Marginal versus conditional causal effects

Abstract

Conditional  methods  of adjustment  are often used to quantify  the effect  of the exposure on the outcome.  As  a  result,  the  stratums-specific  risk  ratio  estimates  are  reported  in  the  presence  of interaction   between   exposure  and  confounder(s)   in  the  literature,  even  if  the  target  of  the intervention on the exposure is the total population and the interaction itself is not of interest. The reason is that researchers and practitioners  are less familiar with marginal methods of adjustment such as inverse-probability-weighting  (IPW) and standardization and marginal causal effects which have causal interpretations for the total population even in the presence of interaction. We illustrate the relation  between  marginal  causal  effects  estimated  by IPW and standardization  methods  and conditional  causal effects estimated  by traditional  methods in four simple scenarios based on the presence  of  confounding   and/or  effect  modification.   The  data  analysts   should  consider   the intervention level of the exposure for causal effect estimation, especially in the presence of variables which are both confounders and effect modifiers.