Encouraging Government Higher Education Scholarship Application through Motivation and Information Intervention: Randomized Controlled Trial in Indonesia

The intervention is an information session held in person at the high school the student attends. A teacher from the high school will collect students in the treatment group in the particular high school and take them to a separate classroom in the school. Once in the room, a zoom session is started and one of the principle investigators of the trial will conduct the online live zoom session from online. She will first play a 11 minutes video with information about the scholarship. The video starts with an introduction of a previous student from a economically disadvantaged background in the same city (role model) who was awarded the scholarship and who had been successful in his/her early career. The role model briefly tells his/her story about his/her family and economic background and how the scholarship helped him/her to complete university and succeed in the job market. Then the role model's voice goes through easy to understand information about the scholarship, mainly on who are eligible to apply and how to apply. After the video is played, a live question and answer session is conducted with all the students. Then the same video with a different role model (of different gender) is played again. The whole session is less than an hour. For 29 high schools, a single zoom session was conducted with students sitting in 29 high schools at the same time on January 12, 2023. Because 2 high schools could not join the original session, two separate but same zoom sessions are conducted on January 16, 2023 and January 17, 2023.
To avoid spillover of the intervention to students in the same high school, the teach and the students cannot watch the video again after the zoom session as they do not have a link to the video. Teachers are also told not to reveal the information covered during the session to students who are not in the treatment group.

2023-01-12

2023-01-17

1. Student's motivation about applying for the scholarship.
2. Student's knowledge about the scholarship.
3. Whether the student apply for the scholarship.

For outcomes 1 and 2, an intermediate survey will be conducted with students in both control and treatment groups a few weeks after the intervention. The survey has questions about the degree to which the student intends to apply for the scholarship, and questions that evaluate the student's knowledge about the eligibility and application process of the scholarship. Various measures of motivation and knowledge will be constructed from the answers from this survey.
For the third outcome (whether the student apply for the scholarship), the principle investigators will try to find out whether the student has applied for the scholarship through various sources (from the student, school, and government agencies), and will try to verify the information.

The target population of our trial are high school students in their last year who are eligible for the scholarship. These are students who come from economically disadvantaged background. We are able to convince 31 high schools (most of the high schools) in one major city in Indonesia to participate in our trial. From these high schools, we identified around 1300 students who would be eligible for the scholarship based on information provided by the high schools. We conducted a baseline survey using online survey tools that students can use through their mobile phone with these students. 1281 students completed the survey. Then we did stratified randomization to allocate these 1281 students into one control and one treatment group. We stratified along household head income quartiles, initial intent to apply for any scholarship (low, medium, high), and knowledge about the particular scholarship we provide information on (low, medium, high, very high). The randomization allocates 628 students into control group, and 628 students into treatment group (25 students were dropped either because they were in small strata or because they were misfits in the stratified randomization process). We did not conduct clustered randomization (either at school or class level) because power calculation shows that we would not have enough power if we conducted clustered randomization. We only have one treatment group due to power concerns, too, and as a result we are not able to separate the effect of motivation and the effect of information. We evaluate the effect of these two channels combined.

A few weeks after the intervention is conducted with the treatment group, an intermediate follow up survey will be conducted with both control and treatment group (online survey that can be completed using mobile phone) that evaluates the students' intent to apply for the scholarship and the knowledge about the scholarship (who are eligible, how to apply). We will also track which students apply for the scholarship until the application deadline (August, 2023).

After all the outcome data are collected, we will conduct regression analysis to analyze the effect of the intervention on students' intent to apply for the scholarship, knowledge about the scholarship, and whether the student applies for the scholarship.

Randomization is done in office using Stata based on the baseline data we collected. Stratified randomization is conducted. We first analyze the relationship between the intent to apply for any scholarship/knowledge about the particular scholarship we provide information on and family characteristics, student characteristics, and high school fixed effects. We find that the intent to apply for any scholarship/knowledge about the specific scholarship we provide information on is significantly correlated with household head income but not with high school fixed effects. These findings direct us to conduct stratified randomization along household head income quartiles. We also stratify along the intent to apply for any scholarship (low, medium, high) and knowledge about the scholarship our intervention provides information on (low, medium, high, very high) as we anticipate these characteristics to be strongly correlated with our outcome variables. The fact that high school fixed effects are largely not correlated with the intent to apply to any scholarship or with the knowledge about the particular scholarship we provide information on is reassuring for out decision to not cluster our randomization at school level.

We randomize at the student level.

No

1256

1256

628 in treatment group, 628 in control group.

Assuming 50% of students in the control group will apply for the scholarship, and assuming a standard deviation of 0.5 for the outcome whether apply for scholarship, the minimum detectable effect size is 7.9%, i.e., the intervention increases the application rate by 7.9 percentage points. Assume 12% of students in the control group will apply for the scholarship (equal to the percentage who have very high intent to apply to any scholarship at the baseline), and assuming a 0.325 standard deviation, the minimum detectable effect is 5.14%, i.e., the intervention increases the application rate by 5.14 percentage points.