Experimental Design
The intervention features three teacher training programs focusing on either (i) teaching methods (didactics), (ii) content knowledge, or (iii) a combination of both inputs. Between mid May 2022 and early October 2022, primary school teachers participate in one of these professional development programs that share a common basic framework combining (a) seven face-to-face meetings, (b) coaching elements, (c) and six self-study modules.
1. In February / March 2022, we conduct school visits in the Salvadoran Departments (i) La Unión, (ii) Morazán, (iii) San Miguel, and (iv) Usulután to inform school heads and teachers about the professional development program and the study. Teachers register voluntarily with the representative conducting the school visit or later via phone.The target population comprises all teachers that instruct math to at least one class in grades four or five in 2022.
2. We invite those teachers that voluntarily registered to participate in the program to the baseline teacher assessment (April 2022). At the baseline assessment, (i) teachers provide written consent to participate in the program and study, (ii) take a 90 minutes mathematics assessment, (iii) fill in a socio-demographic survey, and (iv) verify/provide information on their weekly teaching timetable.
3. Informed by our power calculations and based on the list of participants at the teacher assessment, we select 338 teachers to participate in the study. To avoid distortion of treatment effect estimates via spillovers (e.g. Miguel & Kremer, 2004), the teachers are drawn from 338 different schools.
4. We conduct the baseline asssessment (+ short survey) with the fourth and fifth grade classes that are taught in mathematics by teachers in our study sample (April/May 2022). For each teacher, one class per target grade levels 4 and 5 is assessed. Note that some teachers only teach fourth graders (or fifth graders). Very few teachers instruct multiple classes on the same grade level; for those teachers, students of stream A, e.g. students of class 4A rather than class 4B, participate in the assessment.
5. During the student baseline assessment we randomize teachers into treatment and control. Randomization is conducted by the research team using the software R. The random assignment is stratified by region and terciles of baseline teacher performance; overall, this yields 12 strata (4 regions x 3 performance terciles).
6. Teachers are informed about their assignment to one of four experimental groups (T1, T2, T3, Control).
7. The three treatment arms are implemented from 21 May to 1 October.
8. Between mid August and mid September 2022, enumerators conduct classroom observations in one mathematics lesson (45min or 90min) of each teacher in the study sample; they use an adapted version of the Stallings Tool to collect information.
9. Midline assessments with students are conducted in October 2022.
10. Midline assessments with teachers are conducted in November 2022.
11. Depending on how many classes are assigned to new teachers between the two school years 2022 and 2023, a mid-implementation "baseline" student assessment is conducted in February 2023.
12. In August 2023, enumerators conduct classroom observations in one mathematics lesson (45min or 90min) of each teacher in the study sample; they use an adapted version of the Stallings Tool to collect information on teaching practices.
13. Endline assessments with students are conducted in October 2023.
14. Endline assessments with teachers are conducted in November 2023.