Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
According to the DHS 2016 in Uganda, 21.43% of people use modern short-term contraceptive (Injectable and Pills) in Ankole region (sample weighted).
Following the power calculation of two proportion test, the treatment group has to have 29.57% take up of family planning with power set at 0.9. This is the 8.14 percentage points increase from the baseline.Our target sample size is 1,000 women. With 1,000 subjects, we can test various hypothesis as outlined below.
Primarily, we compare 400 women who will be offered the pregnancy test under the treatment group and another 400 women in the control group who will not be offered the pregnancy test. The main outcome is the take-up of the modern family planning methods. This sample size of 800 women allows us to detect the standardized effect size 0.2 with power 0.8 and significance level 0.05. Assuming conservatively that the standard error is 0.5, we expect the treatment effect is 10 percentage points increase in the family planning take-up.
We will additionally test if the previous experience of pregnancy test affects the willingness to pay for the pregnancy test, among 400 women (200 in treatment and 200 in control). This sample size allows us to detect the standardized effect size 0.28. Although this is a pilot study to test out the hypothesis using the small sample, we plan to scale-up in the future using the sufficient sample size, should we see any potential effect through this pilot study.