Interpretable Conditions for Identifying Direct and Indirect Effects
Abstract
This paper translates the conditions necessary for the identification of natural direct and indirect effects into a transparent language, thus permitting a more informed judgment of the plausibility of these conditions. We show that the conditions usually cited in the literature are overly restricted, and can be relaxed substantially, without compromising identification. In particular, we show that natural effects can be identified by methods other than adjustment. The identification conditions can be further relaxed in parametric models with interactions, and permit us to compare the relative importance of several pathways, mediated by interdependent variables.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 07, 2012
- Accession Number
- ADA564093
Entities
People
- Judea Pearl
Organizations
- University of California, Los Angeles