Colin Aitken

GPT5 having some trouble with the gold standard "can you draw me a spooky picture" benchmark

One thing I admired Americans for was they provided efficient and effective aid for people around the world.

I thought this made America great.

The PEPFAR program saved plausibly about 25 million lives and prevented at least 5.5 million babies from being born with HIV.

See: pepfarreport.org

PEPFAR Report The State Department, which administers PEPFAR, says the program has saved 25 million lives. We are an independent team of journalists, academics, and technologists who decided to investigate whether ...

Most of the lives PEPFAR saved, like most HIV patients in Africa, were likely women and children.

The long answer on what exactly we learned and how we learned it can be found here:
pepfarreport.org

For context: 7.5 million is roughly the population of Arizona. 30 million is roughly the population of Texas. These are *enormous* numbers of people who are alive who otherwise wouldn't be.

It did so extremely cost-effectively: costing between $1,500 and $10,000 per life saved. (That's about three *thousand* times cheaper than the price the US government is willing to pay to save American lives.)

PEPFAR Report The State Department, which administers PEPFAR, says the program has saved 25 million lives. We are an independent team of journalists, academics, and technologists who decided to investigate whether ...

Has the President’s Emergency Plan For AIDS Relief (PEPFAR) actually saved millions of lives?

We weren't sure, so we spent the last two weekends looking deeply into multiple lines of evidence. The short answer: PEPFAR saved between 7.5 and 30 million lives in the first 15 years of its existence.

The discussion ending at this tweet is very smart on both sides ( @lugaricano.bsky.social and @shengwuli.bsky.social ), reasonably spicy, and fun to read.

Hello #econsky. I am building a grad-level course for next semester on Macro Development with a focus on agriculture and the environment.

Please share and shamelessly (self-)promote any JMPs, WPs, and underrated bangers that you think would be worth including in the syllabus.

Here are some I have used from econ for teaching purposes aside from LaLonde: Fraker and Maynard (1987, JHR); Heckman and Smith (1999, JEP); Cook et al. (2008, JPAM). In other fields: Heineken and Shadish (1996, Psych Methods); Glazerman et al. (2003, Annals); Shadish et al. (2008, JASA).

LURKING INFERENTIAL MONSTERS? QUANTIFYING SELECTION BIAS IN EVALUATIONS OF SCHOOL PROGRAMS This study examines whether unobserved factors substantially bias education evaluations that rely on the Conditional Independence Assumption. We add 14 new within-study comparisons to the literature,...

Here’s a nice one:

onlinelibrary.wiley.com/doi/abs/10.1...

Also, this person has cataloged a bunch of them:

gwern.net/correlation

Because I've seen the law of iterated expectations, Jensen's inequality, and the central limit theorem mentioned in the past few days, I'll migrate one of my early Twitter posts about the tools necessary to master econometrics -- which includes each of those. Here it is.

I literally showed that child mortality video this in class today. Everyone seems only vaguely impressed until I describe how my first born almost died shortly after birth (www.ryancbriggs.net/blog/2015/10...). It’s wild to me how much we need stories to even start to grasp something numerical.

best films about economics? i have lagaan, wonka (2023), seven samurai, dune, killers of the flower moon and it's a wonderful life.

I often say there's lots of low-hanging fruit where research could answer important qs with well-designed studies, but they don't happen bc of poor incentives, neglected areas, etc.

@ruben.the100.ci suggested writing a thread of 'studies I would fund in a heartbeat'. So here is an ongoing thread.