Experimental Design Details
In July 2021, we collaborate with a local government in Japan and conduct a field experiment toward 4,000 young adults (30-39 years old) residing in that area. In the experiment, we randomly divide the subjects into either of one control group or three treatment groups (A, B, and C). The local government mails coupons, which are required for COVID-19 vaccination bookings and post-vaccination records, to their residents. In the treatment groups, we respectively put a sticker with a nudge-based message on the envelope to mail the coupons. The message for Group A is “Vaccination will protect you and your dear ones from COVID-19,” the message for Group B is “Vaccination enables you to visit your dear ones with peace of mind,” and the message for Group C is “A vaccine is reserved for you.”
We explain the details of randomization procedure. We first provide background information. The local government stores the envelopes containing coupons in boxes for each age group. Each box contains 450 envelopes. In the box, the envelopes are rubber-banded together in pairs of 25 envelopes. Thus, there are 18 bundles of envelopes per box (25 envelopes per bundle x 18 bundles = 450 envelopes per box). Which box an envelope will be placed in, and in which order it will be arranged in that box, is determined by the address number. The address number is assigned by sorting the postal code, town name, and place number. Thus, envelopes in the same box are likely to be mailed to residents of the same district. In addition, those in the same bundle are likely to be mailed to the thirties who live close to each other.
The local government staff carry out cluster randomization, according to the following procedure. First, they pick up 10 boxes containing envelopes addressed to the thirties (450 envelopes x 10 boxes = 4,500 envelopes). Here, they select the boxes so that the district composition of the selected 10 boxes is similar to that of the local government area. Second, they pick up 16 bundles per box (25 envelopes x 16 bundles = 400 envelopes, leaving two extra bundles), set up 25 rubber-banded envelopes as a cluster, assign a sticker with either of the above nudge-based messages to each cluster, and attach the sticker to the envelopes. They change the order in which they assign the stickers depending on the boxes, to prevent, for example, the same sticker being attached on the first bundle of any box.
In July 2021, for the first time, this local government mails the coupons to ordinary people in the thirties and start accepting reservations for vaccination. However, those who wished to receive the vaccine earlier for some reasons could apply to the local government and obtain the envelopes and coupons before the mailing. In the bundles including such subjects, the number of envelopes is less than 25. In such cases, the staff replenish the envelopes from the remaining two bundles (25 envelopes x 2 bundles = 50 envelopes), which belonged to the same box. As a consequence, 1,000 subjects are assigned per group. The total number of subjects is 4,000, and the total number of clusters is 160.
The local government merge information on which group the subjects belong to, whether they receive the vaccine, and when they receive it, and share the list with us. The list is completely anonymized.