Lord's Paradox
Updated 2026-08-01
INTRODUCTION
English translation pending.
CORE DEFINITION
Described by statistician Frederic Lord. Two analysts studying the same dataset, one comparing group means adjusted for baseline and the other examining individual change, can reach opposite conclusions about whether a treatment had an effect. Both analyses are technically correct; they answer different questions because they condition on different variables. The paradox shows that the choice of adjustment determines the conclusion, and that no analysis is uniquely correct without a stated causal question.
SCAFFOLDING EFFECT
Reduce cognitive load
- Fix the question: state the causal question before choosing which variables to adjust for - Compare models: run both the between-group and within-group analyses and explain the divergence - Read studies: check what a paper conditioned on, since that choice produced its headline result
Anchor fast decisions
Adjusting for a baseline variable answers a question about relative position, while analyzing change answers a question about within-person movement. When the variable lies on the causal pathway or is affected by selection, the two conditioning sets imply different comparisons, so the estimates diverge. The data are identical in both cases; only the question and the adjustment differ, so causal structure must decide between them.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Lord's_paradoxverified
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