Understanding Simpson's Paradox

@article{Pearl2013UnderstandingSP,
  title={Understanding Simpson's Paradox},
  author={Judea Pearl},
  journal={ERN: Other Econometrics: Econometric \& Statistical Methods (Topic)},
  year={2013}
}
  • J. Pearl
  • Published 19 September 2013
  • Philosophy
  • ERN: Other Econometrics: Econometric & Statistical Methods (Topic)
Simpson's paradox is often presented as a compelling demonstration of why we need statistics education in our schools. It is a reminder of how easy it is to fall into a web of paradoxical conclusions when relying solely on intuition, unaided by rigorous statistical methods. In recent years, ironically, the paradox assumed an added dimension when educators began using it to demonstrate the limits of statistical methods, and why causal, rather than statistical considerations are necessary to… 
Simpson's paradox in epistemology and decision theory
I discuss the implications of Simpson’s paradox for epistemology and decision theory. In Chapter One I outline the paradox, focussing on its identification, nature, and type of reasoning that it
Lord ’ s Paradox Revisited – (
Among the many peculiarities that were dubbed “paradoxes” by well meaning statisticians, the one reported by Frederic M. Lord in 1967 has earned a special status. Although it can be viewed, formally,
Lord ’ s Paradox Revisited – ( Oh Lord ! Kumbaya ! ) Judea Pearl
Among the many peculiarities that were dubbed “paradoxes” by well meaning statisticians, the one reported by Frederic M. Lord in 1967 has earned a special status. Although it can be viewed, formally,
Probability Reversal and the Disjunction Effect in Reasoning Systems
TLDR
It is argued that evidence supports that beliefs are not classical statistical variables and they should, in the general case, be considered as superposition states of disjoint or polar outcomes.
The Trouble With Binaries: A Perspective on the Science of Reading
In this article, we critique the science of reading when it is positioned within the reading wars as settling disagreements about reading and how it should be taught. We frame our argument in terms
Addressing Methodological Challenges in Culture-Comparative Research
We address methodological challenges in cross-cultural and cultural psychology. First, we describe weaknesses in (quasi-)experimental designs, noting that cross-cultural designs typically do not
Allocation to groups: Examples of Lord's paradox.
  • D. Wright
  • Education
    The British journal of educational psychology
  • 2019
TLDR
It is shown that a common method used to measure school effectiveness is biased against schools that serve students from groups that are historically poor performing, and the ANCOVA is preferred, but otherwise the gain score model may be appropriate.
Simpson's Paradox in COVID-19 Case Fatality Rates: A Mediation Analysis of Age-Related Causal Effects
TLDR
Causal inference, in particular mediation analysis, can be used to resolve apparent statistical paradoxes; help educate the public and decision-makers alike; avoid unsound comparisons; and answer a range of causal questions pertaining to the pandemic, subject to transparently stated assumptions.
A Bayesian Solution to the M-Bias Problem
TLDR
The fact that this work is able to offer a solution to the M-Bias problem in Bayesian terms shows that it is indeed possible to represent causal relationships within the Bayesian paradigm, albeit in an extended space.
Reply to commentary by Imai, Keele, Tingley, and Yamamoto concerning causal mediation analysis.
  • J. Pearl
  • Economics
    Psychological methods
  • 2014
TLDR
The comment explains the key difference between traditional and modern methods of causal mediation, and demonstrates why the notion of mediation requires counterfactual rather than Bayes conditionals to be properly defined.
...
...

References

SHOWING 1-10 OF 30 REFERENCES
The Simpson's paradox unraveled.
TLDR
It is shown how the paradox disappears when the statistical analysis is appropriately guided by subject-matter knowledge and the problem is stripped of its causal context and analyzed merely in statistical terms.
On Simpson's Paradox and the Sure-Thing Principle
Abstract This paradox is the possibility of P(A|B) <P(A|B') even though P(A|B)≥P(A| B') both under the additional condition C and under the complement C' of that condition. Details are given on why
An Introduction to Logic and Scientific Method
THIS book, by two American teachers of philosophy, helps to show how far the best representatives of logic in modern universities have travelled from the Aristotelian tradition which formerly
Discrete Multivariate Analysis: Theory and Practice
TLDR
Discrete Multivariate Analysis is a comprehensive text and general reference on the analysis of discrete multivariate data, particularly in the form of multidimensional tables, and contains a wealth of material on important topics.
NOTES ON THE THEORY OF ASSOCIATION OF ATTRIBUTES IN STATISTICS
THE simplest possible form of statistical classification is "division" (as the logicians term it) "by dichotomy," i.e. the sorting of the objects or individuals observed into one or other of two
Conditioning and Intervening
We consider the dispute between causal decision theorists and evidential decision theorists over Newcomb-like problems. We introduce a framework relating causation and directed graphs developed by
Identification, Inference and Sensitivity Analysis for Causal Mediation Effects
Causal mediation analysis is routinely conducted by applied researchers in a variety of disciplines. The goal of such an analysis is to investigate alternative causal mechanisms by examining the
All of Statistics: A Concise Course in Statistical Inference
TLDR
This book covers a much wider range of topics than a typical introductory text on mathematical statistics, and includes modern topics like nonparametric curve estimation, bootstrapping and classification, topics that are usually relegated to follow-up courses.
Causal diagrams for empirical research
SUMMARY The primary aim of this paper is to show how graphical models can be used as a mathematical language for integrating statistical and subject-matter information. In particular, the paper
Confounding Equivalence in Causal Inference
TLDR
The test for deciding, from a given causal diagram, whether two sets of variables have the same bias-reducing potential under adjustment is extended to include treatment-dependent covariates by broadening the back-door criterion and establishing equivalence of adjustment under selection bias conditions.
...
...