A null hypothesis articulated with specificity sufficient to allow the researcher to fill in a hypothetical value for each unit’s missing potential outcomes
Usually, use the hypothesis of absolutely no individual treatment effect (Yi(0) = Yi(1) for all units).