On Identifying Total Effects in the Presence of Latent Variables and Selection bias

By Manabu Kuroki, Zhihong Cai. 2008

48 citations recorded by the source corpus.

Assume that cause-effect relationships between variables can be described as a directed acyclic graph and the corresponding linear structural equation model.We consider the identification problem of total effects in the presence of latent variables and selection bias between a treatment variable and a response variable. Pearl and his colleagues provided the back door criterion, the front door criterion (Pearl, 2000) and the conditional instrumental variable method (Brito and Pearl, 2002) as identifiability criteria for total effects in the presence of latent variables, but not in the presence of selection bias. In order to solve this problem, we propose new graphical identifiability criteria for total effects based on the identifiable factor models. The results of this paper are useful to identify total effects in observational studies and provide a new viewpoint to the identification conditions of factor models.

Subfields: Artificial Intelligence, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Causal Effect Inference with Deep Latent-Variable Models (2017, neutral): “Recent work by Cai and Kuroki [9], Greenland and Lash [18], building on the work of Greenland and Kleinbaum [17], Selén [47], has studied conditions for causal identiﬁability using proxy variables.”
3. The Foundations of Causal Inference (2010, methodology): “Graph-based methods for effect identification under measurement errors are discussed in (Pearl, 2010c; Hernán and Cole, 2009; Cai and Kuroki, 2008).”
Policy Evaluation with Latent Confounders via Optimal Balance (2019, background): “In particular, there is a variety of work that studies sufficient conditions for the identifiability of latent confounder models [6, 34, 37].”
On Measurement Bias in Causal Inference (2010, background): “…identiﬁable (up to a constant var ( Z )) whenever each latent variable (in our case Z ) has three independent proxies (in our case X, W , and V ) 6 Cai and Kuroki (2008) and Kuroki and Pearl (2010) further showed that c 0 is identiﬁable from measurements of three proxies (of Z ), even when these…”
The Causal Foundations of Structural Equation Modeling (2012, background): “3Causal relationships among latent variables are assessed by treating their indicators as noisy measurement of the formers (Bollen, 1989; Pearl, 2010c; Cai and Kuroki, 2008).”
An Extended Class of Instrumental Variables for the Estimation of Causal Effects (2011, background): “Other possibilities for identification arise when further assumptions such as normality are imposed, as in Cai and Kuroki (2008), for example.”