The Labor Market Impacts of a College Role Model Intervention

In Spring 2026, students in randomly selected principles of economics classes received visits by two charismatic career women that majored in economics at the same university. The role models were chosen from the roaster of economics alumni with the assistance of two female students who were majoring in economics at the time. The students shortlisted alumni and conducted scripted interviews with them. They chose the role models based on their interest in their field of work, as well as their communication skills and overall charisma. Each role model visited the treated classes individually on separate days and delivered a 15-minute talk in which they shared their reasons for majoring in economics and described their career trajectories.

In Spring 2026, students in randomly selected principles of economics classes received visits by two charismatic career women that majored in economics at the same university. The role models were chosen from the roaster of economics alumni with the assistance of two female students who were majoring in economics at the time. The students shortlisted alumni and conducted scripted interviews with them. They chose the role models based on their interest in their field of work, as well as their communication skills and overall charisma. Each role model visited the treated classes individually on separate days and delivered a 15-minute talk in which they shared their reasons for majoring in economics and described their career trajectories.

The study began in the spring of 2015, when students enrolled in Principles of Economics courses completed an end-of-semester survey. To enable a consistent research design, the researchers secured the cooperation of the economics department to ensure that the same courses would be offered by the same instructors in the following academic year. Randomization was implemented in the spring of 2016, creating a study design that included both a control year (2015) and a treatment year (2016), during which both control and treatment classes were present.

2016-03-15

2016-04-15

The original outcomes of interest were measures of interest in the economics major, i.e., the decision to take intermediate microeconomics the year after the principle class, the number of economics classes taken, and the decision to major in economics. See Porter and Serra (2020).

The follow-up study utilizes LinkedIn data on labor market outcomes. The primary outcomes of interest are:
1) current wage
2) first wage post-graduation
3) earning potential

The wage outcomes are obtained from Revelio Labs, which provides modeled annual salary estimates (in USD) based on role, seniority, company (recorded from LinkedIn, when available) using large-scale labor market data, including job postings, government filings, and self-reported compensation records. For more information, see: https://www.data-dictionary.reveliolabs.com/data.html

1) Seniority
2) Job prestige
3) Work conditions: sentiment score

Explanation of secondary outcomes: Revelio Labs provides estimates of these variables. Seniority, i.e., the position’s level within the organizational hierarchy (e.g., entry-level, manager, executive), is based on job title and career history. Job prestige is a proxy for how prestigious or desirable a job is, based on the reputation of the employer, the role, and educational background of employees in similar positions. The sentiment score is generated by employee reviews and feedback, capturing how positively people feel about their jobs. For more information see: https://www.data-dictionary.reveliolabs.com/data.html

In Spring 2026, students in randomly selected principles of economics classes received visits by two charismatic career women that majored in economics at the same university. The role models were chosen from the roaster of economics alumni with the assistance of two female students who were majoring in economics at the time. The students shortlisted alumni and conducted scripted interviews with them. They chose the role models based on their interest in their field of work, as well as their communication skills and overall charisma. Each role model visited the treated classes individually on separate days and delivered a 15-minute talk in which they shared their reasons for majoring in economics and described their career trajectories.

The study began in the spring of 2015, when students enrolled in Principles of Economics courses completed an end-of-semester survey. To enable a consistent research design, the researchers secured the cooperation of the economics department to ensure that the same courses would be offered by the same instructors in the following academic year. Randomization was implemented in the spring of 2016, creating a study design that included both a control year (2015) and a treatment year (2016), during which both control and treatment classes were present. For more information, see Porter and Serra (2020).

The randomization was originally done in office by a computer.

The randomization was originally done at the class level, and stratified by class size, as there were small classes (capped at 40 students) and large classes (with over 100 students). Class sizes were determined by the department of economics and not by student demand.

Yes

12 clusters from the original experiment, when considering the Spring 2015 and the Spring 2016 student cohorts. Nine classes (i.e., sections of principles) remained unchanged between the two years, while 3 were different.
In the original paper (Porter and Serra, 2020), we also expanded the sample size and tested for parallel trends by including the Spring 2014 student cohort (all control students).

Original sample size: 1397 students, who were enrolled in principles of economics either in Spring 2015 (712) or Spring 2016 (685). An additional 671 students who took the same course in Spring 2014 were included in a parallel trends analysis, bringing the total sample to approximately 2,068 students. The current study includes all individuals from the original sample who, as of December 2024, have a LinkedIn account with publicly available education and employment information, as recorded by Revelio Labs, and can be matched to the original sample using their name, graduation year, and university affiliation. We expect the current study to include approximately 70–80% of the original sample, which may be reduced further if data are missing for any of the key outcome variables.

We expect the current study to include approximately 70–80% of the original sample, i.e. approximately 1400 to 1700 former students, including the original 2014 cohort (all control students). Since the set of former students will include the original 2014 cohort (all control students), the original 2015 cohort (all control students) and the 2016 cohort (half treatment students), we expect between 15 and 25 percent of the former students to be belong to the original treatment group, i.e., to one of the Spring 2016 treatment classes.