Title: Effect Heterogeneity and Variable Selection for Standardizing Experimental Findings

Abstract: Randomized trials are rarely representative of the general population. To account for this, results may be standardized to a target population which may be different from the one in which the trial was conducted. In this paper, we discuss three different approaches for reasoning about which covariates must be standardized over. This includes the traditional approach based on standard effect measures such as the risk difference or the risk ratio, Bareinboim and Pearl's graphical approach based on selection diagrams, and a new approach based on Counterfactual Outcome State Transition parameters. We compare and contrast these approaches, and argue that the choice must be informed by expert beliefs about which assumptions are most likely to be approximately true in the specific scientific context.