CoChat
Sign Up
Cited by 5,293Advanced Causal Inference TechniquesPhilosophy and History of ScienceBayesian Modeling and Causal Inference

Statistics and Causal Inference

Holland, Paul W.·1986·Journal of the American Statistical Association
DOI 10.1080/01621459.1986.10478354
Open this paper in CoChat — free account

Abstract

Problems involving causal inference have dogged at the heels of statistics since its earliest days. Correlation does not imply causation, and yet causal conclusions drawn from a carefully designed experiment are often valid. What can a statistical model say about causation? This question is addressed by using a particular model for causal inference (Holland and Rubin 1983; Rubin 1974) to critique the discussions of other writers on causation and causal inference. These include selected philosophers, medical researchers, statisticians, econometricians, and proponents of causal modeling.

Integrity signals

Registry

Crossref · 10.1080/01621459.1986.10478354

Cites

62 of 62 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Due to the fundamental problem of causal inference, only one of the two potential outcomes is observed for each unit.

    “As is usually the case, due to the fundamental problem of causal inference (Holland, 1986), we only observe one of the two potential outcomes for each unit i .”

    From What's trending in difference-in-differences? A synthesis of the recent econometrics literature (2023) · cites this paper as Holland, 1986

  • Causal effects cannot be observed directly.

    “causal effects can never be observed directly”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Holland 1986

  • In uplift modeling, the actual outcome that is estimated is unobserved, so the error made by the model cannot be measured at the individual customer level.

    “In uplift modeling, however, the actual outcome that is estimated is unobserved, and therefore the error made by the model cannot be measured at the individual customer level.”

    From Why you should stop predicting customer churn and start using uplift models (2021) · cites this paper as [34]

  • Historical data containing information on observed customer behavior never allows classifying customers in the four groups shown in Table 1, as the net effect of treatment cannot be observed for an individual customer due to the fundamental problem of causal inference.

    “Historical data containing information on observed customer behavior never allows classifying customers in the four groups shown in Table 1, as the net e ↵ ect of treatment cannot be observed for an individual customer due to the fundamental problem of causal inference [34].”

    From Why you should stop predicting customer churn and start using uplift models (2021) · cites this paper as [34]

  • Uplift, or whether a customer is persuadable, a lost cause, do-not-disturb or a sure thing, cannot be observed due to the fundamental problem of causal inference.

    “Uplift or, in other words, whether a customer is persuadable, a lost cause, do-notdisturb or a sure thing, cannot be observed due to the fundamental problem of causal inference [34].”

    From Why you should stop predicting customer churn and start using uplift models (2021) · cites this paper as [34]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Holland, P. W.
Affiliations
Educational Testing Service (US)
Corresponding author
Holland, Paul W.
ORCID
Holland, Paul W.
Venue
Journal of the American Statistical Association
Publisher
Informa UK Limited
Published
1986-12
Volume
81
Issue
396
Pages
945-960
ISSN
  • 0162-1459 (ISSN-L)
  • 1537-274X
DOI
10.1080/01621459.1986.10478354
OpenAlex
W1978108654
Cited by
5,293
References
62
Citations by year
2026: 188 · 2025: 260 · 2024: 246 · 2023: 266 · 2022: 274 · 2021: 323 · 2020: 302 · 2019: 269 · 2018: 259 · 2017: 202 · 2016: 212 · 2015: 229 · 2014: 271 · 2013: 224 · 2012: 187
FWCI
14.28
Type
journal article
Language
en
Open access
closed
Locations
Journal of the American Statistical Association (publishedVersion)
Topics
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
  • Social Sciences › Arts and Humanities › History and Philosophy of Science › Philosophy and History of Science
  • Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Modeling and Causal Inference
Keywords
causal inference, causal models, correlation versus causation, randomized experiments, counterfactual reasoning, philosophy of statistics

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review