txshift.RdComputation of targeted maximum likelihood (TML) estimates of the population-level causal effect of a continuous-valued intervention based on a simple stochastic treatment assignment mechanism, defined as the counterfactual mean of the outcome of interest under an additive shift of the observed (natural) value of the intervention. To accommodate settings in which two-phase sampling is employed, an inverse probability of censoring weighted (IPCW) TML estimator for the counterfactual mean under the stochastic intervention is available, alongside procedures facilitating robust estimation of nuisance parameters in a manner ensuring nonparametric efficiency of the IPCW-TML estimator of the causal parameter.
txshift()