· Nature published several many-authored papers on replication. Brodeur et al. report on the Institute for Replication’s replications and robustness checks of 110 published studies in economics and political science. “We found that more than 85% of published claims were computationally reproducible. In robustness checks, our reanalyses showed that 72% of statistically significant estimates remain significant and in the same direction, and the median reproduced effect size is nearly the same as the originally published effect size (that is, 99% of the published effect size).” Miske et al. assess reproducibility in a sample of 600 papers across social and behavioral sciences – they could only get data for 182 of them and found 77% were precisely reproducible with “higher reproducibility for papers from political science and economics compared with other fields (business, psychology, sociology, education), for more recent papers compared with older papers and for papers from journals that require data sharing.”
· Can Claude Code replace a Dave Evans conference round-up? Jacobus Cilliers uses Claude to channel Dave and (along with his own reflections) summarize all the papers at the recent CSAE conference. He highlights the increased number of African researchers presenting, the increasing use of long-term studies, admin data, and lots of papers that unpack graduation programs.
· Notre Dame BIG Faculty spotlight on Rebecca Thornton has discussion of her work on early childhood education in Uganda, and some interesting examples of ethics issues related to research work “I think my perspective on research ethics has been shaped by experiences early in my career, when I underestimated the impact that researchers can have simply through the process of data collection. Two studies in particular made me realize that we don’t always fully internalize how our presence and our methods can affect people’s lives….we interviewed men about their beliefs related to HIV prevention and collected information about their partners. A year later, after receiving additional funding, we returned to interview those partners, whose names we had recorded during the initial survey. All of this had received ethical approval, and we were transparent about our plans. Over the course of the year, some partnerships had changed. In some cases, individuals were now married to someone else. When our team approached women and asked for them by name—for example, “Are you Mary?”—we inadvertently revealed information about past relationships, including infidelity. No one had fully anticipated that simply referencing a name from a prior survey could create tension or conflict within a household.”
· Princeton RPDE conversation with Elisa Macchi on how she decides what research projects to do, her favorite work so far, and what makes for a great economics paper “I think we often put a lot of emphasis on the idea and the results. Of course those matter, but in my experience, many of the best papers distinguish themselves by how they deliver the results. The writing, the structure, the way the evidence is presented and built up: these choices are less glamorous, but they have a lot of impact.”
· A new SIEF from Evidence to Policy note covers work I did with many co-authors on how education systems can ensure online learning works when crisis hits. In experiments in Ecuador, we found centralized online management was effective, whereas “teacher-level nudges, including benchmarking emails, encouragement messages, and administrative reminders, did not improve student participation. Small financial incentives for students increased study time but did not translate into more knowledge.”
· On the World Bank Voices blog – the role of public sector workers in developing countries and reasons to invest more in it “developing country governments are relatively understaffed….[and] face staffing shortages in core government functions… Due to an underdeveloped private sector in low- and middle-income countries, the public sector accounts for 38% of global formal employment and a sizeable 59% in FCV countries. Women represent 46% of the public sector paid workforce, compared to 33% in the private sector…[but] Poor skills often stem from a lack of merit-based and transparent recruitment.”
· Jason Kerwin argues for very short pre-analysis plans that say your plan for analyzing the data and nothing else.
· Lant Pritchett on how using a prosperity gap measure and an array of poverty lines that go much higher (up to P$21.5/day, or ten times the “dollar-a-day (P$2.15) can be used to provide a more inclusive and better measure of development progress. “The understanding of development cooperation before the advent of low-bar poverty lines was not “growth at all costs” rather, it was that when national development, including broad based economic growth, was achieved, one of the many goals that would be accomplished was the (near) eradication of extreme poverty (which is true, demonstrated here). But this was not symmetric. The (near) eradication of extreme poverty did not imply a country was developed.
An analogy is that sometimes people list as a side effect of the USA mission to go to the moon was the invention of Tang (an orange flavored drink). The response that going to the moon was a very cost-ineffective way of inventing Tang is just goofy. Sure, you could invent Tang at much lower cost, but Tang was a side effect of the main goal, going to the moon, and not a very central one. Psycho-social nudges bundled with cash transfers in Niger (discussed here) are the Tang of development economics. The four-fold transformation of national development is the moon landing.”
Join the Conversation