EVIDENCE-BASED RESEARCH ON CHARITABLE GIVING

In 2005, Richard Martin and John Randal conducted a field experiment in an art gallery manipulating the social information available to art gallery visitors by varying the perceived currency contents of a transparent donation box placed in a prominent area of the gallery’s lobby.

There were three treatments and one control:

• The control: empty box
• Treatment 1 ($50 Treatment): a few large denomination bills
• Treatment 2 ($5 Treatment): several small denomination bills
• Treatment 3 (50 Cents Treatment): many coins

Placing many coins in the box (50 cents treatment) led to the largest increase in proportion of visitors who gave. However, placing small or large denomination bills in the box did not significantly increase giving. Although placing many coins in the box increased the propensity to give, the average amount that donors gave actually decreased compared to the case when donors only saw an empty box.

Martin and Randal concluded that the 50 cents treatment provided social information that 1) individuals give frequently, and 2) individuals make small donations.

Richard Martin and John Randal, 2008. How is Donation Behavior Affected by the Donations of Others?

In 2003, Rachel Croson and Jen Shang partnered with a local public radio station during its annual mailing campaign and on-air fund drive campaign. For the on-air campaign, the researchers had experimenters work as phone attendants taking donation calls. The experimenters would determine whether the callers were new members (new donors) or renewing their membership (previous donors). If the callers were renewing their membership, right before being asked the amount they would like to contribute, they would be randomly assigned one of three treatments:

• Treatment 1: Upward social information treatment:
        Inform member that a real donor made a higher contribution than the amount given by the member in the previous year
• Treatment 2: Downward social information treatment:
        Inform member that a real donor made a lower contribution than the amount given by the member in the previous year
• Treatment 3: Neither upward or downward social information:
        Inform the member a real donor made a contribution equal to the amount given by the member in the previous year

The researchers would then record the difference between the members' recent donations and their previous year's donations. For the station's mailing campaign, the station would mail renewal letters to members. Each participant member would be randomly assigned a letter with one of three treatments as noted in the on-air fund drive.

Upward and downward social information pulled average overall donations up and down, respectively. The upward social information treatment increased average donation sizes by $12, while the downward effect reduced average donation sizes by $24.

Rachel Croson, and Jen Shang, 2008. The Impact of Downward Social Information on Contribution Decisions

Rachel Croson and Jen Shang came back and again partnered with a public radio station during its annual fundraising campaign to conduct 2 more experiments.

The first experiment randomly assigned each caller/donor (either new donor, obtaining new membership; or previous donor, renewing membership) to one of three treatments to be administered before the caller made his/her contribution decision. The donor received the treatment information that another donor made a contribution equal to the 50th ($75), 85th ($180), or 95th ($300) percentile of the previous year’s fund drive contribution distribution. Callers randomized into the control group received no donation information at all.

The second experiment looked to find the optimal level of social information to maximize contribution levels. The researchers strictly focused on previous donors (renewing members) and their behavior in response to receiving social information between the 95th-99th ($600) or at the 99th ($1000) percentile of the previous year’s contribution distribution of only renewing members.

The 95th percentile treatment led to a 12% increase in contribution amount relative to the control condition, and it led to the highest overall contribution level at $119.72 (the control receiving $106.72).

(Incidentally, the researchers found that donors renewing their membership had different distributions of giving than their new membership peers. Renewing members gave more on average compared to their new membership donors.)

The study also found that increased contributions as a result of social information led to significantly higher membership/renewal rates. This implies that 1) there was no crowd out effect of higher contributions on participation rates and 2) that contributions may increase in subsequent years due to higher membership/renewal rates:
  23-32% = renewal membership rate interval for treatment groups
  12% = renewal membership rate for control group

Jen Shang and Rachel Croson, 2009. Field Experiments in Charitable Contribution: The Impact of Social Influence on the Voluntary Provision of Public Goods