Motivating Disadvantaged Students during the COVID-19 Pandemic. The Role of Information, Aspirations, and Perceptions

Our study motivates students by providing information about a merit- and need-based scholarship to attend college. Created in late 2011, Beca 18 is the first full scholarship program for higher education funded by the national government in Peru. This trial conducts a large-scale experimental intervention that aims to motivate public school students, specifically those of low and medium-low socioeconomic status, through text messages.

We randomly assign schools into three treatment arms, a placebo treatment group, and a control group. The three treatment groups received general information about the eligibility criteria and coverage of Beca 18, but specific text messages separately to each treatment arm. The first treatment arm provides information about the labor market returns of Beca 18. Then, the second gives details about the current gap between the eligibility academic criteria of Beca 18 and the academic performance of of children. The third arm, instead, seeks to expand the aspirations window of both parents and their children through specific information of past Beca 18 beneficiaries in their region of residence. Additionally, the placebo treatment group only receives information about Aprendo en Casa, the Ministry of Education's multimedia platform (website, TV and radio), created for remote learning. Finally, the control group does not receive a message.

2021-10-04

2022-12-03

Our study design aims four primary outcome variables: high school graduation, Beca 18 eligibility, college application, and future aspirations.

The high school graduation outcome will be measured using information on grade attainment by Year 3, average GPA of all years, and on-time high school graduation.

The Beca 18 eligibility outcome will be measured using information on relative class standing, whether the student applied, the test score in the Beca 18 national test, and whether the student is selected as a beneficiary.

The college application outcome will be measured using information on college characteristics (e.g., 3-year vs 5-year college, private or public, selective college, etc.) and the major choice: social science, STEM, humanities.

These outcomes come from administrative data thanks to our collaboration with MineduLab.

We will further evaluate four secondary outcome variables: learning progress, parental involvement, and college aspirations.

The learning progress outcome will be measured using information on GPA, remedial education, and relative standing in class.

The intervention focuses on rising sophomores from public schools. Unlike senior students, sophomores can still change their academic performance to qualify as eligible for Beca 18. Out of the total secondary education, the sampling frame was restricted to 6,081 public schools with a minimum of 10 sophomore students and those with at least 25% of its students with a contact cellphone number. The poll of schools was stratified for parental wealth and previous educational achievement. Then, we randomly assigned the selected schools into the three treatment arms, the placebo treatment group, and the control.

Not available

We utilize the randtreat Stata command to randomize at the cluster level, considering a bi-dimensional stratification (four strata) on previous mean students' academic achievement and mean students' family income at the school level.

We randomize at the cluster level: schools.

Yes

6,080 schools.

134,256 students.

1,219 schools pure control, 1,214 schools receives employment gains information, 1,219 schools receives current educational achievement information, 1,218 schools receives role models information, and 1,211 placebo treatment.

The significance level of the test (α) was defined at 0.05 and the power of the test (β) at 0.8, that the minimum detectable effect would be 0.20 standard deviations (SD = 91.93) on the average score in mathematics, 560.29, that the average number of treated students per cluster would be 25 students, that the intraclass correlation coefficient (IIC) is 0.224 and that the coefficient of variation of the cluster size is 1.21. The results indicate that the we need to intervene 1,000 schools.