On Efficient Adjustment for Micro Causal Effects in Summary Causal Graphs
Isabela Belciug, Simon Ferreira, Charles K. Assaad
TL;DR
The paper addresses identifying micro causal effects from observational data when the underlying causal structure is represented as a summary causal graph (SCG). It provides a simplified, kinship-based reformulation of identifiability conditions and introduces the SCG-back-door criterion to enumerate multiple valid adjustment sets for estimating ${\Pr}(y_t \mid \text{do}(x_{t-\gamma}))$. It additionally defines a quasi-optimal adjustment set, $q\text{-}opt(X_{t-\gamma}, Y_t)_{\mathcal{G}^s}$, that tends to minimize asymptotic estimation variance across compatible full-time DAGs and relates this set to the classical ${\mathbb{A}}^1$ and ${\mathbb{A}}^2$ bounds. The results offer both theoretical advances and practical tools for efficient causal inference in abstracted time-series graphs, with discussion of limitations, sampling considerations, and future extensions (e.g., front-door in SCGs) that broaden applicability to cohort data and imperfect measurements.
Abstract
Observational studies in fields such as epidemiology often rely on covariate adjustment to estimate causal effects. Classical graphical criteria, like the back-door criterion and the generalized adjustment criterion, are powerful tools for identifying valid adjustment sets in directed acyclic graphs (DAGs). However, these criteria are not directly applicable to summary causal graphs (SCGs), which are abstractions of DAGs commonly used in dynamic systems. In SCGs, each node typically represents an entire time series and may involve cycles, making classical criteria inapplicable for identifying causal effects. Recent work established complete conditions for determining whether the micro causal effect of a treatment or an exposure $X_{t-γ}$ on an outcome $Y_t$ is identifiable via covariate adjustment in SCGs, under the assumption of no hidden confounding. However, these identifiability conditions have two main limitations. First, they are complex, relying on cumbersome definitions and requiring the enumeration of multiple paths in the SCG, which can be computationally expensive. Second, when these conditions are satisfied, they only provide two valid adjustment sets, limiting flexibility in practical applications. In this paper, we propose an equivalent but simpler formulation of those identifiability conditions and introduce a new criterion that identifies a broader class of valid adjustment sets in SCGs. Additionally, we characterize the quasi-optimal adjustment set among these, i.e., the one that minimizes the asymptotic variance of the causal effect estimator. Our contributions offer both theoretical advancement and practical tools for more flexible and efficient causal inference in abstracted causal graphs.