Exact variance formula for the estimated mean outcome with external intervention based on the front-door criterion in Gaussian linear structural equation models (Q2048113)

The authors assume that cause-effect relationships among variables can be represented by a Gaussian linear structural equation model and a corresponding directed acyclic graph (DAG). They formulate the variance of the estimated mean outcome with an external intervention in which a treatment variable is set to a specified constant value. The proposed variance formula is exact, in contrast to those in most previous studies on estimating total effects. Based on the variance formula, they formulate the mean square error (MSE) between the future sample and the estimated mean outcome with the external intervention (Theorem 2). These results seems to extend some results given by \textit{M. Kuroki} [J. Appl. Stat. 39, No. 3, 673--694 (2012; Zbl 07263499)] or \textit{M. Kuroki} and \textit{T. Hayashi} [Scand. J. Stat. 43, No. 2, 505--519 (2016; Zbl 1384.62235)]. Some numerical experiments are given to illustrate their results.