Abhijeet Singh

We can't say definitively (no center-based data).

Likely: low time use on cognition. Pub preschools are under-resourced, a single AWC worker expected to work miracles!

This *can* be fixed (as we show for COVID-19 remediation in the same state elsewhere)

jhr.uwpress.org/content/earl...

Our new paper in EJ shows facts I've been curious about for ages.

I'm biased but, IMO, these results are important.

Most of all, though, this paper was fun to write --- coauthors who worked well together, steady progress, and referees who were demanding but made the paper SO much better!

Not on human rights or democracy, for sure. But, for skilled workers, moving to Dubai or Singapore is often more attractive than Europe. It’s easier for companies to expand, salaries are higher, taxes/COL lower etc.

EU has many natural advantages for attracting talent but little joined up strategy

That, sadly, is only one of the things that are wild about 2025…

I remember meeting Karthik & @singhabhi.bsky.social w/ the Mindspark/EI team when I was working at Global Innovation Fund back in 2017—and being blown away by how they were designing for impact @ scale from day 1.

A stellar example of academics + do-ers partnering for scale 🚀📈👇

Really cool result here -- edtech intervention using personalized adaptive learning in RCT in India reduces mismatch between students and the (overly ambitious) curriculum

No randomly assigned diff in dosage. Some externally induced through implementation changes which we evaluate with observational methods in Year 3 of the program.

The dose response survives school FE, grade FE, attendance, lagged achievement etc. pretty well

In first ppr, we also validated w/ IV

@woessmann.bsky.social @deonfilmer.bsky.social @matthewakraft.com @andydebarros.bsky.social

Big picture: `science of scaling’ matters enormously for welfare.

Unlike medicine, social programs need iterative adaptation & testing beyond finding "what works" in small pilots.

Without this, as
@johnlist.bsky.social
notes, “we are performing efficacy tests on steroids”. 16/16"

These results also matter for rich countries.

High-dosage tutoring has been found to be very effective but has been hard to scale since good tutors are scarce, and costly.

PAL could make personalized tutoring feasible at scale. 15/16

These results matter beyond India.

Most LMICs face a “learning crisis”, and have large numbers of students below grade-level standards.

Middle-school years are especially challenging with few evidence-backed scalable ideas.

Our results offer a promising way forward. 14/16

Implementation protocols developed/tested in this study have *already* informed policy at scale (+ other EdTech initiatives).

Mindspark is now operational in over 2200 public schools in 13 Indian states covering >260k students.

Further scale-ups currently in the works. 13/16

Why is this a big deal?

Governments spend billions on hardware, but evidence shows hardware alone ≠ learning.

What matters is thoughtful integration of PAL software into classrooms.

Our study shows how to make EdTech actually work in public schools. 12/16

A key result is that the "dose-response" relationship between time spent on the platform and learning gains is linear and constant over time.

Thus, monitoring usage is a low-cost way to track and improve implementation quality at scale 11/16

One disappointment: we find no gains on school exams.

Students were so far behind that even large gains didn’t show up on grade-level tests.

Highlights trade-off between teaching “at the right level” vs “at the curricular level,” and the value of early remediation. 10/16

This model was 2x as cost-effective as the original study because: (a) hardware was used more intensely, and (b) many inputs were already present in schools

Program was 1.5 to 4x more cost-effective than default spending patterns.

Costs should reduce further over time. 9/16

A key new result, enabled by multi-year RCT:

PAL significantly reduced mismatch between students and curriculum over time as seen in the pivot over time in the Figure below: 8/16

All students gained comparably, consistent with personalization

BUT, since weaker students learn less in control, the same *absolute* effect is a much larger *relative* improvement over business-as-usual gains for weaker students. 7/16

Our main results:

After 18 months, treated students scored 0.2-0.22SD higher in both Math/Hindi.

That’s 50–66% more learning per year than control schools.

Effect size is in ~90th percentile of effects from large education RCTs (N>5000).
@daveevansphd.bsky.social
6/16

Our study design:

RCT with 80 public schools (40 treatment and 40 control) across 4 districts in Rajasthan.

All students in Grades 1-8 in treated schools were included (~6500/year).

We focus on independent measures of student learning after 2-years of treatment 5/16

Why might PAL work here?

Because the learning crisis is severe:

(i) In grade 8, average math skill is at Grade 4 level
(ii) within class variation in learning spans many grade levels.

Even the best teachers struggle to handle this problem, which PAL can address 4/16

This study:

Working with
Educational Initiatives,
we developed an in-school model where students spent 1 period/day in a Mindspark lab for Math/Hindi.

This displaced ~25–50% of regular class time

Impacts not obvious since tech-induced classroom disruptions can often REDUCE learning. 3/16

n our 2019 AER paper, we found that PAL (Mindspark) had huge positive effects in Delhi.

But that RCT treated ~300 students in an after-school setting with extra instructional time on PAL.

The big question: Can it work in public schools, at scale, within the school day? 2/16

Many interventions “work” in small trials but fail at scale

Also, EdTech often promises much, but delivers little

In a new paper (bit.ly/3JKLgVn)
@karthik-econ.bsky.social
& I show how personalized adaptive learning (PAL) software can sharply improve learning outcomes at scale🧵1/16

📢Our CESifo Conference on Economics of Education is about to start!🤩

This will be so exciting!!!🥳

🔗https://www.ifo.de/en/cesifo/event/2025-09-05/cesifo-area-conference-economics-education-2025

@raffasadun.bsky.social @sandramcnally.bsky.social @christinafelfe.bsky.social @singhabhi.bsky.social …

Results so stunningly clear they inspired this classic xkcd (xkcd.com/2400/):

This would also be devastating for current US Econ PhD programs. Nobody finishes in 4 years, even 5 is rare!

It would be strange if US unis were forced to move to the old UK/France model (separate Masters + 3y PhD) now when finally the top end of European programs finally converged to 5y programs…

I thought Chris Walters’ recent Handbook chapter on Empirical Bayes in Labor Economics was very useful.

@marome1.bsky.social and I leaned on it a lot to deal with some tricky EB shrinkage issues in a recent paper (with my PhD student, Petter Berg, who did most of the work figuring out the EB lit)

In my papers, I think of that part as:

“If I write another paper on this, or had a do-over given what I *now* know, what would I really want to know?”

There’s prob a 50% chance I’ll return to it within 3-5 years with a new project. Not to fluff up future research but bcos I’d really like to know!

He’s definitely helped European academics feel happier about their jobs 😁

It does annoy me that Mincer returns to education in Sweden are only 1/3-1/2 of the US, and academics have similar tenure bars but half the pay.

But at least we don’t (yet) need to worry about ICE!