Objection Overruled: A Comment on Gastwirth, Krieger, and Rosenbaum

Jim Mintz, Wilfrid Dixon

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Gastwirth, Krieger, and Rosenbaum used statistical evidence from a 1987 sex discrimination case in a Canadian court as the basis for a November 1994 article in The American Statistician on the “third variable” problem in correlational analysis. We reexamine their unsettling conclusion that the Court “accepted an impossible explanation.” A revised explanatory model that includes an unmeasured latent variable leads us to a very different conclusion. The result demonstrates that references to unmeasured covariates as “third variables” in analyses of spurious association can be quite difficult or even impossible to evaluate statistically. Judging such models may depend ultimately on assessing their plausibility rather than their statistical possibility.

Original languageEnglish (US)
Pages (from-to)117-119
Number of pages3
JournalAmerican Statistician
Volume51
Issue number2
DOIs
StatePublished - May 1997
Externally publishedYes

Keywords

  • Law and statistics
  • Plausibility
  • Unmeasured third variable
  • Unobserved third variable

ASJC Scopus subject areas

  • Statistics and Probability
  • Mathematics(all)
  • Statistics, Probability and Uncertainty

Fingerprint Dive into the research topics of 'Objection Overruled: A Comment on Gastwirth, Krieger, and Rosenbaum'. Together they form a unique fingerprint.

Cite this