7.

A 2019 article in the American Journal of Public Health (AJPH) reports the results of a randomized study designed, as per the researchers, to "assess the effect of 2-way short message service (SMS) with a nurse on postpartum contraceptive use among individual women in Kenya". The reported results include the following prose:

"We randomized 260 women to 2-way SMS or control. At 6 months postpartum, 69.9% of the 130 women receiving SMS reported postpartum use of highly effective contraceptives (HEC), compared with 57.4% of the 130 women in the control arm"

Consider this study, and other studies comparing outcomes (binary, continuous or time to event) between multiple populations. For such studies, based on random samples from such populations, why do we use confidence intervals?

a. Confidence intervals guarantee that the study results will be scientifically useful.

b. Confidence intervals help validate that a study was well designed.

c. Confidence intervals account for the uncertainty in sample estimates that comes from the variation in such estimates across different studies of the same size.

d. Confidence intervals account for uncertainty that arises because of problems with the study: for example, people answering a question incorrectly, or an incorrect measurement.