Tutoring, Professional Development, and Educational Improvement: Evidence from Colombia

Our study aims to evaluate the individual and combined effectiveness of two interventions –teachers’ professional development and tutoring– on the test results of struggling students. The first intervention, teachers’ professional development, consists in one visit per month that randomly selected teachers in public schools will receive over the next three academic terms. The purpose of these visits, which combine classroom observations and feedback, and that are conducted by tutors and pedagogical advisors hired by the Carvajal Foundation, a local NGO, is to provide teachers with pedagogical strategies to strengthen their educational practices aimed at improving the basic language and math skills of struggling students. These visits will be reinforced with group workshops (three per term). The second intervention, the tutoring intervention, consists in two one-hour tutoring sessions per week that students in the bottom of the distribution within each grade/school included in the study will receive over the next three academic terms. Tutors will be hired by the Carvajal Foundation and will work directly with the students in reinforcing their basic language and math skills.

2017-08-01

2018-12-31

The Early Grade Reading Assessment (EGRA) and the Early Grade Math Assessment (EGMA). School Progression. Dropout.

In April 2017, out of the total 93 public schools in the city of Cali, Colombia, the Carvajal Foundation, a local NGO, randomly selected 7 schools (treated schools). These 7 schools will receive, starting in August 2017, the interventions that we describe below. The NGO collected language and math test scores of students in grades 2 through 5 in these public schools. In addition, out of the remaining public schools, the NGO will randomly select 4 schools (control schools) and will administer the same tests. These schools will not receive any intervention.

The two interventions in the 7 public schools will be assigned in the following way:

• Teachers’ professional development intervention: Out of the total 76 teachers in grades 2 to 5 in the 7 treated public schools, 38 teachers were randomly selected to receive this intervention. Randomization was conducted at the teacher level.
• Tutoring intervention: Out of nearly total 6,000 students in grades 2 to 5 in the randomly selected 7 public schools, 912 students will be selected to receive the tutoring intervention. These 912 students were selected considering the 12 students per school/campus/grade in the bottom of the distribution within each of these school/campus/grade units. Tutors were randomly assigned to each school/campus/grade.

At the end of each academic term language and math tests will be administered to all students in the sample of 9 public schools.

In April 2017, out of the total 93 public schools in the city of Cali, Colombia, the Carvajal Foundation, a local NGO, randomly selected 7 schools (treated schools), 19 institutions. These schools will receive, starting in August 2017, the interventions that we describe below. The NGO collected language and math test scores of students in grades 2 through 5 in these public schools. In addition, out of the remaining public schools, the NGO will randomly select 4 schools (control schools) and will administer the same tests. These schools will not receive any intervention.

The two interventions in the public schools will be assigned in the following way:

• Teachers’ professional development intervention: Out of the total 76 teachers in grades 2 to 5 in the 7 treated public schools, 38 teachers were randomly selected to receive this intervention. Randomization was conducted at the teacher level.
• Tutoring intervention: Out of nearly total 6,000 students in grades 2 to 5 in the randomly selected 7 public schools, 912 students will be selected to receive the tutoring intervention. These 912 students were selected considering the 12 students per school/campus/grade in the bottom of the distribution within each of these school/campus/grade units. Tutors were randomly assigned to each school/campus/grade.

At the end of each academic term language and math tests will be administered to all students in the sample of 9 public schools.

In April 2017, out of the total 93 public schools in the city of Cali, Colombia, the Carvajal Foundation, a local NGO, randomly selected 7 schools (treated schools), 19 institutions. These schools will receive, starting in August 2017, the interventions that we describe below. The NGO collected language and math test scores of students in grades 2 through 5 in these public schools. In addition, out of the remaining public schools, the NGO will randomly select 4 schools (control schools) and will administer the same tests. These schools will not receive any intervention.

The two interventions in the public schools will be assigned in the following way:

• Teachers’ professional development intervention: Out of the total 76 teachers in grades 2 to 5 in the 7 treated public schools, 38 teachers were randomly selected to receive this intervention. Randomization was conducted at the teacher level.
• Tutoring intervention: Out of nearly total 6,000 students in grades 2 to 5 in the randomly selected 7 public schools, 912 students will be selected to receive the tutoring intervention. These 912 students were selected considering the 12 students per school/campus/grade in the bottom of the distribution within each of these school/campus/grade units. Tutors were randomly assigned to each school/campus/grade.

At the end of each academic term language and math tests will be administered to all students in the sample of 9 public schools.

Randomization done by computer; private.

No

21 educational public institutions

Approximately 7000 students in grades 2 to 5.

38 teachers assigned to the first intervention, and 912 students assigned to the second intervention and 528 students in the control group.

Power calculations were estimated. We did several robustness exercises—different powers and effects sizes. Our most conservative estimate is that, in order to detect effect sizes of 0.15 sd on our measure of achievement with a power of 0.8, we need a minimum sample size of 1,396 students. Considering a conservative effect size of 0.15 sd, with a treatment group of 912 students and the control group of 528 students, our power is 0.78. Also, in order to confirm the patterns we observe in the control group, we will compare the results in the 4 randomly selected control schools to a different set of 4 randomly selected control schools in the city.