The Asad Islam — GDRI saga
Understanding what we know and need to know
[I will update this entry, as new information becomes available, and as my time allows. 1 March 2025, AO. Updates at the end of this blog entry.]
On February 25, at 12:55, the Institute4Replication (I4R) dropped, simultaneously on X and bluesky, a bomb of the special kind: links to the first two of 15 reproduction reports that several I4R related investigative teams of experienced replicators had undertaken. These reports (62 and 45 pages, respectively) can be found on OSF here (on a paper published in AEJ — Applied) and here (on a paper published in REStat). There you can also find statements of authors that have removed their names from these papers (as of now, Siddique, Vlassopoulos, Rahman).
Here is the I4R announcement
In a follow-up tweet the I4R laid out the impressive scope of their work.
Paul Novosad has a useful set of highlights on these two published reports.
Josh Merfeld has useful threads on the 1st report here and on the 2nd report here.
In yet more follow-up tweets the I4R wrote:
The replication effort has also identified multiple other potential issues in these papers which are detailed in the individual reports. Author responses are mixed, with some asking for withdrawing their name while others are working on responses.
We have two additional completed reports which have been shared with some of the same authors. We will make those publicly available shortly and keep you posted.
(1) “Parent-Teacher Meetings and Student Outcomes: Evidence from a Developing Country’’ by Islam (2019) European Economic Review.
2) “Partisan Effects of Information Campaigns in Competitive Authoritarian Elections: Evidence from Bangladesh” by Ahmed et al. (2024) Economic Journal.
The I4R folks also noted that they had several other reports in the works including reports on articles scheduled to be published in JPE — Micro and JEEA. Read the whole thread here.
Days b4 there were indications that some such bomb was about to drop.
First, a set of tweets by the Chair of I4R on 19 February indicating the massive investigation that I4R related investigative teams had done, pro bono, for 3 months. Subsequently, and even before (because the reports were shared with them weeks earlier), several authors tried not to be drawn into this bloodbath-in-the-making by trying to retract their authorship, or even retracting papers.
Second, one of the authors, Abu Siddique, posted on X several days earlier:
Third, the GDRI website disappeared from the web around the time of the Siddique tweets (when, somewhat ironically, Siddique was apparently at Monash, the home institution of Asad Islam although he was hosted by another co-author.)
Fortunately, the WayBackMachine is relentless. Among the concerns and questions raised on social media is that the GDRI, a not-for-profit, was not only (co-)founded by the central actor-author in this tangled web of problematic studies, Asad Islam, but that it was/is a corrupted nepotic family affair, with hundreds of thousands, and possibly millions, of dollars provided by prominent grant agencies having been channeled into it. Whether indeed this not-for-profit (which provided the data for all the papers that are now being investigated by I4R) collected the data is at this point another open question in need of an answer.
An insider calling her/himself A Concerned Researcher contacted me several months ago anonymously with related concerns. These concerns did not relate to the present batch of indicted papers but papers that were published earlier and involved, for example, sex workers in Bangladesh (see here and here). Among the concerns was that these papers did not have proper IRB approval and were shady in other respects (the Concerned Researcher detailed these concerns and provided ethics approvals which indeed look fishy and are hard to reconcile with the papers that allegedly build on them.) The Concerned Researcher reported that s/he had contacted Monash more than a year ago but that they seemed to stonewall her/him. If true, that’d be unacceptable.
Understanding what we know and need to know
First, it seems clear that Monash needs to look hard, and at arms-length (!), at the complete body of work that Asad Islam has produced (including papers, now with new title r&r at QJE, where he recently was dropped as co-author). That includes taking a hard look at the ethics approvals for all papers. The evidence out already, the numerous retractions of names on papers, and retractions of papers themselves, do(es) not negate the presumption of innocence per se but surely there is enough out there already to ask plenty of other hard questions. These questions need to be answered in public and transparently.
Second, it seems clear that Monash needs to look hard, and again at arms-length, at the exact nature of the relationship between Asad Islam and that not-for-profit (GDRI) and its operatives. Again, there are plenty of hard questions out already. These questions need to be answered in public and transparently.
Over to you, Monash. Do the right thing.
Third, many of the co-authors of Asad Islam (some rather prominent!) will have some serious explaining to do. Why, for example, did no one question the extra-ordinary effects sizes reported in some of the papers, or the data collection and generation process, and for that matter the nature of that curious entity in Bangladesh that produced the data.
Let’s make no mistake about it. What we know so far suggests the making of a scandal easily as big as the Ariely — Gino affair about which I have written here.
Update 3 March 2025:
I4R has an update on X on the number of papers that they investigate. Same update on bluesky.
They also write:
Here is a list of studies, with links to I4R reports, authors’ responses/statements, and emails sent to the editors and authors (with dates): https://osf.io/23x64/. This excel document is being updated every day.
Updates 4 March 2025:
Another I4R update out just now:
#GDRI_rep I4R sent a report to the original authors for the PLOS One article "“Food insecurity and mental health of women during COVID-19: Evidence from a developing country". All authors have now emailed PLOS One to withdraw their names. https://osf.io/7tzek/
And wouldn’t you know it, the WayBackMachine copy of the GDRI site has disappeared.
Update 8 March 2025:
Gelman on the (Ariely -) Gino saga and the academy’s hesitancy to clean up its act. Written while the Islam — GDRI saga is brewing and Monash seems to try to Duke it out.
“In summary:
Business academia needs to reckon with this inconvenient truth: Committing fraud is, right now, a viable career strategy that can propel you at the top of the academic world.”
Timely piece.
Update 9 March 2025:
Another one bites the dust. Latest I4R update in the Asad Islam — GDRI saga. Very satisfying.
Institute for Replication @i4replication.bsky.social
hashtag#GDRI_rep Update 3: The paper “Improving Hygiene and Sanitation through Parental Skill Training” by Asadul Islam, Umair Khalil and Tabassum Rahman was conditionally accepted at Economic Development and Cultural Change. We requested a replication package. It has now been withdrawn.
Editor in chief wrote to us that two authors (UK and TR) wanted to withdraw their names and one author (AI) has suggested that it doesn’t make sense to continue with a sole authorship and has requested that the paper be withdrawn. Journal is working with UChicago Press now.
We contacted the Journal of Health Economics to obtain a replication package for a R&R study (Unlocking Immunity: Strategies for Cost-Effective Promotion of COVID-19 Vaccination in Developing Countries) by A. Islam, S. Khetrapal, B. Kim, A. Sim, D. Pakrashi, M. Yadav…
Editor in chief wrote to us that the submission to JHE was withdrawn by the corresponding author, who cited an ongoing inquiry. [We are assuming that’s I4R inquiry.]
We contacted the editors at the Quarterly Journal of Economics about this study R&R “Edutainment, social network, and ethnic discrimination in rural Bangladesh”. We requested a replication package. Nathan Nunn (editor) responded that “[…] we feel the best strategy, since the…
paper is being revised (and for the near future will still be changing due to requests of the referees) is for us to review the files internally. If the paper ends up (after going through our own replication of their files and the referees’ scrutiny) being on a path to …
publication, then we will ask them to prepare a set of files that I4R can review. We would not publish the paper until there is a successful review from I4R.”
We agree and and are happy to review the paper if the paper is on path for a publication.
PLUS MUCH MORE:
See here for status on 30+ GDRI studies the I4R is investigating: osf.io/23x64/files.
Later in the day, … https://x.com/Josh_Merfeld/status/1898862670310453261
Update 10 March 2025:
An important question, waiting for an answer. A related question asked by many: Can Abu Siddique just withdraw his name from the papers that got him his position(s) and keep it (them), scot-free?
Update 11 March 2025:
Indeed. The saga started however at least a decade earlier …
Update(s) 12 March 2025:
Another day and sure enough another I4R report (41 pages) pertaining to the Islam — GDRI saga. Interestingly, “Report and author’s response here: https://osf.io/2axzm/files/osfstorage. We decided not to reply to the response. Our comment is under review at EER.”
No word from Monash yet.
Excellent summary of the report by this guy. Will save you having to slog through 41 pages of dense prose. Well done, Josh.
Last tweet in his thread: “TLDR: I don't believe a word in the original study.”
The Threadreader summary is here.
Update 14 March 2025:
I was curious about the referral sources … interesting to see how much more referrals come from twitter/X rather than bsky. Also, what is econjobrumors.com? Gotta check it out. Conversion ratio from look-ups to reading close to 80 %.
Update 15 March 2025:
Another day, another I4R update on the Islam — GDRI saga. And what a doozy it is. Not only a new report (and responses) but we learn that an R&R at JPubEcon by Islam et al was withdrawn and an R&R by Islam et al at JPE — Micro has now been rejected. No word from Monash yet.
Hidden deep in the thread is this deplorable detail:
Update 17 March 2025:
A couple of people have dug into the reports that I4R dropped Friday arvo about the AHI EJ paper and have read same; people elsewhere have suggested that the data in the AHI EJ paper are likely fake. There certainly are plenty of questions in need of an answer.
A useful summary it seems:
Abu Siddique Feb 18:
Recently, I took a major decision by withdrawing my name from a series of papers on covid that I coauthored. …
The following publications are ones for which I have formally requested the respective editorial offices to withdraw my name … :
Telementoring and Homeschooling During School Closures: A Randomized Experiment in Rural Bangladesh (with Hashibul Hassan, Asad Islam, & Liang Choon Wang). The Economic Journal, 134(662): 2418–2438. 2024.
Raising Health Awareness in Rural Communities: A Randomized Experiment in Bangladesh and India (with Tabassum Rahman, Debayan Pakrashi, Asad Islam & Firoz Ahmed). The Review of Economics and Statistics, 106 (3): 638–654. 2024. Online Appendix, Ungated version.
Improving Women’s Mental Health During a Pandemic (with Michael Vlassopoulos, Tabassum Rahman, Debayan Pakrashi, Asad Islam & Firoz Ahmed). American Economic Journal: Applied Economics, 16(2), 422–255. 2024. Online Appendix, Ungated version.
Emotional and Behavioral Impacts of Telementoring and Homeschooling Support on Children (with Hashibul Hassan, Asad Islam, & Liang Choon Wang).
American Economic Association: Papers & Proceedings, 2023. 113: 498–502. Ungated version.
All data for these papers supplied by GDRI.
Feb 24 — first two I4R reports drop on two of the four papers above (RESTat, AEJ- Applied). Additional authors remove their names from these papers. I4R indicates that it is investigating 15 papers in which Islam is involved and for which the data were provided by GDRI which he either founded or co-founded. (As of today, I4R has more than 30 papers under investigation. See https://osf.io/hjc9z )
Around the same time the GDRI website suddenly undergoes construction.
A few days later the cached copy on the WayBack Machine gets mysteriously disconnected.
March 3: I4R updates and reports that an author of a paper forthcoming at Economic Development and Cultural Change has requested to withdraw their name… EiC has contacted I4R and pauses publication.
March 4: I4R updates again; drops new report on PLOS One article whereupon all authors email PLOS One to withdraw their names. The orphaned paper not withdrawn yet.
March 10: I4R updates again; reports that EDCC paper has now formally been withdrawn by the sole remaining author (Islam). I4R also reports that a paper at JHR for which they asked for the replication packages was formally withdrawn by the corresponding author (Islam).
March 12: I4R updates again; drops a report (and author response) on solo-authored paper in EER (author: Islam). I4R: “We decided not to reply to the response. Our comment is under review at EER.” Mic drop.
March 15: I4R updates again; drops yet another report (and author responses) on paper entitled “Partisan Effects of Information Campaigns in Competitive Authoritarian Elections: Evidence from Bangladesh” by Ahmed, Hodler and Islam published at @EJ_RES. Data supplied by GDRI for a(n undisclosed) pittance.
I4R reports also:
We requested a package for this study: “Can public recognition backfire? Field experimental evidence on the retention and performance of volunteers with heterogeneous social-image concerns” by Islam, Malek, Tasneem and Wang, R&R at JPubEcon… . One of the Editors in chief wrote to us that the submission was withdrawn by the authors.
We requested a package for this study: “Determining the Extent of Taste-Based and Accurate Statistical Discrimination: Evidence from a field experiment in India” by Islam, Pakrashi, Wang and Zenou, R&R at JPE Micro … We have now learnt that it has been rejected. We do not know why, but this means that we have no clear path to accessing the data at present.
Still 17 March 2025:
Nicholas J. L. Brown identifies a number of copypasta in the EJ data here:
https://pubpeer.com/publications/FDD0FC8EA7815FA485F0CF0DAEFFC9
