Andrew Foster

@YujungHwang3 Years ago I would tell students not to use a procedure they didn't fully understood or couldn't code. Of course, overtime we tended to be more trusting of packages and the theory underlying them. It saves time, but we lost something too. Seems like that rule should apply here.

@YanagizawaD Years ago I would tell students not to use a procedure they didn't fully understood or couldn't code. Of course, overtime we tended to be more trusting of packages and the theory underlying them. It saved a lot of time but we lost something too. History does tend to rhyme.

A new project. RQ: Can we automate policy evaluation? Not today, obviously. But maybe soon. To reliably, cheaply and quickly figure out what policies work and don't work, seems potentially super valuable to society. ape.socialcatalystlab.org

@JohnHCochrane Comments tend to be a lot of work and may be low impact. But every paper is in a way a comment on the literature and I have had some success in asking authors to turn "comments" into good "short papers" that can then go through a normal review process.

It's an explicit policy. I once submitted a 10 page paper showing a mistake in a top 5 pub. And a general misunderstanding typical of a literature, not just algebra of one paper. Response "we only publish original research." I think it's considered impolite to debate, who gets published ore important than what's right.

@leightjessica Thanks. Will consider this…

@afosterri It generally prompts me to write a more focused report, + presumably (hopefully) this has ancillary benefits for the speed / efficiency of the overall process

Over the last couple years I've been running a mini-experiment to see how many different journals I can serve as a referee for. The cumulative count is 80 + here to report this has been a fun experiment, 100% endorse

@FareedZakaria @TheEconomist @zannymb @RayDalio Beddoes was typically insightful. Dalio sounded a bit like he was in a hostage situation.

Will President Trump fight another day with Fed Chair Jerome Powell, and will it shake markets each time? That and more in Part 2 of my interview with @TheEconomist editor-in-chief @zannymb and Bridgewater founder @RayDalio:

@I4Replication COPE has guidelines on this: publicationethics.org/guidance/flowc…

We have no insights or knowledge about the practice of withdrawing author names. This is new to us. Some journals are also confused and trying to gather information.

@paulnovosad I asked chatgpt if there had been any useful comments on the original article. It pointed to the datacolada post. I think serious adjudication increases the incentives for quality replications; it is not clear the "comment" approach does that. AI can then take it from there.

JAMA takes and posts response letters directly on the journal web site, with the articles. The quality is admittedly so-so, but at least the comments are findable. If a replication is published in Journal of Applied Econometrics, nobody will find it — Google Scholar sorts by cites, replications don't get cited. Take this datacolada post, by a referee of an AER paper: datacolada.org/91 It's a really useful discussion! It's pretty much completely unfindable, if you were reading the original paper for a class, for a project, for any reason, there is no way you would find this. If it was published in J Applied Econometrics, it also would be unfindable. I agree adjudication is a challenge, I think there are good solutions here, but maybe not at the margin. e.g., Glen Weyl had the idea that authors could "sign off" on papers they've read closely and think are well done. I guess you are concerned that if we opened the doors, every journal article would get dozens of low quality comments. A simple voting system? I think you can tell pretty quickly whether a replication is above the bar of "worth at least thinking about."

Lots of good advice on venues that accept replications in this thread. But we're still being weird about this. The comments should live on the same web page as the original journal article. When I visit an article, there should be links to all the independent replications.

@paulnovosad I would consider them for the JDE. We do not publish comments per se or condition on whether the paper overturns the previous result. We do ask that the paper make it clear why the question is worth reexamining in a different frame or setting, given current knowledge.

Is there a venue for "I replicated this AEJ paper using a different dataset, it holds up" Or "it kinda holds up about as much as you would expect." Or "Data extended 20 yrs, does/doesn't hold." Will they take it as a comment, or does a comment have to claim to overturn?

@chidiya_ @Brown_Economics @alterelim Check out Aarushi's amazing work on toxic social media at aarushirita.github.io/papers/

@Brown_Economics Thank you so much! One with the true big fam @afosterri @alterelim

Econ Grad Student, Saxena Affiliate and PSTC Trainee Aarushi Kalra ( @chidiya_ ) wins best paper award for NEUDC 2024 at Northeastern University. Her paper is entitled " Hate in the Time of Algorithms: Evidence from a Large-Scale Experiment on Online Behavior". Congrats!

@leightjessica I have been looking into this issue at the JDE. I think we can improve instructions and do a better job of explaining how linked and non-linked tables and figures are treated. I'd like to see it modified to accept subdirectories. Other suggestions from the hive?

@JustinSandefur @marcfbellemare @eeshani_kandpal @thom_katherina Interesting paper but if I am reading it right the .03 is a transfer elasticity in Table 3 not an income elasticity. Divide by the transfer share of income (.1 to .2) to compare it with previous estimates--so .15 to .3.

Re-analysis of 4 RCTs: Cash transfers don't affect poor people's food consumption in quite the way we thought. cgdev.org/publication/in… from @marcfbellemare @eeshani_kandpal @thom_katherina

@dmckenzie001 @JakeVigdor @AEAjournals @ecmaEditors @ASR_Journal @apsrjournal JDE holding steady at 1849 for 2023. Seems odd to me that the variability over the last four years has been so low.

@JakeVigdor @AEAjournals @ecmaEditors @ASR_Journal @apsrjournal Interesting, here are some stats from different development journals taken from my annual blog post (blogs.worldbank.org/impactevaluati…) - I think there was a big 2020/21 bump in submissions at some journals, in part driven by lots of quick Covid papers.

The other day I noted a decline in @AEAjournals submissions since 2020 and there was some discussion over whether this is just an idiosyncratic thing. So I'm going to collect here a 🧵 of data points from other publications, in econ and beyond. Editors welcome to add more!

I am not sure what possessed me to get an extended warranty on my refrigerator purchased in 2019 but my experience with @New_Leaf_SC after my KitchenAid was "condemned" last month has been incredibly frustrating.

@jmb112485 @BarbaraBiasi No eduroam at sonesta. But yes the brown works at almost every academic institution.

@BarbaraBiasi Do they have eduroam at the Sonesta? I feel like that is a reasonable expectation given the volume of economics that happens there. If so I need to see this tomorrow. I have never seen it connect successfully.

Is there anyone that has ever successfully gotten eduroam to work?

Brown grad student Henrique Pita Barros tests a low-cost intervention to connect displaced persons and members of their host community in Mozambique blogs.worldbank.org/impactevaluati…

3/3 Urban, housing size, deregulation, India geetika-nagpal.com Bargaining, income hiding, work incentives, Cambodia alex-zhou.com

2/3 Childcare, health, environment, Brazil jmbvgarcia.github.io Religion, media, women, Brazil sites.google.com/site/marcelame…

1/3 Do take a look at Brown's fascinating and talented crop of development students on the market (A-Z). Some keywords to get you started-- Conflict, refugees, community meetings, Mozambique henriquepitabarros.com Water, women, religion, Middle East giuliabuccione.com