Chris has been threatening it for a while but looks like he’s finally done it (click through to see the full thread and description of why his answer’s not what you might expect):
If you can’t wait till then, much of his materials are in his class slides and syllabi on this very site.
And a really good related thread on criminal governance – when gangs replace functions of government – from Brazil. In this case the gangs extended from prison out into neighborhoods and actually made them safer. They set rules like not doing drugs in front of children, murders went down, and it became safer for government employees, like healthcare workers to come in and serve the community again. (Not to say this was altruistic – presumably, this is all in the service of making it easier to sell more drugs to people coming from outside the neighborhood.)
A new study found “spin” in abstracts and titles of British psychiatry (and one psychology) journal. When the main findings in an RCT were null in, authors would write the abstract or title to focus on other findings. One reason this is an issue is that in medicine, busy doctors often only read the title and abstract so might miss the null main finding (original research article).
I’ve mentioned before the study that found a major source of misleading news on health research wasn’t the news media overhyping the findings, but exaggerated claims given to them by university press releases. The same research team followed that up by going back to the source, with a clever and impressive randomized controlled trial of university press releases. They persuaded university press offices to allow them to intercept and (randomly) re-write press releases to more accurately reflect the research findings, such as adding caveats. They found the altered press releases resulted in more accurate coverage, with no reduction in likelihood to make it into news. Article here, but I’ll drop you into this mega thread from one of the authors covering both studies here – he describes the backlash from offended journalists over the first study, and how they had to persuade them and the press offices to agree to be randomized into this study.
The French Development Agency is loaning Benin 20 million euros to Benin for a museum to house artifacts looted by France (pending French parliamentary and legal challenges). Well, some of the artifacts at least:
The 26 artifacts, which include statutes and thrones looted by French troops during a military raid against the once powerful West African Kingdom of Dahomey in 1892, are among some of the 5,000 artifactsrequested from France by Benin.
And that’s just Benin, there are an estimated 90,000 looted African artifacts in France.
A holy grail in economic development, and really all of business investment, is figuring out which small businesses will grow when given the opportunity. A few years ago David McKenzie evaluated a very successful program in Nigeria involving an intense business plan competition with $50,000 for the winners. The program’s being copied by other countries, but David and Dario Sansone went back to use all the data collected to compare different methods of predicting later business success, including machine learning. The upshot? As their title suggests “Predicting entrepreneurial success is hard” (ungated here, and let’s hear it for clear titles!):
Business plan scores from judges, simple ad-hoc prediction models used by researchers, and machine learning approaches. We find that i) business plan scores from judges are uncorrelated with business survival, employment, sales, or profits three years later; ii) a few key characteristics of entrepreneurs such as gender, age, ability, and business sector do have some predictive power for future outcomes; iii) modern machine learning methods do not offer noticeable improvements; iv) the overall predictive power of all approaches is very low, highlighting the fundamental difficulty of picking competition winners.
Bonus: listen to David and Lariat Alhassan, one of the Nigerian business plan competition winners, on this Planet Money episode.
IPA isn’t all RCTs. Our young Right Fit Evidence Unit does M&E consulting to other orgs and governments to help them use data better, and is looking to hire an Engagement Manager and a Senior Associate, but is looking for with folks with different backgrounds than often end up at IPA. Management consulting, previous M&E work, or other research management could all be good backgrounds to work on help orgs use data better.
Dina Pomeranz and Paul Goldsmith-Pinkham are taking grad school & career advice questions and sharing them with the larger econ twitter community under the hashtag #EconAdvice. If you’d prefer to remain anonymous, feel free to DM them and they’ll remove identifiers before sharing.
Johannes Haushofer shares this really nice introduction for RAs who work with him on how to work productively together and what to expect. Note the emphasis on what they should expect from him and on not being afraid to ask.
Gray Kimbrough re-upped advice for econ grads applying to non-academic jobs
J-PAL North America has released a pretty spiffy-looking toolkit for doing evaluations in North America, covering conceptual things like assessing feasibility as well as technical things like power calculations, with links to code and more resources (most of the info isn’t specific to North America).
It is the policy of the American Economic Association to publish papers only if the data and code used in the analysis are clearly and precisely documented, and access to the data and code is clearly and precisely documented and is non-exclusive to the authors. Authors of accepted papers that contain empirical work, simulations, or experimental work must provide, prior to acceptance, information about the data, programs, and other details of the computations sufficient to permit replication, as well as information about access to data and programs.
Shameless plug for working with my colleagues at IPA, where there’s a dedicated data curation team that deidentifies and cleans the data, checks code, makes sure results are reproducible, and posts the data in AEA-approved archives.
Nature Human Behavior published an editorial alongside its first two registered reports, vetted (down to checking the code for at least one) and accepted before the studies were run. When the findings came in they contradicted what had previously been found in the literature.
Check out this interview with Brian Nosek on the Circle of Willis Podcast (iTunes) about his work with the Center for Open Science, trying to rethink how the scientific process could work if the incentives were to advance knowledge, rather than career pressure.
(BTW, after discovering I had 850 unlistened to podcast episodes, I recently declared podcast bankruptcy, deleted them all and am resubscribing judiciously. Circle of Willis – the name comes from a neuroanatomy thing – is one of the few I’m back on board with)
Also, one point I hadn’t thought much about, open access journals are free to access and the EU will be transitioning to requiring research they fund be published in open-access journals. But those journals often collect fees from authors, does that discriminate against researchers with less funding and from lower-income countries? Follow this thread for a discussion from Deborah Ghate & Duncan Green and calls for papers on the topic.
And I’m told some people might be using Sci-Hub to get to paywalled articles, make sure to avoid that or using any of these tips on how not to do it.
Cancer researcher Peter Bach points out how Novartis used anchoring to set price expectations high for a new immunotherapy, making a 2 million dollar drug price (much higher than other life-saving treatments) sound reasonable years in advance by consistently referring to the new treatments in development as “million dollar therapies.”
Duke University researchers apparently violated their IRB requirements by compiling a video database of thousands of students on their way to class to use in facial recognition development. The data was made available widely and has been used by companies in China, where AI and facial recognition have been used in state surveillance and repression (though it’s unclear the role of this specific dataset).
Noah Smith reviews studies showing the effectiveness of management consulting helping businesses in low-income countries. I’ve heard several people who’ve observed firms say that what they see missing is basic middle management skills (operations, marketing, etc.)
Andrew Foster, Dean Karlan, Edward Miguel, and Aleksandar Bogdanoski wrote a post on the IPA and World Bank Development Impact blogs reviewing what they’ve learned from the Journal of Development Economics’ experiment having a pre-results review track. Overall it seems to have gone well, and they’re making it permanent.
There’s been a dust-up online over this study, which tested encouragements/incentives for university students in Hong Kong to participate in anti-authoritarian protests and caused some surprise online (I believe it went through several university IRBs). (Update: Berk Ozler discusses the paper and how to potentially avoid misunderstandings around ethical considerations.)
For students or anybody dealing with busy people – how to manage up for working with your academic adviser – be aware of their time constraints, and be very specific on what you need (via Lindsay Page).
Jobs at IPA:
Chris Blattman is looking for an RA to work in Colombia studying gang politics (must be very good with local Colombian contexts).
Work on my team, in the policy group! We try to get evidence used around the world, making results understandable, supporting embedded research labs in governments, and lots of other stuff to make sure good findings get put into practice.
Thanks for being patient while the links were sleeping, expect some summer disruptions of schedule as well
Measles cases are up 300% over last year with outbreaks in the U.S., Europe, The Philippines, Myanmar, and several African countries. I heard a PSA that adults vaccinated before a certain period (when the vaccine process changed) might no longer be immune. So I got checked and sure enough I wasn’t, and had to get a new vaccination. My doctor said about half the people she tests have lost their immunity (though it’s obviously not a random sample). So if you’re planning summer travel, or even not, you might want to talk to your doc about getting checked (it’s a simple blood test).
An impressive pair of papers on a randomized controlled trial of a national program in Ghana to give disadvantaged kids meals at school (if I’m reading it right, it was a national policy and the order of communities where it was phased in was randomized). The research teams found big impacts on nutrition, growth, education, and cognitive performance, particularly among girls and more disadvantaged kids (via Justin Sandefur).
Liberians are protesting today against the government. Shipping containers with the equivalent of over $100 million of newly minted currency went missing from a port last year, and it’s not clear what happened to the $25 Million the U.S. put into the economy. The government is restricting access to the internet and social networks. (h/t Grace McLain)
A nice profile in WIRED of John Arnold, an admired energy trader who became the youngest billionaire in the U.S., then retired at the age of 38, and with his wife Laura, has since been supporting efforts to improve replication and reproducibility in science.
On the topic, the BITSS blog interviewed Irenaeus Wolff, one of the guest editors of a trial of pre-results review, at Experimental Economics, accepting papers based on study design, before the results are known.
Which you’ll be even sadder about once you hear the new podcast by Michael Lewis, Against the Rules, which looks at the referees of society, and how they’re increasingly under attack. The episode Baby Judge School (Apple) looks at the myth of robot-like impartial judges, the statistics showing systematic bias, and profiles a judge who left the bench to run a training academy teaching judges about combatting their own biases.
I’m still working my way through the episodes, but The Neutral (Apple) about Kenneth Feinberg, the go to guy called to (pro bono) run victim compensation funds after national tragedies like 9/11, and has to put values on the lives of the victims, was also great.
Ever wonder why conference audiences get sleepier later in the day? Astronomer Adam Ginsburg measured CO2 levels at a conference and found that the room full of breathing warm people heated up and went past recommended levels of CO2 pretty quickly, but dropped during a coffee break when organizers opened doors and windows. So if people are falling asleep while you’re on your second set of supplementary slides going through regression coefficients, it’s probably because of the CO2. Read the thread below or the Washington Post write-up. (h/t John Branch)
IPA’s looking for a Director of Poverty Measurement. In particular the job involves overseeing the Poverty Probability Index, a short, country-specific tool practitioners use to estimate poverty rates, and developing new non-monetary measures (requires strong quant background). Please share with anybody who might be interested.
How the government of Odisha, one of India’s poorest states, moved a million people out of the path of a cyclone.
Can a whole field of researchers be wrong? On Slate Star Codex Scott Alexander describes how an initial paper on how the gene 5-HTTLPR might relate to depression spawned decades of research and hundreds of papers into the gene’s relationship with all manner of disorders and parts of the brain. But now a group of geneticists has written a blistering reprimand based on a genetic database of more than 600,000 people. The authors of the new paper argue that it’s impossible for that one gene to play such a massive deterministic role, given how multiply causal gene-disorder relationships are. If they’re right, what does that mean for the research process? Alexander argues:
First, what bothers me isn’t just that people said 5-HTTLPR mattered and it didn’t. It’s that we built whole imaginary edifices, whole castles in the air on top of this idea of 5-HTTLPR mattering. We “figured out” how 5-HTTLPR exerted its effects, what parts of the brain it was active in, what sorts of things it interacted with, how its effects were enhanced or suppressed by the effects of other imaginary depression genes. This isn’t just an explorer coming back from the Orient and claiming there are unicorns there. It’s the explorer describing the life cycle of unicorns, what unicorns eat, all the different subspecies of unicorn, which cuts of unicorn meat are tastiest, and a blow-by-blow account of a wrestling match between unicorns and Bigfoot.
There will be another pandemic. We don’t know the date or how many will be infected, but the extent of the latter is largely up to us. CGD’s Jeremy Konyndyk reviewed responses to the 2014-2015 Ebola outbreak, particularly from the U.S. government and makes recommendations. In particular he considers the political realities of different response measures.
Health officials are warning that the current Ebola outbreak in Congo risks expanding past the 1,000 already dead, and infection rates accelerating as health and safe burial teams are violently attacked. While there was good news of an experimental vaccine, there are far fewer doses than needed, and an infection hotspot is near the Rwandan border, with a lot of back-and-forth traffic. And as if all that wasn’t bad enough:
Arthur said mistrust of outsiders is common in North Kivu and a social media disinformation campaign has led many to believe that the Ebola scare is a hoax or that vaccinations actually cause the disease.
For Arthur Brooks’ podcast on productive disagreements he interviewed psychologist John Gottman (Apple), a world expert in how couples get along and argue. One of his main findings, disagreement, even argument, can be productive, but contempt is destructive. It’s a good conversation (also useful tips if you’re in a relationship). They summarize four rules, which I’ll copy from the Gottman Institute’s summary of the conversation:
Rule 1: Focus on other people’s distress and focus on it empathetically Empathy is a cornerstone of emotional intelligence, an essential quality for successful relationships. You don’t have to agree with someone to empathize with them.
Rule 2: Keep your positive vs negative comments and interactions at a ratio of 5:1 You have power to do this. The positive things you say versus the criticisms that you level should be at a 5:1 ratio at least. That means five affirming, praising, and loving tweets and Facebook comments for every critical one.
Rule 3: Avoid contempt with everybody, all the time No exceptions. It’s bad for you and it’s bad for the country if you treat anybody with contempt.
Rule 4: Learn to cooperate and have dialogue with those of whom you disagree Seek out and be around people who are different than you are. Before you speak, see if you understand what the speaker before you has said. Listen to understand, and then frame your rebuttal.
Nudge approaches have had success showing that you can encourage a behavior by making it easy, but the opposite is true also. “Sludge” is when a desired behavior is discouraged by making it hard, like confusing forms, or Arkansas requiring poor people receiving benefits to report their work hours on a website, but closing the website every night from 9PM to 7AM. Cass Sunstein proposes government “Sludge Audits” to find these inefficiencies in their processes.
A few years ago, the “Worm Wars” broke out when a team reanalyzed data from a classic finding on the benefits of treating kids’ intestinal parasites and failed to reach the same conclusion. Owen Ozier reflects back on what it means for replication in a new paper and explanatory tweetstorm.
Great article on the history of the U.S. Census, and how the need to count the growing U.S. population faster spurred technological innovation. If you want to follow current developments on adding a citizenship question to the census, I’d encourage you to follow NPR’s Hansi Lo Wang.
The government has most of your tax information (that’s how they know if you’re lying), and in some countries doing your taxes takes a couple minutes b/c the government fills out the form for you and asks if it looks right or if you want to make adjustments. But in the U.S., tax prep companies like Intuit have lobbied hard to keep the process complicated, so people have to use software to navigate it, effectively creating a tax on doing your taxes. A bill recently passed the House preventing the IRS from making it easier to file. One argument the tax prep companies have used is that, through an agreement with the IRS, they already offer and advertise a free version that most Americans are eligible for (anybody making under $66k). But only 3% of those eligible actually use the free version. Why? ProPublica has been doing some bang-up reporting on how tax companies have effectively hidden the free versions of their software that they’d agreed to offer. Among other things, they prevented the websites for the free versions from being indexed by google, while using fake versions to make it look like customers had found the free version, but then were redirected to the paid version, and directed staffers not to tell customers about the free version. Thanks to their reporting, the bill seems to have been stalled in the Senate.
A good Planet Money episode on the Stanford law professor who nearly got California to offer a pre-filled state tax version (in the pilot it was very popular).
Two African mobile health startups have each won $1.5 million from the Skoll awards: Uganda’s mPedigree lets customers confirm the authenticity of the medicine and agricultural products they’re buying with their phones (counterfeits are rampant in both markets). In Ghana, mPharma helps pharmacies with the drug supply chain to make sure medicines are stocked and priced appropriately.
This was interesting: Investment company Vanguard developed a method to avoid paying taxes on ETFs. I’m not sure I understand it but it involves temporarily borrowing stocks from friendly banks for a day or two when they have to make a payout to investors. Their method of not paying taxes is so clever they patented it.
Two more former students have joined the lawsuit against Dartmouth, with really disturbing allegations of rape and sexual coercion by prominent professors using threats against the students’ future academic careers.
A great article on how the world came to slowly realize how pervasive mental health problems are in the developing world, and some efforts to do something about it.
I had the chance to hear Dr. Dixon Chibanda, whose work is profiled in the piece, talk a while back, one thing he mentioned was how difficult it is to get research grants, because local researchers like him are often not included in research, so they don’t get publications, and it becomes a self-reinforcing cycle. (You can follow him on twitter)
The Research Meets Africa conference taking place in Ouagadougou in October has a call for papers. They particularly encourage financial inclusion researchers who want to collaborate with African researchers to submit. Deadline is May 15th.
Jobs postings: Busara (in Kenya, Nigeria, & India), EPoD at Harvard, & IOM doing impact evals in West Africa.
A wonderful back and forth between David Evans and DFID Deputy Chief Economist Nick Lea, ostensibly about regressions, but to me resonated more broadly on methods. Papers seem to have to need the magical pixie dust of a regression to get accepted for publication, but is it the case that every problem in development is a nail waiting for a regression hammer? Lea wonders if methods are constraining the kinds of questions economists ask. See his thoughtful response to David’s post here.
I’m continually stunned by how prevalent intimate partner violence is in places where development economists work, and how under-studied it is in development. For example, the WHO estimates over 50% of women sampled in Uganda have experienced that kind of violence. Seems like that would have as big an impact on people’s daily life as plenty of more commonly studied topics. The good news is that there’s reason to think plenty of things that researchers do work on – education, livelihoods, skills training – may also reduce violence at home. So IPA is offering money to add on research on intimate partner violence to existing or planned studies, and to do research on how to measure it. More information here, deadline May 17, and please share.
What if everything we’ve been told about the giant impacts of early childhood interventions is wrong? The conventional “Heckman Curve” wisdom argues that the earlier the childhood intervention, the higher the returns, but I’ve also heard child development folks quietly say that interventions work at all ages and we shouldn’t privilege any one window. Andrew Gelman discusses a meta analysis that indeed disputes the early childhood window idea. He suggests original analysis may be heavily skewed by two small and unusual studies.
Johannes Haushofer explains his new paper with Ingvild Almås and Jeremy Shapiro about whether a calorie-based poverty trap exists. Do people not get enough food to work effectively, which keeps them poor and not getting enough calories? Using GiveDirectly cash transfer experiment data, they don’t think so.
As much as it pains me to link to both David *and* my other Friday links competitor, Tim Ogden of NYU’s faiV, (which focuses on financial inclusion) he’s got a really good piece on CGAP’s blog. It’s ostensibly on what can we expect to learn from financial inclusion research, but really about systematic reviews and meta-analyses in general, and how we’re limited by the scope of very specific studies, and lack of standardized reporting. Studies are often limited in scope to begin with (for instance, analyzing effects of a financial product on individual users, but not spillovers on the economy as a whole), then once you start reducing and reducing to just what’s common among studies AND reported in a way that’s comparable, you’ve limited the scope of what can be concluded. Standardized reporting might be a helpful solution.
The National Academies had a task force of big brains assigned to figure out how to cut U.S. child poverty in half in a decade (which the U.K. did between 2001-2008) . Here’s a good summary of what they recommended. Even shorter highlights:
Without current programs, child poverty would be higher than it is now, so we’re already helping
Just expanding two existing programs, the Earned Income Tax Credit and Child and Dependent Care Tax Credit would do half the job by themselves
Reducing child poverty would likely save the country substantial amounts of money in the long run through increased employment, lowered healthcare costs, and reduced incarceration
If you want to know what Chris is teaching now, see his lecture slides from the first couple weeks of his Order and Violence class.
From the Arnold Foundation, programs which have positive RCT results may have their effects fade over time, which you never know if you don’t do long-term follow-up. This is similar to what Chris, Nathan Fiala, and Sebasian Martinez found with IPA in Uganda doing a 9-year follow up of a cash grant/transfers program. Grantees who got $400 increased their earnings for a number of years (compared to a control group which didn’t), but by 9 years out, the control group had caught up and had similar earnings. (All those intermediate years of increased earnings were more than the amount they received, so it worked, but we should be careful about extrapolating beyond the time period for which we have data.) h/t Marc Gunther
Managing Director for Teaching at the Right Level, a multi-organization collaboration (including J-PAL, IPA, Pratham, Young1ove and others), aiming to help scale a proven education program to reach three million students in several African countries over the next five years.
First, he’s back! David Evans, ensconced in his new digs at the Center for Global Development, brings us a roundup of over 275 papers from the Center for the Study of African Economies (CSAE) conference, in a fancy new expandable format indexed by topic. (Honestly it’s probably better than going to the conference to have someone review all those papers and give you the Cliffs Notes).
It reminded me of this study, which looked U.S. media coverage of disasters abroad, and how it influenced U.S. responses (for 5,000 disasters from 1968-2002). First, it found that more dramatic disaster deaths got coverage more easily – it took one death from a volcano to make the news, two from an earthquake, and 38,920 if from hunger. And, U.S. officials were less likely to declare a disaster if other things were in the news, like an Olympics.
ICYMI this video has been tearing up the academic conversations. Psych PhD student Mike Morrison thought about how to redesign conference posters changing the starting point from mini-article on a wall, and using usability psychology to find a design that makes it easy for passersby to quickly get the message and decide if they want to stop and learn more. Here it is: Punchline in plain English in the middle in big type, supporting tables on the right, basics of the study (original poster content), bullet pointed on the left, with a QR code (just a link shortener that you can google to easily make). People can then photograph the QR code on any phone to get the link to the full paper.
Here’s an actual example converted from traditional “wall of text” format to a new one
It’s been a while since we had a podcast roundup, some great ones from the last few months I’ve stumbled across:
Planet Money did a great 3 parter (I know, sounds like a commitment, but it’s good), on anti-trust law – how do we decide as a group when bigger companies are good for the economy and when too big is bad? There’s an interesting history, first involving Standard Oil. The second episode to me was the most interesting, on how one law student, who took some classes in the University of Chicago econ dept, caused a virtual halt to anti-trust enforcement. And finally, what happens when tech giants like Amazon sell products and also control the marketplace for other sellers?
On the Financial Times Alphachat, Leah Platt Boustan of Princeton, and Margaret Peters of UCLA (Apple, or find Alphachat in your podcast app ) give a historical overview of immigration to the U.S. (and Europe a bit), and go into the findings on job competition between native and immigrant workers. A really good primer if you want to get some rational and data-based insights into the debate.
And for something fun, the podcast Good One takes a good joke and sits down with the comedian to understand how they came up with it. (It’s worth listening to the recent Gary Gulman one just for the great Trader Joe’s story they play before discussing, and definitely google for his state abbreviation bit if you haven’t seen it. The Good Place writers panel was great too if you watch the show. And if you don’t, have I got a recommendation for you!)
Scott Cunningham put out a call for advice on teaching the secrets of RCT methods and people responded with readings, syllabi, and other suggestions. A good thread to bookmark if you think you ever might be teaching it.
IPA’s Peace and Recovery initiative, led on the academic side by Chris, has an open call for funding. We define peace and recovery pretty broadly:
Reducing violence and promoting peace
Reducing “fragility” (i.e. fostering state capability and institutions of decision-making)
Preventing, coping with, and recovering from crises (focusing on conflict, but also including non-conflict humanitarian crises),
It also funds a variety of types of work, including pilots and exploratory work, specifically earmarked for junior researchers (PhD student through untenured professors). See page 1 of this doc explaining the purpose of the fund. More details here, examples of currently funded studies here. Deadline for expressions of interest is March 15th.
And if you’re in Chicago, on Monday see Dr. Rebecca Wolfe from Mercy Corps talk at UChicago about one funded project, from Nigeria, on whether playing communities audio recordings of former Boko Haram members apologizing for what they’ve done, can help with reintegration.
The World Bank’s Pandemic Emergency Financing Facility (PEF) has paid out its second payment, $20 million, for Ebola response in the DRC. As I understand this is a relatively recent funding mechanism (from 2017), developed after the slow response to the 2014 Ebola outbreak. Instead of waiting for a disease to spread, then gathering funders together to cobble together funding commitments, this is funded more like insurance. The money is funded in advance by a bond, and then pays out immediately when particular criteria are met.
A nice podcast conversation between Diane Rehm and journalist Francisco Toro explaining the situation, history, and politics of the crisis in Venezuela.
Jonathan Morduch explains positive early results from Grameen America’s RCT on a microcredit program in New Jersey (summary: the 6-month results look promising on satisfaction, some financial indicators, and business creation but it’s early). Full report link here or in his last tweet.
Happy International Women’s Day:
Here’s Justin’s one tweet summary of Eble and Hu’s research on gender attitudes and bias in Chinese schools:
I just learned about Lise Meitner, the Jewish refugee who overcame many barriers to women in academia, barely escaped Germany in WWII, and went on to co-discover nuclear fission. Her colleague published the paper without her name, and was awarded the Nobel Prize alone. Read about her fascinating life from Brain Pickings.
A lovely tribute to Dave Evans, who’s been a boon to the field, and a prolific producer of public goods, from David McKenzie and his Development Impact Blog colleagues
I ran a quick search, and I’ve cited him about 50 times in my links
It’s fitting that Dave’s final Dev Impact post is in one of his specialities, making research more understandable to non-researchers, in this case for education. While researchers often report learning outcome changes in standard deviations, he describes his new paper with Fei Yuan on how to express outcomes in increased years of schooling – a much more intuitive measure.
Whatever you think of the replication crisis/credibility revolution, the rubber meets the road on good scientific practices when it comes to medical trials. A number of top medical journals have signed onto the CONSORT guidelines for publications, requiring hypotheses to be prespecified. But what good are they if nobody goes back and compares the publication to the registration? Well, when a team compared the actually published studies in a number of top medical journals to the pre-registered hypotheses, they found 25% of the studies had switched outcomes – reporting different ones than they’d originally identified (outcome switching might mean reporting quality of life instead of overall survival, for example, even if survival had been the original goal). Almost half dropped secondary outcomes, and others added new outcomes. When the research team wrote to the journals (like JAMA, BMJ, and NEJM) fewer than half their letters were published, suggesting that top journals aren’t adhering to their own guidelines, but also that there’s no mechanism for checking and enforcing the guidelines.
The University of California, Berkeley is ending their contract with Elsevier journals after negotiations (co-chaired by chief librarian, informational science professor, and MIT-trained economist Jeff MacKie-Mason), failed to find an agreement on pricing for journals and also making publications open-access.
Their tips for finding alternative ways to access paywalled articles are helpful for all non-academics looking for paywalled articles
In econ you can often find a version on the author’s personal website, or by googling the title for a working paper version. If you can’t, try emailing an author to request a copy.
A new group of prominent economists has formed the network Economists for Inclusive Prosperity, publishing a set of concrete, research-based policy suggestions for addressing problems of wealthy economies, such as financial system stability, taxation, labor markets in the age of AI and automation, and the like. You can read a summary in the Boston Review from Suresh Naidu, Dani Rodrik, and Gabriel Zucman.
I enjoyed the conversation between Tyler Cowen and Daniel Kahneman (Apple). It might be the accent, but I feel like you get a sense of the wisdom of a lifetime of studying human intuition and life satisfaction when you hear him reflect. Two things that struck me were how often he passed on questions – it seemed like if he wasn’t an expert on a topic, he didn’t feel like his opinion was worth more than anybody else’s – and how often he discouraged Cowen from labeling parts of human nature as biases.
The Indicator, from NPR’s Planet Money, interviewed economist Nina Banks about Sadie Alexander, the first African-American economist, who held both a Ph.D. and JD from Penn (Apple). For more, see Cardiff’s longer interview with Banks when he was at the Financial Times Alphachat (Apple). And Banks has two books coming out on Alexander – a biography and edited volume of her speeches.
The Sadie Collective, which encourages Black women to enter economics and related fields recently held their first conference (and I believe was crowded to capacity). You can watch the video here.
And they’ve started their their new course with Josh Angrist on basics of econometrics. It’s difficult to make an intro video about econometrics, but I have a lot of respect for the production quality and effort they’ve put into it. Here’s one on Ceteris Paribus and counterfactuals (if it moves too slowly you can always speed up youtube videos in the settings):
Berk Ozler counts the numbers of men vs. women asking questions during a seminar speaker’s talk, and guess how the ratio came out (it’s worth also checking out the discussion below, including a code of conduct being considered at one department).
In a follow-up to his informational intervention, he found a few days later the ratio changed for Seema Jayachandran’s talk there
Watch Seema’s talk on gender preferences in India here. (And note the all-star list of previous talks. Tip: you want to take them to go, podcast-style, there are any number of YouTube -> mp3/mp4 converters that will let you download to listen to offline)
Also, a really great conference video from Penn on Child Poverty and Brain Development in Global Context. A nice combination of economist, education folks, and child brain development researchers discussing what’s known about how poverty interacts with cognitive development.
How his growing up in India helps inform his research topics, but that also collaborations with researchers from other countries bring fresh eyes to things he always took for granted.
The relative effect sizes of behavioral interventions compared to others, and why standard reporting can make it hard to compare across interventions.
On the latter, I’d also encourage researchers to collect cost data (here are a primer and templates to use from J-PAL). That lets people who come later compare cost effectiveness of different programs on standard, understandable scales. When you take the extra step of reporting in terms that are more understandable, your research can have a much bigger impact (this from that Penn conference)
One note, the route wasn’t preachy, in fact a version that emphasized legal punishment for men failed. The way which it seemed to work was peripherally, by changing the perception of public norms about reporting and talking about domestic violence (more in this tweetstorm).
A favor if you’re in an org that uses monitoring and evaluation data – some friends at LSE are conducting a survey of how this data is produced and used. If you could take or pass it along to colleagues we’d appreciate it.
Marginal Revolution University announced a new series profiling inspiring women in economics. At the bottom of the page you can sign up on the website for when the videos profiling their work are released (seems like a great teaching tool). There’s also a form to suggest an economist to profile who has inspired you.
This was a beautiful and also sad story about designing a school in Oklahoma specifically for homeless kids. And a reminder to talk to the actual users of your program/product. They started by asking homeless kids what they wanted:
One of the requests was just to have space where they could hang out with friends. “We have kids who don’t get to go to playdates. They don’t do birthday parties,” says Agel. At public schools, homeless children are typically left out of the social calendar of sleepovers or visiting friends’ houses, in part because they have nowhere to host friends themselves.
Rachel Strohm’s Africa Update newsletter has tons of great news stories, research findings, and fellowship opportunities (subscribe at the very bottom).
In the context of the discussion around the new World Bank head, Paul Romer has an op-ed in the Financial Times (ungated version here) where he suggests two improvements to the World Bank’s process: Insulating their agenda-setting from politics and diplomacy, and building future thinking into infrastructure – building capacity for where the population will be five years from now, not just at the moment.
GiveWell is expanding its interests beyond the narrowly measurable RCTed interventions of the absolute best deals, and is also expanding into government advocacy.
When a working professional tells me, “I’m thinking of getting a Ph.D.,” they usually mean, “I’m thinking it would be nice to have a Ph.D.” A Ph.D. is sort of like the perfect gym body. The having is nice, but the getting can be painful—and once you’re there, you never stop.
A fun and helpful discussion from Goats and Soda with researchers about how cultures around the world linguistically label and think about different sub-types of anger (in India for example, political anger vs. anger against a loved one), and how to use that as a technique for dealing with your own anger. (I propose a new word for “online outrage”)
First, congratulations to Dave Evans, everybody’s favorite public good generator, on his upcoming move to the Center for Global Development, where he’ll join an impressive bench, including Pam Jakiela, Susannah Hares, and Kristaps Porzingis. (And subscribe to his blog at that link for great book reviews, and other interesting stuff.)
Jan 30th was Fred Korematsu Day, named for the Japanese-American who fought WWII internment camps up to the Supreme Court and lost. One of my favorite podcasts, More Perfect (Radiolab about famous Supreme Court cases), did an amazing profile of Korematsu (the man and the case), and explained why some legal scholars still think the case was decided correctly.
There’s been a lot of news on the proposal for a minimum income guarantee in India. According to this article (h/t Justin Sandefur) Piketty and Deaton are advising the government on it. But in a recent BBC Worldview interview, Tavneet Suri says she thinks the technical capability is largely in place, but they might want to slow down and study the actual economic effects (audio below, and links continue below that):
A cautionary tale (h/t my IPA colleague in Burkina Faso, Aliou Baguissa Diallo) about measuring spillovers in cash transfers. In the Philippines, Filmer, Friedman, Kandpal, and Onishi found that a cash transfer targeted to the poor helped them, but raised the price of nutritious food in the area. The children of the people who got the transfers ate better themselves, but the general rates of growth stunting for everybody else, went up 11 percentage points! Mothers and children who didn’t get the cash also used less healthcare. As Berk has pointed out before, if you help a few people but potentially hurt far more, you’re doing the opposite of what you intended.
The latest online scam targets academics, with an email purporting to be from a department chair who’s stuck in a meeting, but needs them to go out and buy a lot of gift cards, right away. Yale School of Management economists Florian Ederer and Jason Abaluck decided to see how committed the scammer was. Turns out, pretty committed:
Rohini Pande, Vestal McIntyre and Lucy Page write in the New York Times about a paradox in aid. As big diverse countries like India and Nigeria move from low-income to middle-income, they’ll still have the greatest overall number of poor people. But development aid is scaled back as countries get richer, leaving the greatest number of poor people with the least aid. It’s sort of reminiscent of social safety net programs that cut off as soon as people earn a little part-time income, making them worse off overall.
And to compliment last week’s Mongolian heavy metal band, here’s the Indonesian teen girls’ heavy metal band Voice of Baceport (Google them and you’ll find a lot of articles). Notice the guitar’s sound cuts out two minutes into the show and they carry on really well till it gets fixed.
Oxfam releases a report around the same time as Davos every year on who owns what portion of global wealth. Their spin on it is designed to make headlines, but Dylan Matthews explains why it’s really hard to measure.
About sixty percent of people in sub-Saharan Africa live in rural areas and many of them are dependent on farming, but the low amount of food those farmers get out of the ground compared to other regions of the world has been a particular puzzle. A all-star team in VoxDev summarizes what we know about improving agricultural output through experiments with ag extension workers trying to help farmers be more productive.
But not so fast, economists – Gollin & Udry have a new paper using panel data from Uganda and Tanzania suggesting that even measuring how much farmers are producing (or can) is harder than we think:
We find that measurement error and [unobserved] heterogeneity [in inputs] together account for a large fraction – as much as ninety percent — of the dispersion in measured productivity. In contrast to some previous estimates, we suggest that the potential for efficiency gains through reallocation of land across farms and farmers may be relatively modest.
An all-star team of behavioral scientists, led by Katy Milkman and Angela Duckworth, are running a nationwide experiment on sticking to exercise regimens. Members of the 24-Hour Fitness chain can join here by Jan 31.
I’m usually a bit tempered in excitement about “nudge” style research – often the behavior change is statistically significant but small or short-lived (in my experience). The exception seems changing the default option – that can be huge, and a new meta-analysis quantifies the effects over 58 studies and find they can be substantial, but not all defaults are equal:
Our analysis reveals two factors that partially account for the variability in defaults’ effectiveness. First, we find that defaults in consumer domains are more effective and in environmental domains are less effective. Second, we find that defaults are more effective when they operate through endorsement (defaults that are seen as conveying what the choice architect thinks the decision-maker should do) or endowment (defaults that are seen as reflecting the status quo).
The RISE Programme conference on improving global education is accepting submissions, deadline March 15th.
My IPA colleagues have a series of blog posts about our experience moving evidence into policy. The first lays out the org’s strategic ambition for what we plan on doing differently over the next several years. The second is on how to get non-research-oriented partners (like governments and NGOs) involved in the research process from the start to make sure they have ownership and the questions address their needs. The third is about what to do after you have the findings, to make sure they get used and don’t just languish in a report on a website.
Brilliant coverage of Brexit from a Financial Times Southern Africa correspondent writing about it the way foreign correspondents cover African politics.
University of Virginia economist and public policy prof Sally Hudson is running for the State House of Representatives there, challenging the incumbent Democrat in the primary. This is not an endorsement (I don’t know enough about the candidates), but it did remind me of something I saw Dartmouth economist & public policy prof, journalist, speechwriter, and former political candidate Charlie Wheelan say. Talking to a bunch of policy students, he said it’s unfortunate that the people who are really good at policy usually don’t have the stomach for politics and the personalities who are successful in politics usually aren’t the types who want to get into the weeds on policy.
I haven’t listened to all the episodes of the IRC & Vox podcast Displaced yet (I discovered on my phone I have 836 podcast episodes I’ve individually chosen and downloaded but not yet listened to), but I have yet to hear a bad episode of that show. One of the many interesting parts of the interview with Reshma Saujani, founder of Girls Who Code, about innovation and failure, comes about 43 minutes in, where she talks about how she tries to surround herself with reminders of failing to make it a normal part of daily life and nothing to fear. One way she does this is keeping the rejection letter from her attempt to get on her local community board on her refrigerator. She has a book coming out in a few weeks on the topic.
ICYMI there was a fascinating and troubling discussion on Twitter this week following the Nairobi attacks, where the New York Times apparently showed (I didn’t look) graphic photos of dead bodies, even though they wouldn’t have done it for a U.S.-based school shooting or terror attack for example (even the death of U.S. soldiers is usually communicated through returning caskets with flags rather than bodies on the battlefield.) I’ll refer you to Jeffrey Paller’s great as always This Week in Africa newsletter for links to the specific arguments and defenses from the Times but it’s probably Ken Opalo’s words that stuck with me the most:
The Innovation Growth Lab conference on entrepreneurship and business growth is looking for submissions of RCT/field experiments (deadline Feb 11th).
The African Institute for Development Policy is offering a policy communication fellowship for Ph.D. students in their 3rd to 5th years from a list of 20ish countries studying some specific topics related to gender and health.
And if you have’t seen it, this AeroMexico commercial came out last year apparently, but it’s just making the rounds now*:
* don’t take the science of genetic ancestry too seriously (h/t Tim Ogden).
Hope everybody’s off to a great new year, and good luck to all the job candidates interviewing at ASSA. Also, remember from the last links that Ben Casselman, who’s been co-reporting on sexual harassment in economics for the New York Times, is there and happy to meet confidentially with anybody who wants to tell him about their experience. If you’re not on twitter, feel free to email me and I’ll put you in touch with him (confidentially of course).
At the Data Colada blog Uri Simonsohn realizes that publishing articles with links online (such as to news articles) is problematic as links die or change over time. He reviewed links in his articles since 2005 and found over half didn’t work anymore, and recommends a simple fix: Instead of the direct URL, link to the Internet Archive version of it.
Preanalysis plans (PAPs), where you specify your analysis before you see the data, can be a bit controversial. Some say a PAP ties your hands and prevents you from exploring things you might only find out about later. Psychologist and open science advocate Sanjay Srivastava offers six strategies which you can use instead of or alongside a PAP to allow for more flexible analysis without letting you fool yourself.
Daniel Kahneman explains why he’s become less interested in understanding happiness and more in how to live a satisfying life (if the article’s intermittently gated, try in your browser’s incognito mode). You can take Yale’s popular course on happiness, life satisfaction, and well-being from Professor Laurie Santos online for free and decide for yourself. A couple interesting observations from Kahneman, about British economist Richard Layard who started paying attention to the research and bringing improving overall happiness and well-being into policy there:
“The involvement of economists like Layard and Deaton made this issue more respectable,” Kahneman added with a smile. “Psychologists aren’t listened to so much. But when economists get involved, everything becomes more serious, and research on happiness gradually caught the attention of policy-making organizations.
“Much of Layard’s activity on behalf of happiness in England related to bolstering the mental health system. In general, if you want to reduce suffering, mental health is a good place to start – because the extent of illness is enormous and the intensity of the distress doesn’t allow for any talk of happiness.”
Given how common we’re discovering this is, it’s likely Fryer’s not the only one in economics. You can report bad behavior to the reporters, Ben Casselman and Jim Tankersly, who guarantee anonymity and will be at ASSA (I can confidentially put you in touch with Ben if you’re not on Twitter).
If you’re having career issues because of harassment, many senior faculty will help, including Jennifer Doleac, who’s offered to assist people in this situation navigate what to do next and connect them with people who can help.
You can also report to the NSF if you know of any sexual harassment by a PI on one of their grants.
Palm oil may be the worlds most hated product for destroying rainforests, and it’s in everything. But if economics is about anything, it’s about tradeoffs. In Ryan Edwards’ job market paper he looks at Indonesia, where he estimates the rapid expansion of palm oil exports since 2000 led to 2.7 percentage point faster poverty reduction and 4% faster consumption growth, at the cost of more rapid forest loss and more fire. A back of the envelope calculation finds 2.6 of 10 million Indonesians lifted from poverty this century were because of palm oil
Tim Ogden hosted an all-star cast for a discussion on microcredit, and how to think about how it interacts with the rest of developing economies. You can see the recording here.
Job: I believe IPA will be interviewing at ASSA for the new Ph.D.-level lead position using our org’s scale (over 200 RCTs happening now around the world), to develop new methods. It’ll be based in New York or DC but will involve working closely with Andrew Dillon at Northwestern and our network of PIs. Please pass it along if you know anybody who might be interested.
If you get this Friday AM, last I heard there were a few slots left for the webinar this afternoon on the latest thinking on microcredit/microloans (depending on what field you’re coming from). It’s at 1PM (US Eastern Time) from Tim Ogden at NYU’s Financial Access Initiative, featuring Gisella Kagy, Cynthia Kinnan, Karthik Muralidharan and Bruce Wydick.
Two great job market papers:
Amazing work by Anne Karing of Berkeley working with my IPA colleagues in Sierra Leone. Vaccinations have to be done in a sequence over a child’s first year, and it’s hard enough for people in rich countries to remember and keep up with it, let alone somewhere with scarce resources and lots of travel required. It required a massive amount of work upgrading the way health records are kept locally on top of the experimental work, but she tested the effects of handing out simple, color coded silicone bracelets to some mothers that publicly showed where their children were in their vaccination schedules. If I’m reading it right, the bracelets advertised to other mothers in the community that these mothers were keeping on schedule, and that influenced them to do the same. The simple bracelets increased vaccination rates by up to 14 percentage points which is a huge bump from a simple social signal.
A really cool paper from Meera Mahadevan at Michigan, who looked at elections in a large state in India, and then what the constituents of winners were then billed for electricity to nighttime satellite images of what they were actually using. Magically, the constituents of the winners of elections were later billed less for electricity than what the satellites showed they were actually using.
The latest Freakonomics episode features work by Gharad Bryan, James Choi, and Dean Karlan (with the voices of the latter two), on testing the effects of the religious part of a religious aid program for very poor people in the Philippines. It turned out the program worked better with the religious component than without, boosting earnings. The larger episode is about the Protestant work ethic and if its effects are real and measurable. (Apple)
A nice thread from John Holbein on teaching analysis of policies in his class. Every class group analyzing a change in a public policy found zero effect, and he reminds us that journals full of positive results condition us to expect something different than reality. He says we need more of a culture around precise nulls. (He also includes his class syllabus)