Cross-posting from my blog without much refinement. If you spot any non-typo errors, I will upvote you and correct the post on my website.
If you aren’t sure whether to read it, I’ll try to tip you over the edge by mentioning that someone at Charity Science did, and decided to add it to their website.
If you might be writing a thesis at some point in the future, consider picking a topic that could be helpful to the EA community! And if you wrote an EA-ish thesis recently, consider writing a summary for the Forum! I’m really glad I wrote this summary; it helped ~300 hours of work not go to waste.
In 2015, I wrote a senior thesis:
It’s a very long paper, and you probably shouldn’t read the whole thing. I conducted my final round of editing over the course of 38 hours, during which I did not sleep. It’s kind of a slog.
Here’s a PDF of the five pages where I summarize everything I learned and make recommendations to charities:
In the rest of this post, I’ve explained my motivation for actually writing this thing, and squeezed my key findings into a pair of summaries: One that’s a hundred words long, one that’s quite a bit longer.
Super Short Summary
Americans only give about 2% of their income to charity, and most of that money goes to charities that don’t do an especially good job of helping people. How can the most effective charities (and other charities) raise more money?
There are many different techniques that have been shown to work well in multiple studies, but evidence on most techniques is still very mixed, and some popular techniques in the real world have no experimental evidence behind them. Charities really ought to run more experiments to figure out which techniques will work for them.
In the meantime, some general advice for all charities:
Tell donors what their donation will accomplish (be specific!).
Tell stories about individual people you’ve helped.
Make donors feel like they’re on a winning team with lots of other cool donors, making real progress on an important problem.
Also, run experiments. I can’t emphasize that enough.
I’d seen a lot of studies on fundraising techniques, and on techniques for persuading people in general, but it wasn’t easy to find a lot of studies in one place, and it was especially tough to figure out whether any techniques at all had super-strong evidence behind them. It seemed like some were overvalued thanks to the results of a single study that wouldn’t necessarily generalize to most nonprofits.
So I did something foolish. I decided that my senior thesis would attempt to review every experimental study ever conducted on a charitable fundraising technique.
To ensure that I was saying something original, I added a special section on techniques that would apply especially to “effective” charities: Those which could present strong evidence to donors that they were actually making the world a better place (and doing so more efficiently than most other charities).
This isn’t the best-written literature review of fundraising techniques, nor the most comprehensive. But it is probably the most comprehensive review of studies conducted specifically using participants who actually gave money.
This is actually a major problem in the fundraising literature: About half the studies I found didn’t measure the impact of a technique on real donations. Instead, researchers measured how much money the participants claimed they would give if someone asked them, or whether they gave tokens to people playing an “economic game” with them, or whether they helped a research assistant clean up spilled coffee.
(To make an uncharitable comparison, it’s as though Stanley Milgram had conducted his famous obedience experiment by asking participants whether they would be willing to shock the person on the other side of the curtain if he asked nicely.)
I excluded any study that didn’t measure real monetary donations, unless it dealt in some way with evidence-based giving — very little has been written in that domain, so I had to be a bit less selective.
Take everything I say with a grain of salt: I was an undergraduate when I wrote this, and I probably missed important points in some of these papers.
Almost every study here involves a single request for money, even though donor retention is more important for most charities than getting new donors who only give once. Including donor retention would have made this thesis almost impossible to write, but it’s still an important topic. (Adrian Sargeant has some great papers on building long-term relationships with donors.)
There’s not a lot of research on most of the techniques I covered, considering how popular they are. I found about five studies per technique, and many of those were methodologically flawed. Sample sizes and effect sizes varied drastically, and the sheer number of techniques meant that a meta-analysis wouldn’t have made sense.
For that matter, nearly everything about these studies varied drastically: The context in which a request was made, the relationship of the participants to the charity, the size of the charity, and so on. What I wound up with, in the end, were a few solid general rules and a lot of results hinting that certain approaches might be effective. Still, it’s better to have hints than to have nothing.
The Actual Literature Review!
Reminder: This is a very abridged summary of the paper. Citations available from the actual paper.
Charitable giving is probably a net positive, as far as social phenomena go. And even if it isn’t, the most efficient, data-driven forms of giving are certainly good. (This is my first thesis, so I’m defending even the most basic assertions.)
The latter form of giving, or “effective altruism”, clearly helps the recipients of donations, but it’s not entirely clear whether giving actually makes people happier. There’s a good chance that happy people give more, or that people claim to be happy after giving so experimenters will like them. But it’s also quite possible that giving money makes us happier than spending it, especially once we’ve spent a certain amount on ourselves.
But even though charitable giving is a very good thing, we don’t give very much, and the rate at which we give hasn’t really changed since 1970. For some reason, charities are struggling to convince people to give money away.
But science can help! This literature review aims to summarize research on the efficacy of various fundraising techniques — particularly those which could be useful to the most effective charities.
By the way, some charities are more effective than others! (I’ll skip this bit for the EA Forum post, you’ve all heard it before.)
I read hundreds of pages of Google Scholar even more pages in a few specialized databases and lots of books and the reference sections of some truly epic literature reviews (which are linked at the end of this post).
Some of the techniques I reviewed could be used by just about any charity. Others should be especially useful for charities that have something in common with the most effective charities — that is, they help people in other countries, help lots of people, measure their results, etc.
With a few exceptions, I only reviewed studies where participants actually gave real money to charity, because most other ways of predicting giving behavior in the real world don’t seem very effective, and we really want to predict giving behavior! Prediction is the name of the game.
I’m also not measuring religious gifts or gifts to colleges, because “giving back” to an institution that helped you isn’t quite the same as the giving I’d like to measure.
Who Gives? Why?
What motivates people to give money away? I’m not going to say “System One” and “System Two”, because that’s cliche, so instead I’ll say “warm giving” and “cool giving” to reflect the fact that giving is driven by a mix of “cool” motivations (an abstract desire to do good, careful calculation of your impact, strategic giving that will make people like you) and “warm” motivations (empathy toward the recipient, personal connections to the charity, a habit of giving a dollar to anyone who asks).
Yes, this is really just System One and System Two. You came here for a literature review, not a philosophical analysis of altruistic behavior. This section is lazy.
Anyway, who gives the most money away? Sometimes men give more than women, and sometimes the reverse it true. Older people give more until around the time they retire. Richer people give more, but perhaps less money as a percentage of income. Religious people might give more, but it’s really difficult to tell because 1⁄3 of all U.S. donations goes toward churches, which only spend a fraction of their income on traditional “charitable” activities.
Fundraising Techniques that Probably Help
“Legitimizing paltry contributions”: Tell donors that “even a penny will help” (or something like that), and they’ll usually be more likely to give without giving much less. I have a bunch of theories about why this happens, but we have many more techniques to cover, so let’s move on!
Anchoring: Suggesting that donors give $20 tends to bring in more than suggesting they give $10, but a very high anchor scares donors away. Use experiments to figure out the optimal suggestion!
Dialogue: Ask someone how they’re doing, wait for them to answer, and then ask for money. This is a much better idea than asking right away, but so far we’ve only seen it work in person, not over the phone. Bonus points if you mention having something in common with the donor!
(In my favorite “similarity” study, the experimenter lied about having the same birthday as the participant. I can’t believe an IRB let them get away with that.)
Publicity: When someone’s donations will be made public (or even seen by just a single experimenter), they tend to give more. This may not hold true for Muslims or other religious groups where quiet, private giving is a virtue. But that’s a minor exception hypothesized from a single study: Mostly, publicity is a good strategy in the context of these experiments.
Photographs: Adding pictures to donation materials tends to make them more effective, though it’s unclear whether sad children are better than happy children. Especially sad or upsetting photos could backfire.
Individuals: We really like helping individuals, possibly because it’s easier to empathize with one person than with a whole group of people. Rather than talking about the sheer scope of the problem your charity deals with, it’s generally best to talk about how a donation has helped, or could help, a single sympathetic person.
In fact, people will literally give more money to save the life of one child than to save the lives of eight, even when eight lives can be saved for the price of one!
This is a troubling result, but one team of researchers may have discovered how to reverse it with something called “the unit asking effect”. (That paper might be my favorite in the entire thesis — check it out if you can.)
Follow the Leader: Potential donors give more after they learn about the gifts of past donors, especially those who were very generous or who resembled the potential donor in some way. This also works if the potential donor sees another donation happen, or is told that the amount they donate will be known by another person (so that they have the chance to become “leaders” themselves).
Matching donations: Ben Kuhn is better at statistics than I am, and his summary of the literature on matching is very rigorous. If you really care about donation-matching, you should read it.
My shorter summary: If you have some money lying around, you might be able to use it to increase donations by “matching” the gifts of future donors, so that people feel like they can do more good for the same “price” (as though your nonprofit were having a buy-one-get-one-free sale). Ben Kuhn points out that most of the research on matching is sketchy, but it’s no sketchier than the rest of the research on fundraising. Also, matching is “free”, since your charity gets the matching dollars either way, so you might as well experiment.
Seed donations: Announce that you’d like to raise a set amount, then “seed” part of that amount so that “success” seems more likely. Donors like giving to specific campaigns that seem like they will meet their goals, and seed donations work about as well as matching in head-to-head experiments. On the other hand, if you have money you could use to seed a campaign or match donations, you could also try…
Overhead coverage: When a charity announces that donors’ gifts will only cover “programs” (like giving mosquito nets to families) rather than “overhead” (like paying the salaries of professional fundraisers), donors give quite a bit more. This phenomenon can be hacked if a charity uses leftover funds to “cover” its own overhead, or convinces one particular donor to cover all of the overhead so that most donors never have to think about it.
Donors seem to prefer charities with lower overhead even when the overhead is “covered”, but it’s unclear whether that’s true independent of donors’ fear that their own money will pay for overhead rather than programs.
Many nonprofits claim that “overhead doesn’t matter”, because forcing charities not to spend on overhead keeps them from growing or innovating. This is partly true, though especially high overhead can be a warning sign that something weird is going on. Anyway, what really matters is how much good each dollar does, however the charity spends it. (Still, donors speak the language of overhead, so charities may have to do the same.)
Other Fundraising Techniques
This summary is long enough already, so I’ll skip talking about techniques that only work sporadically, or don’t seem to work at all.
With one exception: Offering gifts or prizes in exchange for donations works very badly in every study that tries to do it. This may not be the case for gifts “related” to the nonprofit (like a PBS tote bag), but telling people you’ll give them random chocolate if they donate is a terrible idea.
On the other hand, telling donors they’ll feel great after they give works pretty well, despite playing on the same selfish motivation. And giving people gifts before you ask them to give leads to amazing results (at least in the fourth study mentioned on page 20 of this paper).
Really Obvious Helpful Techniques
Simple, evidence-based things that all nonprofits should probably be doing:
Talk about your beneficiaries a lot. Make them sound like nice, hardworking people who have a lot in common with the donor.
Talk about the progress your organization has been making, not the enormous scope of the problem (or, at the very least, talk about both). People want to be on the winning team.
Look good. Dress nicely. Donors can be shallow! It might even help to play catchy music and smell good, though that study has yet to be funded.
If someone signs up for your mailing list, send them an email right away, and ask them to donate soon after. As of 2014, some of the largest charities in the U.S. didn’t send a single email to new subscribers within 30 days — enough time for a potential donor to completely forget about them.
Use simple, visual language. One clever study took issue with the fact that newspapers tend to use the word “affected” to describe the people who survive natural disasters. Referring to these people as “homeless” (which is what “affected” really means in this context) substantially increases the amount donors are willing to give to them.
This isn’t surprising: I don’t know what an “affected” person looks like, but I can picture a “homeless” person without difficulty, and being able to imagine someone is an important step toward caring about them. Visual language is important.
When we consider the size and sociological importance of the nonprofit sector, it becomes clear that we need more research on fundraising techniques!
Yes, like any person who researches a topic in great depth, I conclude that more research is needed. On the other hand, I’m not going to grad school, so I’m not biased by the need to churn out more papers on things I already know about. You can trust me on this one.
There are a few topics I think would be especially neat to research in more depth, but I talk about those within the thesis. For the rest of this summary, all I’d like to say is that charities should be running more experiments, and publicizing their results.
One cool thing about the nonprofit sector is that it isn’t a zero-sum game. It’s true that charity money is limited. But if we somehow raise charitable spending from 2% to 3% of the U.S. GDP, the gains from that will dwarf the pain of charitable competition. And one of the ways charities can raise the national giving rate is to work together to figure out better fundraising techniques.
What would happen if charities with excellent websites — like Kiva or Acumen or charity: water — shared the results of their A/B testing with the rest of the nonprofit world?
What if the five largest charities in America pooled funds to hire a couple of full-time researchers who could run a dozen experimental replications of important studies over the next year, and begin to figure out which techniques consistently had large effects on charitable giving?
What if the hundred largest charities in America hired a couple of extra lobbyists to push for a U.S. version of Gift Aid, which could push the giving rate from 2% to 2.5% within a couple of years?
I don’t know if any of this would help, but it seems like it would be worthwhile to try. Fundraising experiments are easy to run, and can even be profitable. Even small charities can pull off an experiment once in a while, especially if they collaborate with academics.
Many of the studies I examined found that some techniques can boost donations by 50% or more. Either these results don’t carry over to the real world, or charities can profit enormously from experimentation; I’d really like to know which one is true.
It may be that no technique or set of techniques, however clever, is going to push the 2% giving rate to 3% or higher. If so, we’ll need to figure out other ways to do more good with our giving.
This is why I’m so excited about effective altruism: And… skipping this, since y’all on the Forum have your own reasons to be excited. In retrospect, though, it’s funny that I went from writing this paper and speculating that I’d work for an EA charity someday to not even looking at open EA jobs, GiveWell excepted, for almost three years.
Interesting Papers and Other Links
As I discovered while writing this thesis, the science of fundraising isn’t very rigorous yet. The team at Charity Science explains why.
The Behavioural Insights team designs very cool experiments, many of which use subconscious “nudges” to boost charitable giving.
One of the deepest threats to effective giving is “psychic numbing”: The more suffering we know about, the less likely we are to take an objective approach to dealing with it. When one person is in danger, we’ll make an enormous effort to save them; when hundreds of thousands are in danger, we often fall into despair and stop trying to help. Paul Slovic explains.
In 2010, a group of nonprofit foundations published the Money for Good study, which surveyed thousands of high-income donors in an attempt to figure out how people might be convinced to give more money, and give more to the “highest-performing nonprofits”. The results are fascinating, and a little sad: Only 35% of participants ever did research on their favorite charities, and only 10% of those people used neutral sources rather than the charities’ websites.
* * * * *
Also: I’d never have finished this thesis without the help of my wonderful advisor, Hedy Kober. I’d also like to thank Dean Karlan, my second reader, whose work helped inspire me to pursue this topic in the first place.