Randomized Controlled Evaluation of Workforce Programs
RCTReview
Non-experimental evaluations of employment interventions frequently produce misleading results due to selection bias, as highly motivated candidates self-select into training programs. Without rigorous counterfactuals, policymakers cannot determine whether observed wage increases stem from program quality or unobserved participant characteristics.
Picture this
A randomized controlled trial acts like a scientific coin flip that creates two identical twin groups of job seekers. By randomly assigning qualified applicants either to receive sectoral training or to a control group receiving standard services, any subsequent difference in earnings can be attributed solely to the training program itself.
What the evidence says
Randomized sorting ensured baseline equivalence between groups, establishing that sectoral programs causally increased 2-year earnings by $4,500 in SEIS (18% increase, p < 0.01), 5-year earnings by $2,716 in WorkAdvance (11.5% increase, p < 0.01), and 11-year earnings by $4,616 in Project QUEST (15% increase, p < 0.01).
- Who was studied
- Thousands of human adult and youth job seekers across 9 sectoral programs evaluated in 4 multi-site randomized trials in the United States.
- How
- Randomized Controlled Trial (RCT) methodology comparing treatment and control group outcomes over 2-year to 11-year follow-up windows (Maguire et al. 2010; Hendra et al. 2016; Fein et al. 2021; Roder & Elliott 2021).
What to do
Mandate randomized lottery assignment among qualified workforce applicants during pilot program rollouts to establish rigorous causal evidence of earnings impact before national scaling.
From the source
"Randomly sorting a population into two groups--one that receives a program and one that does not--ensures that the groups are, on average, balanced at the beginning of the study."
Evidence-Review_Sectoral-Employment_2222022_0.pdf