We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Efficient least squares for estimating total effects under linearity and causal sufficiency

Abstract: Recursive linear structural equation models are widely used to postulate causal mechanisms underlying observational data. In these models, each variable equals a linear combination of a subset of the remaining variables plus an error term. When there is no unobserved confounding or selection bias, the error terms are assumed to be independent. We consider estimating a total causal effect in this setting. The causal structure is assumed to be known only up to a maximally oriented partially directed acyclic graph (MPDAG), a general class of graphs that can represent a Markov equivalence class of directed acyclic graphs (DAGs) with added background knowledge. We propose a simple estimator based on recursive least squares, which can consistently estimate any identified total causal effect, under point or joint intervention. We show that this estimator is the most efficient among all regular estimators that are based on the sample covariance, which includes covariate adjustment and the estimators employed by the joint-IDA algorithm. Notably, our result holds without assuming Gaussian errors.
Comments: Minor edits
Subjects: Statistics Theory (math.ST); Methodology (stat.ME)
Cite as: arXiv:2008.03481 [math.ST]
  (or arXiv:2008.03481v4 [math.ST] for this version)

Submission history

From: F. Richard Guo [view email]
[v1] Sat, 8 Aug 2020 09:10:23 GMT (713kb,D)
[v2] Fri, 28 Aug 2020 02:05:47 GMT (570kb,D)
[v3] Fri, 5 Mar 2021 00:11:52 GMT (535kb,D)
[v4] Thu, 17 Mar 2022 21:58:08 GMT (536kb,D)

Link back to: arXiv, form interface, contact.