Errors in causal inference: an organizational schema for systematic error and random error

Etsuji Suzuki, Toshihide Tsuda, Toshiharu Mitsuhashi, Mohammad Ali Mansournia, Eiji Yamamoto

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Purpose To provide an organizational schema for systematic error and random error in estimating causal measures, aimed at clarifying the concept of errors from the perspective of causal inference. Methods We propose to divide systematic error into structural error and analytic error. With regard to random error, our schema shows its four major sources: nondeterministic counterfactuals, sampling variability, a mechanism that generates exposure events and measurement variability. Results Structural error is defined from the perspective of counterfactual reasoning and divided into nonexchangeability bias (which comprises confounding bias and selection bias) and measurement bias. Directed acyclic graphs are useful to illustrate this kind of error. Nonexchangeability bias implies a lack of “exchangeability” between the selected exposed and unexposed groups. A lack of exchangeability is not a primary concern of measurement bias, justifying its separation from confounding bias and selection bias. Many forms of analytic errors result from the small-sample properties of the estimator used and vanish asymptotically. Analytic error also results from wrong (misspecified) statistical models and inappropriate statistical methods. Conclusions Our organizational schema is helpful for understanding the relationship between systematic error and random error from a previously less investigated aspect, enabling us to better understand the relationship between accuracy, validity, and precision.

Original languageEnglish
Pages (from-to)788-793.e1
JournalAnnals of Epidemiology
Volume26
Issue number11
DOIs
Publication statusPublished - Nov 1 2016

Fingerprint

Selection Bias
Statistical Models

Keywords

  • Bias
  • Causality
  • Epidemiologic methods

ASJC Scopus subject areas

  • Epidemiology

Cite this

Errors in causal inference : an organizational schema for systematic error and random error. / Suzuki, Etsuji; Tsuda, Toshihide; Mitsuhashi, Toshiharu; Mansournia, Mohammad Ali; Yamamoto, Eiji.

In: Annals of Epidemiology, Vol. 26, No. 11, 01.11.2016, p. 788-793.e1.

Research output: Contribution to journalArticle

@article{b3685c32765641c59caca218f640f37a,
title = "Errors in causal inference: an organizational schema for systematic error and random error",
abstract = "Purpose To provide an organizational schema for systematic error and random error in estimating causal measures, aimed at clarifying the concept of errors from the perspective of causal inference. Methods We propose to divide systematic error into structural error and analytic error. With regard to random error, our schema shows its four major sources: nondeterministic counterfactuals, sampling variability, a mechanism that generates exposure events and measurement variability. Results Structural error is defined from the perspective of counterfactual reasoning and divided into nonexchangeability bias (which comprises confounding bias and selection bias) and measurement bias. Directed acyclic graphs are useful to illustrate this kind of error. Nonexchangeability bias implies a lack of “exchangeability” between the selected exposed and unexposed groups. A lack of exchangeability is not a primary concern of measurement bias, justifying its separation from confounding bias and selection bias. Many forms of analytic errors result from the small-sample properties of the estimator used and vanish asymptotically. Analytic error also results from wrong (misspecified) statistical models and inappropriate statistical methods. Conclusions Our organizational schema is helpful for understanding the relationship between systematic error and random error from a previously less investigated aspect, enabling us to better understand the relationship between accuracy, validity, and precision.",
keywords = "Bias, Causality, Epidemiologic methods",
author = "Etsuji Suzuki and Toshihide Tsuda and Toshiharu Mitsuhashi and Mansournia, {Mohammad Ali} and Eiji Yamamoto",
year = "2016",
month = "11",
day = "1",
doi = "10.1016/j.annepidem.2016.09.008",
language = "English",
volume = "26",
pages = "788--793.e1",
journal = "Annals of Epidemiology",
issn = "1047-2797",
publisher = "Elsevier Inc.",
number = "11",

}

TY - JOUR

T1 - Errors in causal inference

T2 - an organizational schema for systematic error and random error

AU - Suzuki, Etsuji

AU - Tsuda, Toshihide

AU - Mitsuhashi, Toshiharu

AU - Mansournia, Mohammad Ali

AU - Yamamoto, Eiji

PY - 2016/11/1

Y1 - 2016/11/1

N2 - Purpose To provide an organizational schema for systematic error and random error in estimating causal measures, aimed at clarifying the concept of errors from the perspective of causal inference. Methods We propose to divide systematic error into structural error and analytic error. With regard to random error, our schema shows its four major sources: nondeterministic counterfactuals, sampling variability, a mechanism that generates exposure events and measurement variability. Results Structural error is defined from the perspective of counterfactual reasoning and divided into nonexchangeability bias (which comprises confounding bias and selection bias) and measurement bias. Directed acyclic graphs are useful to illustrate this kind of error. Nonexchangeability bias implies a lack of “exchangeability” between the selected exposed and unexposed groups. A lack of exchangeability is not a primary concern of measurement bias, justifying its separation from confounding bias and selection bias. Many forms of analytic errors result from the small-sample properties of the estimator used and vanish asymptotically. Analytic error also results from wrong (misspecified) statistical models and inappropriate statistical methods. Conclusions Our organizational schema is helpful for understanding the relationship between systematic error and random error from a previously less investigated aspect, enabling us to better understand the relationship between accuracy, validity, and precision.

AB - Purpose To provide an organizational schema for systematic error and random error in estimating causal measures, aimed at clarifying the concept of errors from the perspective of causal inference. Methods We propose to divide systematic error into structural error and analytic error. With regard to random error, our schema shows its four major sources: nondeterministic counterfactuals, sampling variability, a mechanism that generates exposure events and measurement variability. Results Structural error is defined from the perspective of counterfactual reasoning and divided into nonexchangeability bias (which comprises confounding bias and selection bias) and measurement bias. Directed acyclic graphs are useful to illustrate this kind of error. Nonexchangeability bias implies a lack of “exchangeability” between the selected exposed and unexposed groups. A lack of exchangeability is not a primary concern of measurement bias, justifying its separation from confounding bias and selection bias. Many forms of analytic errors result from the small-sample properties of the estimator used and vanish asymptotically. Analytic error also results from wrong (misspecified) statistical models and inappropriate statistical methods. Conclusions Our organizational schema is helpful for understanding the relationship between systematic error and random error from a previously less investigated aspect, enabling us to better understand the relationship between accuracy, validity, and precision.

KW - Bias

KW - Causality

KW - Epidemiologic methods

UR - http://www.scopus.com/inward/record.url?scp=84994845864&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84994845864&partnerID=8YFLogxK

U2 - 10.1016/j.annepidem.2016.09.008

DO - 10.1016/j.annepidem.2016.09.008

M3 - Article

C2 - 27771142

AN - SCOPUS:84994845864

VL - 26

SP - 788-793.e1

JO - Annals of Epidemiology

JF - Annals of Epidemiology

SN - 1047-2797

IS - 11

ER -