Minimum detectable effect size for main outcomes (accounting for sample
design and clustering)
We assume a post-randomization take-up rate of 54 percent (share of the treatment group who enroll in the program), 80 percent power, and a type-1 error rate of 5 percent. We base control group means and standard deviations on the reported outcomes in the Year Up evaluation (Fein and Hamadyk,2018). Under these assumptions we are powered to detect the following effects under simple random assignment: Quarterly Earnings MDE: $1,022 increase among those who enrolled in the program; $552 increase for the entire treatment group (mean = $3,600; SD = $3,600) Indicator for earning more than $30,000 per year: 10.1 percentage point increase among those who enrolled in the program; 5.5 percentage point increase for the entire treatment group (mean = 0.15)