Literature Review: Why Do People Give Money To Charity?

Link post

Notes:

  • 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:

Char­i­ta­ble Fundrais­ing and Smart Giv­ing: How can char­i­ties use be­hav­ioral sci­ence to drive dona­tions?

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:

The Part of the Th­e­sis You Should Ac­tu­ally Read

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.

Reg­u­lar Summary

I be­gan to study the non­profit sec­tor be­cause I’m con­vinced that giv­ing money to the right causes is one of the best ways for an av­er­age per­son to im­prove the world.

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).

The Result

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.

Limitations

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.

Introduction

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.)

Method

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 13 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.

Conclusion

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.

Here’s why:

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.

Last Words

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.

As I dis­cov­ered while writ­ing this the­sis, the sci­ence of fundrais­ing isn’t very rigor­ous yet. The team at Char­ity Science ex­plains why.

My fa­vorite liter­a­ture re­views on char­i­ta­ble giv­ing, be­sides mine: Zagefka & James (2015) and Bekkers & Wiep­king (2011)

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.