Perceived gender composition of programs and college applications

Survey experiment with information treatments

2023-02-18

2023-03-10

We will estimate the impact of the treatment on different outcomes: (1) the submitted list on Parcoursup, (2) the final placement.

(1) Submitted list on Parcoursup: Using the administrative data, we look at the treatment effect on characteristics of the application list submitted on Parcoursup. We look at the dummy for application to at least one CPGE program and at least one Scientific CPGE. Moreover, we measure the maximum and average prestige of the submitted list on Parcoursup (prestige is based on the average grade of accepted students in a program).
(2) Final placement: In the administrative data, we measure whether the final match is a CPGE/Scientific CPGE and the prestige of the final placement.

We will test the impact of the treatments on post-treatment items, in order to investigate the potential mechanisms (explained below).

In this survey experiment, subjects will be randomized into one of two information treatments and a control. We merge their responses to the administrative Parcoursup data to study their application behavior and final matches.

We use social media ads to recruit our sample of Parcoursup participants. The ad redirects participants to our Qualtrics survey.
In the survey, we ask for information related to participant’s high school education and related to their study intentions. We measure their confidence in their academic performance relative to others. We also collect information on risk, competitiveness, and gender homophily, i.e., how comfortable the participants are to be in a group composed of the opposite gender. Additionally, we elicit gender stereotypes on how well male and female students perform in CPGE. These controls will allow us to disentangle the mechanisms behind potential treatment differences.
In the survey, we ask for participants’ beliefs about the share of female students accepted to CPGE and Scientific CPGE in 2021. Afterward, subjects are randomized into treatments on the individual level: No feedback (NoInfo); Feedback on the share of female students in CPGEs (CPGEshare); Feedback on the share of female students in Scientific CPGEs (ScientifcCPGEshare). Depending on the condition, the respondents receive information on the correct share of female students in 2021, taken from admin data.
Ultimately, we ask them to self-assess their fit to CPGE and CPGE scientific and elicit their belief on where they stand in the ability distribution of male- and female-composed groups. This allows us to have post-treatment measures related to three hypothesized channels: stereotypes about gender performance, competitiveness, and homophily.
In the analysis, we estimate average treatment effects by regressing the outcomes on treatment dummies and the interaction of the treatment dummy and prior beliefs. To increase precision, we control for baseline characteristics. Moreover, we analyze heterogenous treatment effects by gender, socioeconomic background, and Bac grade.
We will do subgroup analysis for the sample of participants who indicated that they have considered applying to a CPGE and the sample with a science-related specialization in high school.

Computerized randomization using Qualtrics.

individual

No

We aim for 6,000 individuals but will collect as many respondents as possible during the period, in which the ads are planned (18 February till 9 of March).

6,000 individuals

We aim for 2,000 individuals for each of our three treatment arms. We target a rather large sample size as we can only use individuals for the final analysis whom we can match to the administrative data.