Pubget

This article considers parametric inference of the causal effect of a sequence of treatments on an outcome after the last treatment in the presence of time-dependent covariates between treatments. We construct point parametrization for the conditional distribution of the outcome given all treatments and time-dependent covariates by using the point effects of treatments or time-dependent covariates as the point parameters. With point parametrization, we estimate sequential causal effects by maximum likelihood, improve the estimation by patterns of the sequential causal effects and treatment assignment conditions, and obtain unbiased consistent maximum-likelihood estimates of sequential causal effects even for long treatment sequences. http://pubget.com/paper/pgtmp_14111194/parametric-sequential-causal-inference-in-point-parametrization