Addressing Inequalities in High School Track Choice

We implement the following interventions:

(1) Experiment in teacher survey. We show a series of vignettes with hypothetical students' profiles to teachers and ask them to provide a track recommendation for these hypothetical students. We randomize students' gender and SES status across vignettes to study if they affect teachers' recommendations. We implement two teachers' online surveys: the baseline survey was sent by email to the official school email address in November to recruit the teachers. The second survey will be sent by email to the recruited teachers, through the email address they provided in the first questionnaire.

(2) Information intervention at the school level. Between November 27 and December 17 2023, teachers in treatment and control schools received a personalized report with aggregate statistics about the performance in high school of their past students, along with the province and national average. All teachers received information about their past students' average grades in math, Italian, and English in high school, the average share of students admitted in grade 10, and the average number of subjects in which they failed in grade 9. They also received information about these outcomes separately for students who followed and did not follow their track recommendation, and for students who chose different high school tracks. In addition to this information, teachers in treatment schools received information about inequality in their past track recommendations. In particular, they were shown the percentage of students that they recommended to each track by students' SES; the same statistics were shown separately for students with high and low GPA. Moreover, they were shown their believed and actual share of students from low SES assigned to the high tracks, and the actual performance in high school of students with high GPA in middle school but different SES. The individualized reports were sent to teachers via email. We will evaluate the impact of this intervention on teachers' track recommendations and students' choices. We chose to randomize at the school level rather than at the teacher level in order to avoid contamination between teachers who received the additional information on inequality in recommendations by SES.

In the academic year 2023-24, we have expanded the intervention to include additional schools. The intervention was implemented in November and December 2023. The information treatment remains similar to the previous year, with the following modifications. In addition to the control group, there will be two treatment groups. In the first treatment, teachers receive information on the gaps in track recommendations for their past students, along with information on teachers' beliefs regarding whether students should receive different recommendations based on their socio-economic status (SES). In the second treatment group, teachers receive additional information aimed at correcting their misconceptions about the probability of success for low SES students in high school. We present teachers with reported beliefs on the probability of failure in grade 9 for low SES students (on average and conditional on a high GPA in middle school), comparing this with the truth based on administrative data for the same variables.

2022-11-27

2023-12-23

The outcomes will include:
(1) administrative outcomes on students' high school track choice and teachers' track recommendations
(2) data collected from a teachers' questionnaire that will be implemented in January-February 2023, including teachers' beliefs about their own track recommendations, teachers' beliefs on students' future performance in high school and return to schooling by SES, self-reported track recommendations for their students, and track recommendations for hypothetical students' profiles.

Information intervention with teachers and lab-in-the-field experiment in teachers' survey.

We plan to study how receiving information about inequality in their own track recommendations affects teachers' recommendations for their current students, as well as subsequent students' choices. Teachers provide track recommendations to their students between mid-November and mid-December, depending on their school. Due to time constraints, some teachers received the personalized reports after their school had assigned the track recommendations. We will account for this in the analysis, as we expect to find a smaller effect in schools in which teachers already assigned the track recommendation.

The additional treatment implemented in the academic year 2023-24 will allow us to assess whether correcting teachers' misperceptions about low-SES students' probability of failure in top high schools affects their track recommendations.

Randomization done in office by a computer.

The unit of randomization is the school.

Yes

230 schools in total (85 schools in the first year, 150 additional schools in the second year)

Around 1400 teachers and 10,000 students in total (Around 4,000 pupils and 600 teachers in the first year of implementation, 780 teachers and 6,000 students in the second year).

In the first year of implementation, around 300 teachers in control group and 300 in treatment group.
In the second year of implementation, around 180 additional teachers in control, 300 in first treatment group, 300 in second treatment group.

Considering ICC of 0.07, 230 schools, 40 students per school, the MDE is 0.113 SD or 0.05 in terms of percentage points change assuming 20% of students are recommended toward a top-tier track.