On Simpson's Paradox and the Sure-Thing Principle

By C. Blyth. 1972

Subfields: Statistics and Probability, Game & Decision Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

In Game & Decision Theory

Citations in the corpus

A Survey on Bias and Fairness in Machine Learning (2019, background): “Simpson’s paradox [17] can bias the analysis of heterogeneous data that is composed of subgroups or individuals with different behaviors.”
Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries (2019, background): “, the presence of a keyword or hashtag), and there are multiple independent factors that result in the attribute taking a certain value, analyzing data entries with this value is equivalent to conditioning on it, and may result in spurious patterns of association among these factors (Blyth, 1972; Tu”
Using Simpson's Paradox to Discover Interesting Patterns in Behavioral Data (2018, methodology): “Simpson’s paradox (Blyth 1972; Norton and Divine 2015) often confounds analysis of heterogeneous data, especially social and behavioral data.”
Causal inference in statistics: An overview (2009, background): “(Simpson, 1951; Blyth, 1972; Cox and Wermuth, 2003) lamented by statisticians (Good and Mittal, 1987; Bishop et al.”
Using Loglinear Model for Discrimination Discovery and Prevention (2016, neutral): “This corresponds to the well known Simpson’s paradox phenomenon [23].”
The Bouncer Problem: Challenges to Remote Explainability (2019, background): “sing models that may include forms of bias [27], the automatic detection of bias on the user side is also of interest for the community. For instance, researchers seek to detect the Simpson’s Paradox [8] in the data [4]. Another work makes use of causal graphs to detect [41] a potential discrimina11”
Toward Deconfounding the Effect of Entity Demographics for Question Answering Accuracy (2021, background): “The lack of representation in QA could cause us to think that things are better than they are because of Simpson’s paradox (Blyth, 1972): gender and profession are not independent!”
A critique of the American Psychological Association Clinical Practice Guideline for the Treatment of Posttraumatic Stress Disorder (PTSD) in Adults. (2020, background): “As Ziliak and Teather-Posadas (2014) articulate, Simpson’s paradox (Blyth, 1972; Simpson & Dixon, 1951) remains a risk in RCTs’ studies.”
Simplicity Creates Inequity: Implications for Fairness, Stereotypes, and Interpretability (2018, unknown): “f Dcomes ahead of the corresponding cell consisting of rows of A, and this is why gis not strictly improvable. The surprising property that makes this example work is an instance of Simpson’s Paradox [5], which roughly speaking describes situations in which the unconditional mean over one population”
Comment: Understanding Simpson’s Paradox (2014, background): “…could have causal roots.1 In particular, the word “causal” does not appear in Simpson’s article, nor in the vast literature that followed, including Blyth (1972), who coined the term “paradox,” and the influential writings of Agresti (1983), Bishop, Fienberg, and Holland (1975), and Whittemore…”
A Survey on Causal Inference (2020, background): “s effect of confounder variableAge. From Table 1, we can observe another interesting phenomenon, Simpson’s paradox (or Simpson’s reversal, Yule-Simpson effect, amalgamation paradox, reversal paradox) [18, 42], brought by the confounder. It can be observed that: in both Young and Older patient groups”