Experimental Design Details
Participants are split into four treatment arms. They are randomized at the individual level without stratification.
All participants receive a 1000K gift of airtime at the end of the survey, and a 500K airtime voucher for redemption at any HIV testing site in Zomba Town. They do not have to agree to an HIV test to redeem this voucher.
Prior to applying the results of the randomization, we will elicit preference for a hard commitment device from each participant.
0: 25% are assigned to the control group.
1: 25% of participants will then be offered the hard commitment device. The original preference we elicited from all participants will be honoured. If their preference was to refuse the commitment device, they receive an immediate payment of 1000K. If their preference was to accept, they forego the immediate payment of 1000K, but can receive a 1000K transfer if they show up at one of the HIV testing sites. They do not have to agree to an HIV test.
2: 25% of participants will be offered a scheduled appointment to test for HIV. They are allowed to refuse the appointment.
3: 25% of participants will be offered both the hard commitment device and the appointment.
The effect of the soft commitment device (being offered an appointment) can measured by comparing those randomly selected to get an appointment (groups 2 and 3) to those who were not (groups 0 and 1).
The intent-to-treat effect of offering the hard commitment device can measured by comparing those who were assigned to groups 1 and 3 to those assigned to groups 0 and 2. We can also measure the effect on compliers using the same comparison, but restricting our sample to those who preferred the hard commitment device.