David Reinstein: Answering questions from journalists on charitable giving (updated 2015)

  1. What are the main factors that motivate people to donate money? Is it the desire to do good, the desire to appear as though you are doing good? Do most people have a personal stake, like a family member with a particular disease? Does the balance of what motivates people shift with something as public as the ice bucket challenge (i.e. does it become more about social bonding/perception than charity)?

It is very difficult to know ‘why’ people do good, and diverse motives seem very plausible (and there’s some evidence for such heterogeneity).

There is evidence for social motivation,[1] although many people do give in anonymous settings, and there is also evidence that some people prefer anonymity and give less under recognition[2]. 

The first prominent evidence for ‘reputation-seeking’ in the economics literature came from Harbaugh (1988), who found alumni giving clustered around the minimum thresholds for particular levels of recognition.

There is also evidence [Dellavigna et al] that people will pay costs to  avoid being asked and avoid social pressure, and for this same case if they are asked they then respond by giving.

Economic models also suggest that people may want to use giving as a ‘signal of their goodness to others’; but then advertising that you wan recognition may nullify this. Thus, people may seek an ‘involuntary’ or ‘normal’ way of donating publicly, that allows them recognition without giving off the impression that they are seeking it

There is strong evidence for conforming behavior and people following examples and reference points set by earlier givers.[3]

Peers as solicitors tend to be effective: Meer (2011), conducts analysis of alumni fundraising drives of an anonymous university. As part of these drives, participants are telephoned and asked to make a donation to the university. When participants are called by their former college roommate (to whom they had been initially randomly assigned), they are significantly more likely to make a donation than if they were asked by a stranger.

I could find no connection in the PSID data between a family member getting a disease and individual donations, but there may have been measurement issues. There is strong conventional wisdom (backed by evidence) that people respond more to causes they can relate to and imagine themselves or know someone in such a position.

The ‘ask’ is very important. A conventional wisdom, backed up by some evidence, is ‘people don’t give unless asked’.  This could be about shifting reference points, social pressure, or about the person having asked having themselves ‘paid a cost’ in terms of the discomfort of asking. (Like pouring ice on your head).

Evidence: Levis, 1990; Andreoni and Rao, 2011; Long, 1976

… on this line, there is evidence that ‘gift exchange’ can work. Falk, A. “Gift exchange in the field.” Econometrica 75 (2007): 1501–1511.        

There seems to also be too major classes of donors – those who give regularly and perhaps in a sophisticated way, and those who mainly give small amounts in response to requests and social pressure.

Fun things like the ‘ice bucket challenge’ or marathons are also often tied to fundraising. There are various theories for why this link exists, but it's a bit of a puzzle.

2. Does giving in one place diminish giving in another? If so (and it appears this is the case), to what extent? Might this be happening with the ice bucket challenge, or is the challenge potentially bringing in a crop of new donators?

Yes and yes, although it is very hard to get strong and general evidence on this. I have two papers on this, 1 using the PSID and another lab experiments; I have also done some work on ‘substitution’ responses to natural disasters. In the PSID data the evidence is not overwhelming, but there is a strong case to be made that at least for the larger ‘regular’ donors, when they increase their giving to one category of cause in a particular year they decrease their giving to other categories (at least for certain pairings).

My lab evidence shows stronger results (and I need more cooperation to run better lab and field experiments exploring this).

I am also looking further, into some data from the Netherlands. My preliminary evidence, requiring followup, suggests that those who gave more to natural disaster relief in 2005 gave less to ‘international causes other than natural disasters’ in the same year.

As I noted before, giving behavior seems to be heterogeneous and very sensitive to environment, thus difficult to study.

From my recent paper “Substitution Among Charitable Contributions in Responses to Changes in Prices, Choices, and Information: Experimental Evidence

Some recent work has begun to address issues closely related to the present paper. Van Diepen et al. (2009) offer some field experimental evidence on the crowding-out effects of direct mail solicitations [of later solicitations from the same charity.] …  (Andreoni et al., 1996) examines substitution between giving and volunteering. Cairns and Slonim (2011) find that a “second collection” at a Catholic mass lead to a 17.5% higher overall level of contributions, but a 4.3% decline in first collections.

The converse of this is that for those who don’t regularly donate there seems to be no such ‘substitution.’ The possibility that people ‘learn to donate’ is intriguing and goes with evidence from other domains that people follow ‘inertia behavior’ in, e.g., savings and investments. Also ‘learning how to donate’ might remove excuses one could use in future.[4] 

 I have yet to see clear evidence for ‘learning to donate’ though.

Note that Charities seem to believe and act as if they are competing for donors, so this could be seen as a sort of ‘revealed preference’ evidence by those with a stake in the outcomes.

3. Why are we so susceptible to the power of suggestion, despite information that could persuade us to make another, perhaps more evidence-based, decision? For instance, although ALS is a horrific disease, if one looks purely at the numbers, CDC data reveal that many other conditions far outstrip ALS in terms of their death toll (even accidental injuries, such as falls). 

As you know there is a well known effect that people become much more concerned about risks when they have recent experience with them or they are more visceral. Also, people seem to have a tough time dealing with differences in large number. E.g., it's a common finding in ‘contingent valuation studies’ that people do not distinguish much in their stated willingness to pay for ‘saving 100 dolphins’ versus ‘saving 10,000 dolphins’.

4. From the ALS Association: "The ALS association has received $88.5 million in donations compared to $2.6 million during the same time period last year (July 29 to August 26) These donations have come from existing donors and 1.9 million new donors to The Association, which is incredibly grateful for this tremendous outpouring of support." 

This is about 34 times what the donations were in the same period last year, which is quite a substantial increase, but given the number of people who have participated in the ice bucket challenge (or acknowledged it via social media), might one expect these donation numbers to be higher? 


I haven’t done the arithmetic but I suspect it to be so. $100 seems like a large amount for the young generation that this appeals to, and there’s evidence that too high a suggested donation deters ontribution. However, there are cases in the past where more people wanted to be seen as having done good then actually followed through on it. Remember ‘hands across America”?

Note: On the afternoon of Sunday, May 25, 1986, almost seven million people joined hands to form a line that stretched 4,152 miles (6,682 km) ? from New York City's Battery Park to the RMS Queen Mary pier in Long Beach, California. This nationwide event, called Hands Across America, was intended to raise money to fight hunger and homelessness." Participants were supposed to contribute $10. $20 million raised. 20/7 × 10; 0.29

5. What does it mean that Macmillan Cancer Support is trying to piggyback off the success of the ice bucket challenge? How might this challenge change the way charities try to appeal to potential donors?

Clearly other charities will try to latch on to this, but there’s a danger that there will be saturation and these online ‘flavours of the month’ tend to go stale and people on the web, I think, don’t seem to like ‘me too’ sites (consider the lack of success of the ‘donate for me to make a potato salad’ clones). I think that charities will continue to try to use peer pressure in a ‘fun’ way. It’s not socially acceptable to say “I donated $100 to charity X, will you?” but it is acceptable to do this if you make a joke about it.

If people can gain status from their contribution this is a form of direct benefit to the ‘donor as consumer’; it has sustained charitable behavior for millennia, but in the modern world it is a bit difficult to buy because it is not normalized and our society is disparate (but it can be bought to some extent in small church groups, which may partly explain the greater charitable giving --at least including religious giving – of church-goers).

I hope this encourages charitable organisations to think about trying out and testing new ideas for motivating giving.

[1] Evidence: Harbaugh, 1998; Soetevent, 2005; Alpizar et al, 2008; Reinstein and Riener, 2012

[2] Jones and Linardi, ‘Wallflowers doing Good”

[3]Greater previously observed (or noted) gifts lead to greater donations, at least in some environments and within a range.

  1. Information selection treatments
  1. Alpizar et al (2008): Costa Rican national park: “A reference level of $10 increases contributions by about 4% compared with not providing any reference contribution information. However, providing a lower reference level of $2 decreases contributions by 23% compared to the baseline of no reference level”
  2. Croson and Shang: Radio fund drive [check: compare to no report]
  3. Frey and Meir (2004) students were more likely to donate (a specified discrete amount) to at least one social fund supporting other students when they were told that 64% of their peers donated (in a recent semester) in comparison to when they were told that 46% of their peers donated (on average over the past ten years).
  4. Martin and Randal (2008): Seeded donation box in a museum. “The social information provided had a significant impact on donation composition, frequency, and value.”
  1. Reinstein and Reiner [lab, charitable donation].
  2. “Econometric” strategies with field data        
  1. Smith et al (2012) JustGiving data from a marathon campaign
  2. Sanders, VMG data
  1. Mixed evidence on whether anonymous “previous gifts” are better at influencing later donors then revealed ones.
  1. Sanders finds this, in Virgin Mobile Giving context
  2. Reinstein and Riener (in the lab) finds opposite: stronger influence when the identity of the previous giver is reported.

Women as leaders also seem to have more influence on others' donations. (They also seem to be more successful at asking people to donate). (Reinstein and Riener). See also Peacey and Sanders 2013; and Smith et al. (2013) from the London Marathon.

Evidence: Reinstein and Reiner, 2012; Sarah Smith, 2012; Carman, 2003.

[4] For instance, academics at the Harvard Business School found that experimental participants who were able to purchase information on the recipients of donations withheld money from those who were not seen as ‘deserving’, yet they did not give more when the information was positive. Consequently, they gave less overall.