Hidden Data/Unobserved Evidence
Updated 2026-08-18
INTRODUCTION
English translation pending.
CORE DEFINITION
A class of statistical and inferential problems in which the observations available are a biased subset of the observations that matter. Classic forms include survivorship bias, where only successes remain visible, and attrition in trials, where patients who drop out differ systematically from those who stay. Abraham Wald's analysis of returning bomber aircraft is the standard illustration. The qualifier is that the danger depends on the missingness mechanism: data missing at random is manageable, data missing because of the outcome is not.
SCAFFOLDING EFFECT
Reduce cognitive load
- Missing-data audit: ask who or what is absent from the sample before drawing conclusions. - Mechanism check: determine whether the missingness is random or tied to the outcome. - Correction choice: apply a method that accounts for the missing, such as intention-to-treat.
Anchor fast decisions
If the probability of being observed depends on the value being measured, the observed sample is a biased estimator of the population. The bias does not shrink with a larger sample, because a bigger biased sample is still biased. Only a design or an analysis that models the missingness mechanism removes the distortion.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- imperial.ac.ukhttps://www.imperial.ac.uk/stories/dark-data-bright-light/verified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS