A new strategy for broadening the appeal of effective giving (GivingMultiplier.org)

post by Lucius_Caviola · 2020-10-26T23:18:51.692Z · EA · GW · 12 comments

Contents

  Background
  The strategy
    Donation bundling
    Asymmetrical matching
    Matching as donor coordination
  GivingMultiplier: testing it in the field
  How you can help
    feedback
    to spread the website
  Future plans
  FAQs
None
12 comments

In this post, I introduce an ongoing research project with the aim of bringing effective giving to a wider range of altruists. The strategy combines 1) donation bundling (splitting between your favorite and an effective charity), 2) asymmetrical matching (offering higher matching rates for allocating more to an effective charity), 3) a form of donor coordination (to provide the matching). After conducting a series of experiments, we will test our strategy in the real-world using our new website GivingMultiplier.org. This project is a collaboration with Prof Joshua Greene and is supported by an EA Funds grant.

Background

It is difficult to motivate people to give more effectively. Presenting people with information about charity effectiveness can increase effective giving to some extent (Caviola, Schubert, et al., 2020a; 2020b). However, the effect is limited because most people prefer to give to their favorite charity even when they know that other charities are more effective (Berman et al., 2018). This is because people are motivated by ‘warm glow’ of giving (Andreoni, 1990), which isn’t a good proxy for effectiveness. Another issue is that most people aren’t motivated to proactively seek out information about the most effective charities. But making people care more about effectiveness is difficult. In multiple studies I have found that presenting people with moral arguments makes little to no difference. (Though moral arguments might work for some people and under the right circumstances, cf. Lindauer et al., 2020; Schwitzgebel et al., 2020.) Therefore, the approach we take here is to work with people’s preferences instead of trying to change them.

The strategy

Below is a short summary of the set of techniques our strategy relies on. In our experiments, 2,000 (Amazon MechanicalTurk) participants made probabilistically implemented decisions involving real money. If you are interested in more details about our studies and results, you can find an early working draft here.

1) Donation bundling

We found that donations to effective charities can be increased by up to 75% when people are offered the option to split their donation between their favorite and a highly effective charity (Study 1). We call this technique donation bundling. Most donors find such bundle options appealing because they enjoy nearly all the warm-glow of giving exclusively to their favorite charity, but also gain the satisfaction of giving more effectively and fairly (Study 2). Likewise, we find that third-parties perceive bundle donors as both highly warm and highly competent, as compared to donors who give exclusively to an emotionally appealing charity (warm, but less competent) or exclusively to a highly effective charity (competent, but less warm) (Study 3).

2) Asymmetrical matching

The bundling technique can be enhanced by offering matching funds in an asymmetrical way, i.e. the matching rate increases as more is allocated to the effective charity. In our studies, participants were offered higher matching rates, the more they would give to the effective charity as opposed to their favorite charity. For example, they might get a 10% matching rate for giving 50% to their favorite and 50% to the effective charity, but a 20% matching rate for giving 100% to the effective charity. We found that asymmetrical matching can increase donations to effective charities by an additional 55% (Study 4). A key advantage of offering donation matching is that it provides people with no prior interest in effective giving to visit the site and choose to support a highly effective charity. 

3) Matching as donor coordination

Where does the matching funding come from? We have some initial matching funding from our EA Funds grant and hope that more EA donors are willing to provide matching funding. However, it may also be possible that our donation system could become financially self-sustaining by coordinating the actions of donors with differing priorities.

Some donors want to give exclusively to a personally meaningful charity, even if other charities have higher cost-effectiveness. Some donors—effective altruists—simply want to maximize their impact and are willing to support whichever charities have the best chance of doing the most good. Still other donors (ca. half of participants in our studies) have mixed preferences and want to support both kinds of charities. Donors who are focused on effectiveness—whether completely or partially—can voluntarily supply matching funds. This multiplies their own impact by incentivizing other donors to make more effective donations. And other donors can benefit from these provided matching funds, by enjoying the benefits of bundled giving while having their altruistic impact amplified.

To test this, we offered participants in our studies the option to provide the part of their donation that they had allocated to the effective charity into the matching system instead. Participants were informed that this part of their donation would then not be matched but would instead be used to match other people’s donations, thereby leading to a greater indirect impact. A third of participants chose the option to support the matching system, which left us with a four-fold surplus of matching funds (Study 5). (Note that any matching surplus can eventually be donated to effective charities.) If these results hold up with donors in the real world, the matching system could be financially self-sustaining. But even if it can’t be sustained entirely by users who choose to become “micro-matchers”, this could greatly reduce the need for larger donations from EA philanthropists.

GivingMultiplier: testing it in the field

Our research tells us that many people—perhaps most—like the idea of supporting effective charities. They just don’t like the idea of giving up on the charities that they love. But there is no reason why most donors can’t also be effective donors. Even if just a small proportion of donors were to give a part of their donation to highly effective charities, this could have a big impact. In the US alone, more than $450 billion are donated to charity each year—most of it goes to charities that aren’t considered particularly effective according to experts. And our experiments suggest that non-EA participants who use our donation system give more than half of their donations to effective charities. 

We now want to test whether our strategy can work in the real world. For that purpose, we have created a GivingMultiplier.org — an online donation platform that applies our techniques. Everyone (including you) can make donations to any US charity and get their donations matched with a rate between (currently) 2-20%. Donors have to give at least 10% to one of our recommended effective charities. The matching rate becomes higher, the more you allocate to an effective charity. This should be an appealing deal for most donors. Everyone likes their donations to be matched and most people are open to the idea of giving at least some part of their donation to a highly effective charity. 

Our plan is to keep the website running for a few months and assess its impact. That is, we measure how much the website can counterfactually direct to highly effective charities. And we measure whether the matching system can be financially self-sustaining, i.e., whether the provided matching funds (provided by both non-EA and EA donors) can cover the required funds to match donations by others. 

How you can help

Provide feedback

Any sort of critical feedback about the techniques, the studies and the website would be much appreciated. You can comment here or send me an email (lcaviola@fas.harvard.edu).

Help to spread the website

We’d like to get as many donors as possible during the next few weeks and months, especially for the coming giving season. We therefore need to market the website as widely as possible. 

You can help us by telling your friends and family members about the website and asking them to make a donation. If you’d like, share it on social media and other channels. After all, this may be the first time you can tell people about EA without making them feel bad for giving to their favorite charity!

If you’d like, please consider making a donation yourself. We ask users a few follow-up questions after they’ve made their donation, including whether they are EAs or not. Therefore, you won’t mess up our results if you make a donation. In fact, it would be very helpful to get EAs to donate and support our matching system.

My plan is to reach out to organizations and well-known EAs to share it via their channels. I’m also looking for ways to get celebrities to endorse our website and newspapers to cover it. Please let me know if you know people who could help make the website go viral or if you have other marketing ideas. 

Future plans

This is primarily a research project. We will keep the website running for a few months and then publish an academic paper about the findings. We believe that this could trigger more research focused on developing interventions to increase effective giving.

If the strategy works well, I hope that our techniques could be integrated into existing donation platforms. Our website could then be continued or merged with platforms. 

FAQs

12 comments

Comments sorted by top scores.

comment by MichaelStJules · 2020-10-29T07:20:11.240Z · EA(p) · GW(p)

Are you considering registering as a charity in other countries, too?

Or, maybe you can partner with RC Forward for Canada, the EA Foundation for Switzerland and Germany, and CEA for the UK and Netherlands to extend tax credits/deductions to those countries.

comment by Lucius_Caviola · 2020-10-30T16:32:37.334Z · EA(p) · GW(p)

It would be great to do that at a later point. Note that you already now can donate from anywhere in the world, but donations are only tax-deductible in the US.

comment by CristinaSchmidtIbáñez · 2020-10-27T16:06:30.768Z · EA(p) · GW(p)

Really enjoyed how intuitive the website is! Particularly how the matching is super clear at the end. Very excited to see the results as well. Thank you for all your work.

comment by alexrjl · 2020-10-27T12:22:46.072Z · EA(p) · GW(p)

This is a great idea and I'm excited to see the results!

comment by shaybenmoshe · 2020-10-27T07:39:10.293Z · EA(p) · GW(p)

I just wanted to say that I really like your idea, and at least at the intuitive level it sounds like it could work. Looking forward to the assessment of real-world usage!

Also, the website itself looks great, and very easy to use.

comment by Peterslattery · 2020-10-31T01:29:06.854Z · EA(p) · GW(p)

Amazing work! Donation bundling is a really great application of temptation bundling and I can see how it could really accelerate the adoption of effective giving. I am also really excited to see you translating your research into practice. Some quick thoughts:

  • It would be interesting to have an post test survey to track who gives and why, for example to see the extent to which different demographic factors (e.g., age and gender) and psychographic factors (e.g., attitude toward effective giving/belief in existential risk/self rated knowledge of effectiveness) correlate with different donation choices. Also: How do people feel that exposure to this service has changed them/their views (e.g., do they think they are more likely to give effectively and recommend it?
  • It would be interested to track changes over time and compare if those exposed to donation bundling differ much from a cohort who haven't been in terms of their future attitudes and behaviour. Then we could see if engaging in EG via donation bundling leads to changes in attitudes and behaviour that sustain (potentially even when the EG donation bundling option is removed) 
  • It will be important to maximise your traffic so have you thought about doing some sort of series of media releases (ideally with partner organisation) to make sure people know about the service?
    • I would also think about partnerships options for co-promotion: Intermediary services like workplace giving services, large and well networked (maybe semi-EA aligned) charities, Effective charities and EA recommendation  services like TLYCS and Givewell. 
  • Have you considered running experiments on the website using simple A/B tests? This seems like a great opportunity if you are getting the traffic.
    • You could  test the effect of different numbers of options . I know, for example, that several EA organisation, like EAA, have wondered what the optimum number of options to show and there  choice overload is often an issue in retail settings.
    • Testing other types of appeals/strategies could be interesting. For instance, normative appeals (e,g, X% choose this option), defaults (changing the initial value), chained/escalating approaches, like seeking a small commitment post donation (e.g., to read something about charitable donation or reflect on their donation approach) then seeing if that commitment/action shifts their identify/values and leads to greater donations relative to those who didn't make the commitment).
    • Testing presentation difference - images v logos, stats v images v arguments, different types of narrative (e.g., opportunity v obligation arguments) 
    • You state that "in some preliminary studies I found no significant differences in amounts given to effective charities depending on whether they focused on global poverty, animal welfare or existential risk" - this could be a chance to test this by pooling different charity types and assessing differences. Personally,  I would be surprised if there aren't some significant differences between demographic or psychographic groups, even if there are no differences on aggregate.

(and then I wrote a short essay - sorry - no need to respond to any/all of that but I hope some of it is useful!) 

comment by Lucius_Caviola · 2020-11-08T14:03:25.772Z · EA(p) · GW(p)

Thanks for these really helpful suggestions, Peter!

We are planning to test some of things you suggest. We kept our post-donation survey short because we wanted to focus on our main research question and not try to do too many things at once. But if we end up having a lot of donors, we might send them a survey via email to find out more about their demographics, beliefs and preferences. We're not planning to do A/B testing at this point. But if we're starting to get lots of donors, we'd definitely consider doing A/B testing to optimize the user experience and get more people to donate.

At this point, our primary goal is to test if the technique works in the real world and if we can get enough donors.  Yes, we want to do media releases to get more traffic. And we are trying to partner with organizations and services to spread the word. I like your idea of reaching out to workplace giving services. If you have concrete ones in mind or have ideas how I could find these, please shoot me a DM!

comment by MichaelStJules · 2020-10-29T07:01:59.476Z · EA(p) · GW(p)

Are you considering cause-specific matching funders? Or allow matchers to choose among which charities they'll match (as long as it's at least 2, to ensure regular donors still have some counterfactual impact).

This could allow the individual CEA Funds, GiveWell, ACE,  etc. to put matching funds there, too.

I also worry that if you don't allow cause-specific matching funders, you'll get fewer of them. I'd personally prefer to earmark my donations to specific causes.

comment by Lucius_Caviola · 2020-10-30T16:42:35.437Z · EA(p) · GW(p)

This is a good point and we've considered it. I agree that there are advantages to allowing matchers to support only specific causes (or charities). 

But there are also downsides. In addition to the ones you list below, the matching system would be somewhat less honest. Since the matcher would per default have donated to that cause/charity anyway, you as a donor don't really influence where the matcher's funding goes to. With our current system, in contrast, you do influence to which specific charity/cause the matcher's funding goes to. But this comes at the costs of the matching funder, who has to be willing to support any of the nine effective charities we currently list.

I still think it's worth thinking more about allowing for cause-specific matchings. But we don't plan to implement it anytime soon.

comment by MichaelStJules · 2020-10-30T01:21:54.735Z · EA(p) · GW(p)

Some potential drawbacks are:

  1. Greater risk of running out of matching funds for specific EA charities.
  2. Meta/community EA charities are kind of a public good in EA, so may go underfunded. You could require all matchers to match such charities (or at least one of them), though, although it would be good to ensure there's consensus on what counts as a meta/community charity. E.g. 80,000 Hours is the closest to one of those listed now (and none of the others seem like meta/community EA charities), but they're also pretty explicitly a longtermist organization and have their own cause priorities.
comment by Jess Kinchen Smith · 2020-10-28T23:59:50.211Z · EA(p) · GW(p)

Thanks for making this!

 

A nitpick: the top-left Harvard logo should IMO be a few pixels toward center. 

comment by meerpirat · 2020-10-27T19:18:51.732Z · EA(p) · GW(p)

Really really cool idea, and great to see it executed already! :)

I’m a little bit unsure how I feel about the name. It’s concise and informative, but it sounds a bit odd and un-fuzzy to my ears... hard to put in words. You probably put a lot of thought into this, would be interested what you and others think.