Experimental Evidence on the Effects of a Coursera Program on Labour Outcomes in Costa Rica

How to cite

Freund, Richard; Novella, Rafael; Rosas-Shady, David. 2023. "Experimental Evidence on the Effects of a Coursera Program on Labour Outcomes in Costa Rica." AEA RCT Registry. AEARCTR-0011483. https://doi.org/10.1257/rct.11483-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Richard Freund* , Rafael Novella , David Rosas-Shady

* Primary investigator

Registry ID
AEARCTR-0011483
DOI
https://doi.org/10.1257/rct.11483-1.1
First registered
Last updated
Status
Completed
Trial dates
2020-08-01 – 2023-09-30
Countries
Costa Rica (Latin America and the Caribbean)
JEL codes
E24, J01, J24, J62, I26
Registry record
http://www.socialscienceregistry.org/trials/11483

Abstract

Massive open online courses (MOOCs) are frequently viewed as a tool for democratizing education and job training. However, their effectiveness is largely untested. The existing evidence of the value of online courses is generally based on observational data where the selection of unobservable characteristics is a major concern (see e.g., Castaño-Muñoz and Rodrigues, 2021; Hadavand, 2018).

To address this issue, we carry out a randomized controlled experiment in Costa Rica in which the treatment group received cost-free access to Coursera MOOCs and certificates. Additionally, treated individuals were assigned to one of two behavioural intervention arms, where they either received light-touch reminders, or personalised reminders.

We are primarily interested in investigating three main questions: (i) What is the profile of the individuals who complete a course?; (ii) Is a more comprehensive, personalized reminder more effective in increasing course completion than a standardized email reminder?; and (iii) What are the labour market effects of completing a Coursera MOOC? To better understand the third question, we also analyse the heterogeneity of the treatment effects and study the impact of treatment on educational outcomes.

Intervention

In this intervention, randomly selected applicants whose jobs were affected by the pandemic were offered cost-free access to Coursera online courses and certificates. Treated individuals were also assigned to to one of two behavioural intervention arms - a light-touch reminder group, or a more comprehensive personalised reminder group.

Experimental design

The design followed an oversubscription model, whereby the program sought to attract a pool of eligible applicants that was at least twice the capacity the vacancies reserved for evaluation. From this, eligible applicants were randomly assigned to receive cost-free access to Coursera courses or not. After eligible applicants were assigned to receive treatment or not, a second round of randomization was implemented to assign treatment individuals to one of two behavioural intervention arms.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Individual
Clusters
8543 individuals
Observations
8543 individuals
Arms
4,271 individuals control, 4,272 individuals treatment. Within treatment, 2,136 individuals assigned to either behavioural intervention arm.
Primary outcomes
First, we are interested in assessing treatment take-up. The key outcome variables for this are where individuals: i) Activated their Coursera license ii) Enrolled in at least one online course, iii) Completed at least one online course, iv) The number of courses completed. Second, we are interested in assessing the labour market effects of the treatment. For this, the key outcome variables are: i) Current employment status, ii) Monthly income, iii) Formal job.
Secondary outcomes
As secondary outcomes, we are interested in the effect of treatment on other labour market outcomes. These include whether individuals are active in labour market, engaged in work related to what they applied to in Coursera, self-employed, have a tenured job, and are working full-time. We are also interested in the effect of treatment on educational outcomes. For this, we are interested in whether individuals are currently studying, studying full-time (i.e., not working or looking for work), and are engaged in online education.

Papers from this trial

  1. Novella, R., Rosas-Shady, D. and Freund, R. (2024). Is online job training for all? Experimental evidence on the effects of a Coursera program in Costa Rica. Journal of Development Economics 169, 103285. https://doi.org/10.1016/j.jdeveco.2024.103285

Cited by

No citation data available for this registration yet.

Keywords