consider the study from question 2, and other studies comparing an outcome (binary, continuous, or time to event) between two (or more) populations. for such studies, based on random samples from such populations, why are confidence intervals used?