- On the future development blog, Jishnu Das discusses recent experiments on public-private provision of education in Liberia and Pakistan, takes on Bridge Academies, and highlights the importance of good measurement: “in Liberia, Romero et al. tracked students to ensure that schools could not “game” the evaluation by sending weaker children home: “We took great care to avoid differential attrition: Enumerators conducting student assessments participated in extra training on tracking and its importance, and dedicated generous time to tracking. Students were tracked to their homes and tested there when not available at school. Finding children who have left a school is like finding a needle in a haystack. In a country where only 42 percent have access to a cell phone, it’s heroism.”
- On Straight Talk on Evidence, James Heckman and co-authors get taken to task for torturing data to overstate findings in a 2014 Science article on the long-term effects of the Abecedarian ECD program. Specific criticisms on sample size (and its reporting) and multiple comparisons. Response and a rejoinder follow the post...
Randomization inference has been increasingly recommended as a way of analyzing data from randomized experiments, especially in samples with a small number of observations, with clustered randomization, or with high leverage (see for example Alwyn Young’s paper, and the books by Imbens and Rubin, and Gerber and Green). However, one of the barriers to widespread usage in development economics has been that, to date, no simple commands for implementing this in Stata have been available, requiring authors to program from scratch.
This has now changed with a new command ritest written by Simon Hess, a PhD student who I met just over a week ago at Goethe University in Frankfurt. This command is extremely simple to use, so I thought I would introduce it and share some tips after playing around with it a little. The Stata journal article is also now out.
How do I get this command?
Simply type findit ritest in Stata.
[edit: that will get the version from the Stata journal. However, to get the most recent version with a couple of bug fixes noted below, type
net describe ritest, from(https://raw.githubusercontent.com/simonheb/ritest/master/)
- Excellent VoxDev piece by Donaldson and Atkin on how high intra-country trade costs are in Ethiopia and Nigeria (and how they go about measuring this).
- Uri Simonsohn on why using a quadratic to test for a U-shaped relationship is a very bad idea and what to do instead.
- “Glasses askew and gray hair tousled, Scott Rozelle jumps into a corral filled with rubber balls and starts mixing it up with several toddlers”. So begins a feature in Science on Scott’s experiment in progress on parenting and early childhood education in China…including the challenges of keeping a control group in this setting “Rozelle says that when he sees kids in the randomly selected control villages “I often want to take them in my arms and move them to the treatment villages””.
- WDR 2018
Here is a familiar scenario for those running field experiments: You’re conducting a study with a treatment and a comparison arm and measuring your main outcomes with surveys and/or biomarker data collection, meaning that you need to contact the subjects (unlike, say, using administrative data tied to their national identity numbers) – preferably in person. You know that you will, inevitably, lose some subjects from both groups to follow-up: they will have moved, be temporarily away, refuse to answer, died, etc. In some of these cases there is nothing more you can do, but in others you can try harder: you can wait for them to come back and revisit; you can try to track them to their new location, etc. You can do this at different intensities (try really hard or not so much), different boundaries (for everyone in the study district, region, or country, but not for those farther away), and different samples (for everyone or for a random sub-sample).
Question: suppose that you decide that you have the budget to do everything you can to find those not interviewed during the first pass through the study areas (doesn’t matter if you have enough budget for a randomly chosen sub-sample or everyone), i.e. an intense tracking exercise to reduce the rate of attrition. In addition to everything else you can do to track subjects from both groups, you have a tool that you can use for those only in the treatment arm (say, your treatment was group-based therapy for teen mums and you think that the mentors for these groups may have key contact information for subjects who moved in the treatment group. There were no placebo groups in control, i.e. no counterpart mentors). Do you use this source to track subjects – even if it is only available for the treatment group?
- Let’s start with your approach to teaching development economics at the graduate level. The class when you taught David in 1999 was heavy on the agricultural household model and understanding micro development through different types of market failures. Most classes would involve in-depth discussion of one or at most two papers, with a student assigned most weeks to lead this discussion. There was a lot of discussion of the empirical methods in different papers, but no replication tasks and the only empirical work was as part of a term paper. How has your approach to teaching development changed (or not) since this time?
Try as I might, I have made little progress on changing my basic approach to teaching. The papers and topics have changed, but the essence of my graduate teaching remains the in-depth discussion of a paper or two each class. I’ve tried to expand the use of problem sets, and had a number of years of replication assignments. The first was hindered by my own inadequate energy (it’s hard making up decent questions!). I found that replication exercises required too much time and effort in data cleaning by students relative to their learning gain. Students were spending too much time cleaning, merging and recreating variables and too little time thinking about the ideas in the paper. I’ll reassess assigning replication this year, because there may now be enough well-documented replication datasets and programs available. With these as a starting point, it would be possible to get quickly into substantive issues in the context of a replication.
In the latest JEL, Parker and Todd survey the literature on Progresa/Oportunidades: some bits of interest to me included:
- CCTs have now been used in 60+ countries;
- over 100 papers have been published using the Progresa/Oportunidades data, with at least 787 hypotheses tested – multiple testing corrections don’t change the conclusions that the program had health and education effects, but do cast doubt on papers claiming impacts on gender issues and demographic outcomes;
- FN 16 which notes that at the individual level, there are significant differences in 32% of the 187 characteristics on which baseline balance is tested, with the authors arguing that this is because the large sample size leads to a tendency to reject the null at conventional levels – a point that seems inconsistent with use of the same significant levels for measuring treatment effects;
- Two decades later, we still don’t know whether Progresa led to more learning, just more years in school;
- One of the few negative impacts is an increase in deforestation in communities which received the CCT
- Dave Evans asks whether it matters which co-author submits a paper, and summarizes responses from several editors; he also gives a short summary of a panel on how to effectively communicate results to policymakers.