “ill-defined” causal question, as suggested in much of the potential outcomes work in epidemiology.

Hernan and Vanderweele use new terminology to make points that have a long history in the Cook and Campbell tradition in psychology (eg, Shadish et al3). In that work, causal questions are divided into 2 broad categories: causal description and causal explanation. Causal description is about identifying what the causal effect was in a particular study. The focus is on internal validity. Causal explanation is about the transportability of effects. Shadish et al3 describe 2 issues in causal explanation: construct validity and external validity. Construct validity is the extent to which the treatment effect can be generalized beyond the particular operationalization in the study. In their framework, construct validity examines treatment variation irrelevance—the influence of various treatment versions on the causal effect. External validity is about the extent to which the treatment effect holds across various characteristics of persons and settings, addressing the issues of interference between units and effect modification.

The Cook and Campbell tradition shares with potential outcomes a counterfactual frame, although one starting from the perspective of Mackie rather than Lewis, who is favored in the potential outcomes literature. Shadish et al note the relationship between their view and the potential outcomes view (which they refer to as “Rubin's causal model”):

“Rubin's model is not intended to say much about the matters of casual generalization that we address in this book.”3 p.6

We have argued that integrating these aspects of the Cook and Campbell tradition into epidemiologic discussions of potential outcomes would be useful.2,4 We are glad to see a broader recognition of the problems of generalizability in the potential outcomes framework, which has been so illuminating for the explication of internal validity.

Sharon Schwartz

Department of Epidemiology

Columbia University

Mailman School of Public Health

New York, NY

sbs5@columbi.edu

Nicolle M. Gatto

Ulka B. Campbell

Department of Epidemiology

Columbia University

Mailman School of Public Health

New York, NY

Epidemiology

Worldwide Safety Strategy

Pfizer Inc.

New York, NY

Back to Top | Article Outline

REFERENCES

1. Hernan M, Vanderweele T. On compound treatments and transportability of causal inference. Epidemiology. 2011;22:368.

2. Schwartz S, Gatto N, Campbell U. What would have been is not what would be: counterfactuals of the past and potential outcomes of the future. In: Shrout PE, Keyes KM, Ornstein K, eds. Causality and Psychopathology: Finding the Determinants of Disorders and Their Cures. New York: Oxford; 2011:25–46.

3. Shadish WR, Cook TD, Campbell DT. Experimental and Quasi-Experimental Designs for Generalized Casual Inference. Boston: Houghton Mifflin; 2002.

4. Susser E, Schwartz S, Morabia A, Bromet EJ. Psychiatric Epidemiology: Searching for the Causes of Mental Disorders. New York: Oxford; 2006.

© 2011 Lippincott Williams & Wilkins, Inc.

Twitter  Facebook

Login

Article Tools

Share