Experimental Design Details
For the first component, we use a sample of 2,005 schools, we have randomly assigned them to the following groups:
- Treatment group (T): 1,123 schools will be randomly allocated to receive the training from the UGEL specialists. They will also participate in an intensive data collection activity.
- Control (C): 580 schools will be randomly assigned to a control group. They will not receive the training from the UGEL specialists but will participate in the intensive data collection activity.
- Pure control (PC): To address concerns about the Hawthorne effects--that is, principals and teachers can change their behavior in terms of how they report the data due to the intensity of the data collection process--we will randomly assign 302 schools to a pure control group that will neither receive the training nor will report any data. From these schools, we will collect administrative data on their reopening status.
For the second component, we plan to work with a sample of 1775 school principals distributed across eights regions in Peru from which we have phone numbers. To identify relevant elements for the reopening decision, we conducted five focus groups with 16 principals in January 2022. We found a total of 7 characteristics that are relevant for them: (1) the need for an additional budget for hygiene and cleaning kits, (2) the number of COVID cases in the district, (3) networks (if other principals have reopened), (4) adequacy of school for social distancing and hygiene, (5) liability for reopening, (6) training, and (7) UGEL support.
Using these characteristics, we created vignettes. Each vignette is a hypothetical setting for the reopening with a specific value for each of the seven features. Two vignettes constitute a choice set. We decided to vary only three of the characteristics between the two vignettes. The other four are held constant. Thus, we have a total of 10 groups of varying attributes. Then, following Datta (2019), we created a total of 180 choice sets (with two vignettes each), that is, 12 sets (24 vignettes) by a group of varying attributes. Each school principal will be presented with a total of 10 choices, each of them will be randomly selected from every group of varying attributes. To reduce cognitive burden, the varying attributes will be highlighted.