GWWC Reporting Attrition Visualization

April 19th, 2023
ea, gwwc
A few weeks ago Giving What We Can published their impact evaluation. As an organization that supports people in making long term altruistic commitments, the question of how much people stick to those commitments is pretty important. We don't know what people are actually doing, but whether they're reporting donations to GWWC is a decent proxy, and the evaluation includes a table of how many members were still reporting donations:

This is interesting data, but I found it a bit hard to think about. I tried a few ways of visualizing it, and made a pair of charts:

code

I colored each line with an intensity proportional to the number of people in the cohort, since that represents how much information it represents. The weighted average line fades out over time as it represents progressively fewer members.

Some thoughts, looking at the charts:

  • Later cohorts, starting around 2015, fall off faster than earlier cohorts.

  • Reporting attrition is quite sharp at first, and then slows down, but it doesn't go to zero.

  • After the initial attrition it seems to end up around 2-6% annually, though this depends a lot on where you see "initial attrition" ending.

  • There's probably some sort of effect of the calendar year ("how hard did GWWC push people to report donations") and I haven't looked at this.

Disclosure: my wife used to be President of GWWC. I haven't run this post by her and I don't know her views here. I shared a draft of with GWWC before posting.

Comment via: facebook, lesswrong, the EA Forum, mastodon, substack

Recent posts on blogs I like:

I did a meditation and it sort of changed my personality?!

a trip report

via benkuhn.net November 8, 2026

The bad part about rape is all the rape

I have recently seen a lot of people casually refer to consensual non-consent orgies as “rape orgies”.

via Thing of Things November 6, 2026

Test/verification technique effectiveness with Astra

When we looked at how well agents used test/verification techniques in August1, we saw that GPT-5.6 Sol generally did what you might call a caricature of various techniques when asked to use a technique or library such as TDD, QuickCheck, Lean 4, etc.; no…

via Posts on November 2, 2026

more     (via openring)