The GiveWell Blog

Diving in

As quickly and efficiently as I possibly can, I’m now trying to read up on our five causes, and ask the question: “What is the evidence – independent of particular nonprofits – for and against different methods of helping people?” Given the enormity of the topics and how little time I have, this is difficult, to say the least. That’s why I’d like to keep you posted and get thoughts from anyone who’s interested.

Here’s what I’m thinking right now:

1. I’ve been surprised by how foreign the question above seems to foundations and grantmakers. I’ve had long conversations with several major foundations known for “results-oriented” grantmaking, where I’ve asked this question, and have gotten referred either nowhere or to very general sites where I can start digging in. I’ve gotten practically no concrete statements along the lines of “We believe Method X is more promising than Method Y” (I got exactly one, and it was couched in about 200 caveats about how it’s just a generalization, and I’m not allowed to disclose to anyone what the generalization was or who said it). I would expect a foundation that makes grants in education to have written up its own opinions on what works (preschool? Tutoring? Longer hours? Teacher training? Etc.), with footnotes to the research that makes it believe this. We certainly will.

2. But, there’s certainly plenty of research out there to create this kind of writeup. More than I can realistically cover thoroughly in 2007. I’m doing the best I can to focus on very current research, and methods that are common among charities.

3. Causes 2, 3, and 4 are all very amenable to this sort of research, because “How do we fight poverty in the developing world?” and “How do we improve education?” are both discrete, established topics that academics like to fight over. Cause 1 and 5 is much harder: people tend to ask not “How can we save lives in Africa?” but “How can we fight malaria?” or “How can we fight AIDS?” If I went through all of these papers, I could eventually build up a picture, but I don’t believe I can find literature reviews that directly compare all the different ways of saving lives (for the record, the way I would do this would be to rely heavily on the triage approach). Let me know if I’m wrong. Cause 5 is a very similar situation: people argue about how to help ex-convicts and how to help substance abusers, but directly comparing the two (again, I would do this using the triage approach) seems pretty rare.

So, I’m going to focus my research for now on Causes 2-4; I’ll attack Causes 1 and 5 later, when we know who our strongest applicants are (in terms of their ability to report and evaluate what they do, our main criteria for Round One), and focus on comparing the specific strategies that our strongest applicants engage in.


  • Gillian on July 20, 2007 at 12:20 am said:

    A couple of points –

    Could you say something about how you are focusing your questions? Your causes 1 and 2 are very broad and the responses would depend very much on the local conditions. Here’s a reference about the current education situation in Tanzanian. Their basic problem is to recruit and train tens of thousands of teachers in the next year or two.

    The second issue I’d like to raise is the difference between ‘the best’ and ‘good enough’. You can spend a lot of time deciding the last hair breadth of excellence to decide on the best – and support that one. Or you can set guidelines that establish performance standards and support anyone/someone who passes those guidelines. On this measure, all ten runners in the final of the 100 metre sprint would be worthy of support, not just the person who won by two hundredths of a second.

    Any well-run project that puts more trained teachers in front of classes in Tanzania is worth supporting. The need is recognised and the context is supportive (i.e. the govt is stable and working to improve education — militias aren’t roaming the country burning down schools and killing teachers).

    Preschool education might have value in the US, but in Tanzania there aren’t enough elementary schools/teachers (average class sizes are 60), and there are high schools for only 30% of kids.

    So, answers to the generic question – should it be elementary school or preschool? – will vary with the country.

    We live in complex systems of interacting elements, our questions need to reflect this. You seem very structured, I wonder how your approach accounts for this kind of complexity?

  • Holden on July 20, 2007 at 11:43 am said:

    I’m not sure what you mean by funding all the good charities. We have $25,000 per cause to give out; we can’t fund every good activity, and realisitically, we can’t even fund a single one to completion. That’s why we have to choose – and that puts us in the same position as every other individual donor, which is the idea.

    I predict that the difference between “best” and “good” will not at all be splitting hairs, based on the fact that I never see things play out that way in any other sphere where people are trying to do something difficult. (And I’m surprised you think it will be, given your passionate support for a charity that sounds like it has a pretty unique approach. You’ve chosen School of St. Jude to support, over all the other charities you’ve seen – was that choice a matter of splitting hairs?) But, I could be wrong – if within a particular cause, we do see that the best 5 are too close to call, we’ll make a quick decision and spend the time we save on other causes or more research.

    The important thing is to do as much as we can to assess effectiveness, rather than concluding that “they’re all good enough” and splitting up the money. I believe that demanding the best is the best way to push for constant improvement, as well as the best way to drive a thorough investigation.

    Now, regarding the breadth of our causes. They aren’t as broad as you seem to think – the exact causes are listed here, and they specify different geographical areas of focus (although we’re well aware that “Africa” is far from a unified region). But they are broad. My quick answer to your question is that we are looking for charities to make their own case that they have the best approach to each goal – that means understanding and including the external factors in the areas they work in. Two charities are not equal just because they both build water infrastruture, for example – if one works in an area that is more in need of this infrastructure, it is saving more lives, which would make it better qualified under Cause 1. Does that clarify anything?

  • Gillian on July 20, 2007 at 3:42 pm said:

    Yes, Holden, you’ve confirmed, as I suspected, that your approach is very sensible.

    For me, choosing the St Jude’s was a bit like falling in love. It looked very attractive on first encounter, and further exposure added only positive things!

    You’ll be interested in my post about ‘Good Aid :: Bad Aid’ as it references a study of aid in three African countries and comes up with some principles for good aid.

    In assessing aid projects in Africa, I think that one needs to go beyond what the organisation says about itself and consider possible systemic downsides. I give the example of clothing gifts that prevent local manufacturers from getting started. Or NGO aid that impedes local govt agencies from building capacity by drawing resources away from them (e.g. hiring the few trained people in town).

Comments are closed.