Under conditional exchangeability (i.e., randomization), then allowing for interactions between treatment and baseline in the ANCOVA allows one to estimate both the ATE (main effect) and ATT (interaction).
Contrast with a change score analysis, which only targets the effect on the treated (ATT) and not the full population average causal effect (ATE). These estimands can differ from each other when treated individuals differ from the untreated on variables that modify the treatment effect.
Under conditional exchangeability (i.e., randomization), then allowing for interactions between treatment and baseline in the ANCOVA allows one to estimate both the ATE (main effect) and ATT (interaction).
Contrast with a change score analysis, which only targets the effect on the treated (ATT) and not the full population average causal effect (ATE). These estimands can differ from each other when treated individuals differ from the untreated on variables that modify the treatment effect.