Proxy Variables and the Generalizability of Study Results

• Published in: American journal of epidemiology Vol. 192; no. 3; pp. 448 - 454
• Main Authors: Nilsson, Anton, Björk, Jonas, Bonander, Carl
• Format: Journal Article
• Language: English
• Published: United States Oxford University Press 24.02.2023; Oxford Publishing Limited (England)
• Subjects: Causality; Clinical outcomes; Data analysis; directed acyclic graphs; external validity; Folkhälsovetenskap, global hälsa och socialmedicin; generalizability; Health Sciences; Human subjects; Humans; Hälsovetenskap; Medical and Health Sciences; Medical research; Medicin och hälsovetenskap; Patient Selection; Proxy; proxy variables; Public Health, Global Health and Social Medicine; Risk and Environmental Studies; Risk- och miljöstudier; Standardization; Variables; generalizability; directed acyclic graphs; proxy variables; external validity
• ISSN: 0002-9262; 1476-6256; 1476-6256
• DOI: 10.1093/aje/kwac200

Abstract When individuals self-select (or are selected) into a study based on factors that influence the outcome, conclusions may not generalize to the full population. To compensate for this, results may be adjusted, for example, by standardization on the set of common causes of participation and outcome. Although such standardization is useful in some contexts, the common causes of participation and outcome may in practice not be fully observed. Instead, the researcher may have access to one or several variables related to the common causes, that is, to proxies for the common causes. This article defines and examines different types of proxy variables and shows how these can be used to obtain generalizable study results. First of all, the researcher may exploit proxies that influence only participation or outcome but which still allow for perfect generalizability by rendering participation and outcome conditionally independent. Further, generalizability can be achieved by leveraging 2 proxies, one of which is allowed to influence participation and one of which is allowed to influence the outcome, even if participation and outcome do not become independent conditional on these. Finally, approximate generalizability may be obtained by exploiting a single proxy that does not itself influence participation or outcome.