Pré-Publication, Document De Travail Année : 2025

When Numbers Mislead Us

Résumé

Believing that there is a single, objective way to describe phenomena using numbers is to forget that data does not "speak" for itself. Collecting data involves making choices: what to measure, how, when, on whom, etc. This implies implicit (or even ideological) assumptions about what counts as a measurable fact. And in any data analysis, what is not measured can be as important as what is observed. When an influential variable is overlooked-whether ignored, neglected, or simply unknown-the apparent relationships between other variables can become misleading. This is known as "omitted variable bias": a hidden effect distorts comparisons and can make a correlation appear where there is none, or mask a real one. Sometimes, introducing this "forgotten" variable can even completely reverse the conclusions that would have been drawn from a naive reading of the data.

Fichier principal
Vignette du fichier
numbers_AC_2025.pdf (638.86 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05149008 , version 1 (07-07-2025)

Licence

Identifiants

  • HAL Id : hal-05149008 , version 1

Citer

Arthur Charpentier. When Numbers Mislead Us. 2025. ⟨hal-05149008⟩
34 Consultations
55 Téléchargements

Partager

  • More