MENTAL MODEL · M9096

Collision Factor

TechnicalHigh supportStatistics

Included

Updated 2026-08-15

Loading revision record…

INTRODUCTION

English translation pending.

CORE DEFINITION

In a causal graph, a collider is a variable that two others both influence, so the arrows point into it. Conditioning on a collider, for example by studying only cases where it occurs, opens a path between the two causes and induces a correlation that does not exist in the population. The qualification is that this bias arises from selection or conditioning rather than from the underlying relationship, so the correlation is an artefact of the analysis.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

- Use Graph Drawing: mark which variable is caused by both others before conditioning on anything. - Use Selection Audit: ask whether your sample was filtered on an outcome that several causes produce.

anchor

Anchor fast decisions

When a sample includes only cases where the collider occurred, high values on one cause must be offset by low values on the other for the collider threshold to be met. That compensation produces a negative association between the two causes inside the sample even when none exists outside it.

MINIMUM ACTION

In progress 0/1

Practice this model in one real situation:

Draw the causal graph for the variables involved. 2. Identify any node that two others both influence. 3. Check whether your sample was filtered on that node. 4. Recognise that correlations within the filtered sample are induced. 5. Avoid conditioning on the collider unless the question requires it.

Check to track your progress (stored locally)

Learning progress0%

account_treeGenealogyexpand_more

menu_bookReferencesexpand_more

Source support: Explicit

link
en.wikipedia.orghttps://en.wikipedia.org/wiki/Collider_(statisticsZH · Explicit
verified

RELATED MODELS