· Video of Andrew Goodman-Bacon and Pedro Sant’Anna doing a discussion of difference-in-differences with staggered timing, with lots of nice graphs and intuition about what the problems are with two-way fixed effects and how to deal with this.
· On VoxDev, Paul Gertler and Brett Green summarize their interesting work on how digital collateral both improved loan repayment rates and got more kids enrolled in school. The idea is to make loans secured by something that the lender can digitally repossess in the case of non-payment through lock-out technology – in their case, a solar home system.
· The Washington Post has Katherine Baicker and Zui Song summarizing another randomized experiment showing that workplace wellness programs have very limited effects – this one done with cluster-level randomized assignment at BJ’s Wholesale Club – “We randomly assigned some BJ’s worksites to adopt a new wellness program and tracked employees at sites with and without such programs. After nearly three years, we didn’t see any substantial effects on employment outcomes (such as fewer sick days), health-care spending or objective health measures. We did find some improvements in self-reported health behaviors….if the goal is to save money by reducing health-care costs and absenteeism, or to improve chronic physical health conditions, the evidence so far is underwhelming.”
· On the JPAL blog, Aprille Knox, Aimee Barnes and Isabela Salgado share lessons from an evidence wrap-up of experimental evaluations aimed at reducing crime and violence. Issues covered include evidence on improving effective policing, improving dispute resolution systems, reconciliation initiatives after conflict, and reducing violence against women.
· On the Oxfam blog, Duncan Green has a wrap-up of lessons from six case studies of how research has policy impact, including factors that make it more likely for your research to have impact, some interesting case studies of how anthropology work influenced policy, and the difficulties of ascribing impact to a particular piece of research.
· On the Stata blog, making customizable tables in Stata 17 – how to create a classic table of summary stats.
Join the Conversation