From Knowledge @ Wharton

Suppose you work at an organization -- like a public radio station or library -- that offers a free service of value to society, such as quality radio programming or the ability to borrow books at no cost. And suppose you decide to launch a fund-raising drive to help cover your organization's operating expenses. How would you structure that drive? Are there steps you could take -- besides offering gifts or other incentives -- to increase the amount of money that individual donors commit?

Wharton operations and information management professor Rachel Croson and Jen Shang, a PhD student formerly at the University of Pennsylvania and now at Indiana University, set out to answer this question by examining the influence of social information on contribution behavior. Their specific goal was to find out whether individual donors to a public radio station will give more money if they are told the amount of another donor's contribution. Their paper is entitled, "Field Experiments in Charitable Contributions: The Impact of Social Influence on the Voluntary Provision of Public Goods."

The researchers chose public radio as a representative example of an institution where individuals benefit from a service but are not required to contribute to it; at the same time, the community is better off as a whole when the service is funded. Other examples of "positive externalities" include environmental groups, which enable others to enjoy the benefit of cleaner air without themselves making any donation; or town watch committees, where the whole community benefits from reduced crime, including those who don't contribute time or money to the town watch effort.

Croson notes that social information affects behavior in a variety of economic situations, such as labor markets (when you are trying to decide what a reasonable wage is, you look at what other people earn) and investing (when making decisions about how to allocate your portfolio, you look at what other people with similar assets have done). Her research adds to this area, she says, by "providing evidence that social influence [in this case, data about what another person has donated to the radio station] has an impact on charitable contributions as well."

The 'Right' Contribution

Croson and Shang worked with a public radio station that has three on-air fund drives per year during which DJs and volunteers ask for donations and propose particular contribution levels in amounts ranging from $50 for basic membership on up to $2,500. According to Croson, research suggests that individuals are more likely to be influenced by social information when the situation is seen as "ambiguous" -- i.e., there is no obvious or correct answer -- and when the information received is seen as "relevant or appropriate." The public radio station experiment meets these criteria. Â

First, "the range of recommended contribution levels means that callers have relatively little idea of what the 'right' contribution might be," says Croson. "It's not entirely clear how much they should donate. Pure self-interest says zero. But people don't want to do that. They want to do the right thing, whatever that is."

Second, the station "could tell you how much it thinks you should give. But that's probably not appropriate. Or a volunteer could say, 'Bill Gates gave us $10 million.' That's not relevant. So in order for this information to have an impact on my behavior it should be relevant to me, and it should be appropriate," Croson says.

In their experiment, Croson and Shang analyzed past contribution data from the station and focused on three levels -- donations of $75 (the 50^th percentile, meaning half the donations were above that and half below), $180 (the 85^th percentile, meaning 15% of the donations were above that amount and 85% were below) and $300 (the 90^th percentile; 10% of the donations were above that amount and 90% were below).

Once a person called into the station and indicated he or she wanted to make a donation, the volunteers manning the phones were told to say: "Are you a new member or a renewing member?" Once the donor answered, some donors were then told by the volunteer that: "We had another member" who contributed $75 (or $180 or $300; these amounts were randomly assigned) -- or the donors were told nothing (thereby constituting the control group.) In both cases, the volunteer then said: "How much would you like to pledge today?" No deception was used in this experiment; there had indeed been members who had contributed these amounts on the first day of the fund drive.

The question the researchers wanted to answer was whether those donors who were told how much another donor had given were more likely to pledge a higher amount than donors who were not given this information. During the length of the experiment, 538 donors called the station.

"Our primary result is that social information can influence contributions," the researchers write. Those members who were told that another donor had contributed $300 gave an average contribution of $119.70 while those in the same group who were not told about the other donor's contribution (the control group) gave $106.72. This $13 difference "would translate into a 12% increase in revenue for the station had all callers been offered the $300 social information," the researchers note.

In discussing the long-term impact of this strategy, one concern is whether the increased contributions come at a cost. "Do higher contributions this year crowd out future contributions from these donors?" the researchers ask. The answer is no. They examined the renewal rate and the amount the donors contributed in the subsequent year, and found that the renewal rate is higher when donors are given the social information than when the donors are not given this information (the increase was 12%). "We can conclude that providing social information significantly increases the renewal rate," the researchers write. In addition, the contribution amount one year later is higher in the social information conditions (ranging from $93.97 to $121.13) than in the control condition ($86.11)

The $300 figure given out to donors was especially influential, the researchers found, leading to a higher probability of contributing one year in the future (31.9% vs. 12.3% in the control group) and a higher amount contributed ($93.97 vs. $86.11).  Multiplying these together yields higher expected contributions ($29.95 in the $300 condition vs. $10.62 in the control condition).

The researchers conclude, based on this experiment, that social information has the potential to "influence real-world decisions in the voluntary contribution to public goods. Providing social information significantly increased contributions. The size and significance of these effects varied, with the most effective social information level representing the 90^th percentile of the distribution of contributions. In addition, the increase in contributions due to social influence does not crowd out future contributions. In fact, it generates higher expected revenue than the control condition."

Same Sex Donors

One natural question arises: If the $300 social information increases contributions, wouldn't higher amounts work better? Croson and Shang investigated this "boundary condition" for the influence of social information. "The social influence research suggests that individuals conform only to relevant norms. A contribution that is too high, such as Gates' $10 million, might easily be seen as irrelevant or inappropriate." The researchers used only renewing donors calling into the station and compared social information of $600 (which had previously been demonstrated as effective) with social information of $1,000, the 99^th percentile of contributions.

The experiment was conducted in the same public radio station and involved 280 renewing donors. The results: The $600 condition produced higher contributions than the $1,000 condition ($172 in the $600 condition and $140 in the $1,000 condition). These findings "support the social conformity explanation for the influence of social information rather than an alternative cognitive reference-point explanation," the researchers say. In other words, the boundary effect suggests that using the $1,000 "anchor" did not increase donations.  Instead, contributions increased because of the appropriateness of the number offered.

Aware that their results could be affected by the way the experiments were set up -- using the telephone, for example -- the two researchers participated in a renewal campaign by mail, and found that donors are influenced by social information presented in that setting as well.

Croson and Shang cite other research this year showing that callers who were told that another donor was the same sex (as they) gave significantly more than callers who were told that another donor was the opposite sex -- i.e., women give more if they know other women are giving more. "This result is again consistent with the social conformity story. If another person like me gave a particular amount, then it is more likely to be relevant or appropriate than if another person unlike me gave that amount," the researchers write.

The authors, in their two field experiments, demonstrate that social information influences contributions, with information drawn from the 90^th to 95^th percentile of contributions being the most effective. Lower social information has little or no influence, while higher social information (the 99^th percentile) actually decreases the level of contribution. "Our research provides a deeper understanding of what motivates individuals to contribute to the funding of public goods and other charitable organizations," says Croson. It is also "a first step in understanding where social influence is likely to be an important factor to consider in our attempts to improve predictions and explanations of economic behavior."