This video is my entry for Project for Awesome, an annual charity event on Youtube where we
raise money and vote for our favourite nonprofits to give that money to.
I’ll reveal which organisation I’m supporting in a few minutes.
First though, let me tell you a story about how powerful the scientific method can be.
In the late 90s, the Harvard professor Michael Kremer, was interested in improving education
Instead of starting an education nonprofit, like you might expect, he decided to do a
series of Randomised Control Trials.
This is what they entailed: he randomly split up the schools he was studying in Kenya into
One group would continue with their schooling as usual, and the other group would receive
the new interventions like more textbooks and more teachers.
Finally he’d compare the outcomes of the students, and, provided the sample was big
enough, could conclude that any differences between the groups were due to the intervention.
As you probably know, this sort of Randomised control trial is exactly how we test if a
medicine is effective, but up until this point they were unheard of in the not for profit
For obvious reasons, of course this intervention is going to be effective- why should someone
waste precious resources testing that?
Well Kremer’s idea was not to test all the different interventions at once, but to test
them all in separate trials, to figure which was the most effective at improving the outcomes
for the students.
To his surprise, one of these factors worked ridiculously well.
First they tested giving out more textbooks, then providing flipcharts for the teachers
use, and then improving the teacher to student ratio.
Before I tell you the results, I want you to take 10 seconds to guess which of these
you think was most effective.
Which one would you have funded, if you had to choose?
Ok, let’s go through them.
Surprisingly, the extra textbooks seemed to have no real effect.
A possible explanation is that these textbooks were too high level for the student anyway,
especially since they were written in English, the student’s third language.
So then they tried the flipcharts, reasoning that the teachers could tailor the lesson
to the students ability this way.
Again though, for whatever reason, this had no discernible effect in the sample they trialled.
Then they tried simply having more teachers, but again, there was no effect.
Frustrated that none of these had worked, Kremer decided to try a crazy suggestion he’d
His team gave the children medicine in school that prevented them from getting stomach worms.
Stomach worms are something many of us aren’t familiar with, because they’re fairly rare
in developed countries.
But they’re devastating to a person’s health, and far too common in some developing
It turned out, many children in Kenya at the time were missing school because they were
sick with worms.
What they found in the trial was that students who’d got the treatment missed school far
less, and as a result did far better.
In fact, when researchers followed up with the students a whole 10 years later, those
who’d been treated were earning 20% more than those who hadn’t been.
For just 50c per a child per a year, this intervention had had a massive impact on their
The scientific method tells us that we should make hypothesis about the world but then we
need to test them.
If they don’t hold up, no matter how much it seems like it should be true, we need to
abandon the theory and start the process again.
And in this case you can see how powerful this process can be.
Some nonprofit interventions really are a lot more effective than others, so it’s
worth doing this sort of rigorous analysis to find them.
To put this into perspective, here’s some really mind bending data.
This study compared 108 health interventions in developing countries.
To measure how successful a health intervention is, you want to look at how many extra healthy
years of life it gives someone.
1 extra year of health is what we’ll call a DALY.
This study looked at how many DALYs $1000 spent on an intervention could buy.
What they found was, half the 108 interventions saved people 5 years or less of life for that
In other words, 5 years was the median value in the study.
So you might expect the best interventions save around 10-15 years.
That’s what I would have guessed.
But actually, a whole bunch of the remaining half fall between 5 years all the way up to
100, which is great.
But amazingly, there are 5 interventions out of the 108 that do even better.
The best one saved 300 years for every $1000 spent, A massive bargain, compared to the
median of just 5 years.
The majority of these intervention work just ok, some a few of these interventions work
So doesn’t it make sense to find these and fund them?
That’s where GiveWell, the not for profit I’m supporting, comes in.
They do extensive research on these sorts of interventions, and if you donate to them,
they give that money to their top rated initiatives.
At the moment that includes several organisations focussed on deworming treatments.
But GiveWell apply the scientific mindset that we should always be updating our beliefs
based on the best new evidence.
That’s why they reassess their choices for top non-profits every year, and are very transparent
about how they come to their conclusions.
If you want to support this sort of evidence based giving, there’s two simple things
you can do: one, vote for them on the project for awesome website.
The link is in the description.
Share this video, because a lot of people have the idea that charity is useless and
I hope this video convinced you that it certainly doesn’t have to be this way.
Charity can be an effective way to help those most in need.