@learncollider

Building AI that Expands Human Opportunity Housing, Education, Workforce Nonprofit research lab founded by @peterbergman_ (@UTAustinEcon, @JPAL, @NBERpubs)

Joined April 2020
Learning Collider retweeted
A practical guide to recent econometric advances with shift-share instruments, from @borusyak, @instrumenthull, and @Xjaravel nber.org/papers/w33236
27
163
17,022
Learning Collider retweeted
How can we improve access to quality education for all? Affordable Housing, @GreatSchools & J-PAL affiliate @PeterBergman_ highlight the impact of evidence on housing choices. Watch the full story🔽
1
3
944
Learning Collider retweeted
📣A/B testing, common in tech, has massive potential to maximize impact in the social sector. Check out our new Insight Note @WWHGlobalEd @youthimpactorg: ➡️"A/B testing in education: rapid experimentation to optimise programme cost-effectiveness" wwhge.org/resources/a-b-test…
11
31
2,535
Learning Collider retweeted
#EconTwitter I'm very excited to present our new working paper (nber.org/papers/w33050) at @BrownUniversity tomorrow! Thanks for hosting me! A mid-sized 🧵 TL;DR: "Racial differences in income unable to explain a meaningful share of gaps in ambient air pollution exposure."
2
28
1
96
29,622
A common assumption is that families with Section 8 vouchers prefer to live in low-opportunity neighborhoods. What if that's not the case? Turns out, barriers to higher-opportunity neighborhoods play a major role in where low-income Americans end up living.
1
61
Creating Moves to Opportunity: Experimental Evidence on Barriers to Neighborhood Choice @peterbergman_, Raj Chetty, Stefanie Deluca, @nhendren82, @lkatz42, Christopher Palmer Quotes: opportunityinsights.org/wp-c… More: learningcollider.org/researc…
46
Learning Collider retweeted
Very excited that "Building Non-Discriminatory Algorithms in Selected Data" (with David Arnold & Will Dobbie) is now forthcoming at AER: Insights! This is the third (& final?) chapter of a series of AEA papers using IV methods to measure and reduce discrimination in high-stakes decisions. The table of contents goes something like: 1. The 2022 AER, on measuring discrimination in human decision-making (shorturl.at/2hLuQ) 2. The 2021 AEA P&P, on measuring discrimination in algorithmic decisions (shorturl.at/SWBKa) 3. The forthcoming AER:I, on reducing discrimination in algorithmics (shorturl.at/h6lDB) I learned a ton working with David and Will on these papers; I hope you find them interesting too!
32
149
31,299
🎧#Listen to the latest podcast episode from @vox_dev featuring @WWHGlobalEd Academic Director @angrist_noam where he discusses the gap between policy and practice in education📚 based on an assessment of 50 countries during COVID-19. #VoxDevTalks voxdev.org/topic/education/g…
6
8
1,231
Learning Collider retweeted
Alis paper develops new tools to measure discrimination in eviction filings, and uncovers a lot of it. I learned a ton in talking to her about this important work. Very well deserved! – at Providence, RI
Congratulations to Alison Lodermeier (PhD '24) for winning the 2024 APPAM PhD Dissertation Award for her dissertation "Essays on the Economics of Eviction." Amazing work Alison!
2
8
173
26,638
Today at @TheLBJSchool -> @UTAustinEcon & Learning Collider's very own @peterbergman_ speaking on AI in Teaching & Learning. Go, Peter!
3
82
Learning Collider retweeted
At least one in six workers in management, business, and computer-related occupations is using generative AI at work every day. Also, 22% of blue collar workers use AI. From our new paper, "The Rapid Adoption of Generative AI": static1.squarespace.com/stat…
2
20
3
89
10,699
Looking forward to virtually tuning in. Fantastic line up, @WWHGlobalEd!
Excited to present our A/B testing work at the @WWHGlobalEd conference this week! Sharing insights from 10 rapid A/B tests on scaling and cost-effectiveness-boosting strategies in Ed. @youthimpactorg @angrist_noam @janica_magat
1
104
Learning Collider retweeted
Had a brilliant time at #AFE in London. Inspiring talks by @Econ_4_Everyone @Susan_Athey @orianabandiera @singhabhi @RDMetcalfe +more. A treat to join a panel moderated by @Econ_4_Everyone on the "science of using science" 👇a slide from my talk & link: aeaweb.org/articles?id=10.12…
This quoted post is unavailable.
6
29
7,615
Learning Collider retweeted
Please RT/spread the word about this online causal inference course! Starting September 9th -- I'm very excited for it!
I had a great time teaching a minicourse on design-based econometrics in Bergen last week. See attached for a snapshot of what we talked about! For those outside of Norway, I'll be teaching a version of the course online in two weeks. Sign up here! mixtapesessions.io/session/d…
24
30
12,821
Learning Collider retweeted
Reasons I love this Yglesias-FAS collab: activates a nerdy readership towards policy action and entrepreneurship. More please!
This quoted post is unavailable.
1
2
11
1,473
We're excited to share that our Founder and Director, @peterbergman_, is a proposed speaker for #SXSWEDU on the panel "Seeding Innovation: The Power of Breakthrough Funding." Help make this session happen by voting and commenting here. ⤵️ panelpicker.sxsw.com/vote/15…
2
313
Learn how transformative funding is empowering trailblazers and propelling innovation during the "Seeding Innovation: The Power of Breakthrough Funding" session at #SXSW 2025! Community voting is open til Aug 18! @WaltonFamilyFdn @RenPhil21 @learncollider panelpicker.sxsw.com/vote/15…
2
3
209
Learning Collider retweeted
Excited to see @ALeighMP cite our RCT on the effectiveness of a phone-based math program in the Philippines. @youthimpactorg is proud to collaborate w/ @poverty_action @DepEd_PH @unicefphils on this cost-effective initiative, now scaling & undergoing govt A/B tests! 📈📚✨#RCTs
"Evidence-Based Development”: my speech on randomised trials in development, delivered at an Asian Development Bank conference in honour of Peter McCawley andrewleigh.com/evidence_bas… #auspol @adb_hq @AusAmbPH #randomistas
5
10
1,443
This was an extremely fun paper to be a part of! How can we evaluate whether a transformer has uncovered the right "world model"?
New paper: How can you tell if a transformer has the right world model? We trained a transformer to predict directions for NYC taxi rides. The model was good. It could find shortest paths between new points But had it built a map of NYC? We reconstructed its map and found this:
1
2
24
5,902