Experimental evaluation of an outreach program for lifelong learning
How to cite
Karhunen, Hannu; Kauhanen, Antti; Ollikainen, Jani-Petteri; Riukula, Krista; Suhonen, Tuomo; Virtanen, Hanna. 2023. "Experimental evaluation of an outreach program for lifelong learning." AEA RCT Registry. AEARCTR-0011550. https://doi.org/10.1257/rct.11550-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This experiment evaluates the effects of an outreach intervention on lifelong learning. The intervention is implemented by Finland's Service Centre for Continuous Learning and Employment (SECLE). It seeks to improve the co-operation between companies and education providers with the ultimate goal of encouraging underrepresented groups to increase their educational participation. In practice, SECLE hires service providers who approach companies employing 3 - 149 persons and provide their services for the companies that agree to take part. First, the service providers chart the skill needs of the companies and employees. Next, the service providers organize an event tailored specifically for each company in which they provide information and guidance on different educational options. Evaluation of causal effects is made possible by random assignment of areas to treatment and control groups. The service providers only approach companies located in the treatment areas. We aim to have at least 1,200 companies in the treatment group.
Intervention
In late 2022, SECLE hired service providers from different areas in Finland to carry out the outreach intervention in 2023. The intervention is structured as follows: First, the agents approach companies (3 to 149 personnel) to chart their skill needs. Next, if company is interested, agents will hold an informational event for the employees. In the event, the agents will provide information on local educational options as well as practical guidance on applying, enrollment, and financial support. The exact content of the event will be tailored based on a previous discussion with company management. These events can also be organized for multiple companies simultaneously if deemed suitable. The events are targeted towards individuals with only primary or general upper secondary education although individuals with higher educational attainment are not forbidden to attend.
The service providers receive a payment for each event they organize, provided that at least one attendee has only primary or general upper secondary education and is not currently enrolled in education. If there are more than one such attendees, the agents will also receive additional payments for each one of them. The intervention is provided only in the randomly selected areas for companies employing 3 to 149 employees. The agents are instructed to only approach the companies fulfilling these prerequisites. They also receive the financial reward only for the events organized in these companies.
Experimental design
We use the Enterprises and Establishments register of Statistics Finland to define the target group. The target group includes companies or establishments in Finland with 3-149 employees.The unit of randomization is an area, where smaller municipalities form one area and the seven largest cities (Helsinki, Espoo, Vantaa, Tampere, Oulu, Turku, and Jyväskylä) are further divided into smaller areas based on the first three numbers of zip codes. We exclude regions in Lapland and the Åland islands as well as areas with less than 20 companies/establishments with 3-149 employees. We were left with 262 areas for randomization, 206 of the these are municipalities and 56 zip code areas (approximately 45 000 companies/establishments in the spring of 2022).
We conducted the randomization in August 2022 using the following procedure:
• We used stratified randomization by population size to ensure that different sized areas are represented in treatment and control groups. We sorted the areas by population and divided them into four equal sized strata.
• We performed simple randomization within each stratum by first generating a random variable that took values between 0 and 1 from a continuous uniform distribution (runiform in Stata). The areas were then sorted based on this random variable and the first 50% of areas within each stratum were assigned to the treatment. The rest of the areas formed the control group.
Design details
Cited by
No citation data available for this registration yet.