Resources for Causal Reasoning in Health Services Research

Recently Added

See All

Sign-up for Updates

To receive an email when references are updated, please fill out this form.

Privacy Policy.

This policy covers our treatment of personal information when you register with the website. We use the name and email address you provide only for letting you know that new references were added to this website. We will not sell this information to other people or companies. You will be able to stop you registration with the website at any time when you request it.

Glossary of Terms

Risk
Definition
Risk

probability of an event during a specified period of time

d-separation
Definition
d-separation

variables X and Y are said to be d-separated, if all paths from X to Y are blocked. A path may be blocked by a set of variables S, if the path contains at least one arrow-emitting variable that is in S or if the path contains at least one collider that is not in S and has no descendent in S. A collider is defined as a variable on a path such that the path enters and leaves via arrow heads. X and Y are, therefore, d-separated conditional on S, if S blocks all paths between X and Y. X and Y are marginally independent when S contains the null set

Direct effect
Definition
Direct effect

the expected change in outcome in response to changing exposure while holding all other variables fixed

Instrumental variable
Definition
Instrumental variable

a variable relative to the total effect of exposure on outcome if there is a set of measurements, unaffected by the exposure, such the instrument is conditionally independent of the outcome given the set and the instrument is not independent of the exposure conditional on the set

See All