At one house this Halloween, trick-or-treaters were asked to give candy rather than simply receive it.

From 5 to 8 p.m. Monday night, volunteers from the Yale chapter of Students for Proven Impact and staff members from an affiliate organization, Innovations for Poverty Action, donned costumes and sat on the porch of Yale economics professor and IPA founder Dean Karlan’s East Rock home. They took advantage of a flood of trick-or-treaters to conduct an experiment testing what factors affect children’s willingness to donate candy to charity.

The purpose of the experiment, in which over 300 trick-or-treaters participated, was to test the strength of the “follow the leader” effect by determining whether people were more likely to donate to wealthy charities or poorer ones, SPI member Adèle Rossouw ’13 said. While some may see wealthy charities as more efficient, she said, others may be more sympathetic towards poorer ones.

In the experiment, trick-or-treaters formed a line in front of Karlan’s steps and were asked by volunteers whether they wanted to “play a game.” At the front of the line they received a slip of paper randomly assigning them to either one of two tables or telling them to choose between the two tables. Volunteers then gave each child five Tootsie Rolls and led them to their table. Both tables had two orange pumpkin buckets, with the one representing the wealthy charity already containing 360 Tootsie Rolls and the “poor charity” bucket containing just 15 Tootsie Rolls. Children were randomly assigned to either the full bucket or the mostly empty one, and each trick-or-treater was given the choice of how many Tootsie Rolls, if any, to donate to kids who did not have any.

A cardboard cutout of President Barack Obama was featured at one table. The role of the Obama cutout, Rossouw said, was to see if his presence inspired children to donate more candy.

Although Karlan was traveling in Uganda and not present for the experiment, Rossouw said he suggested many aspects of the experiment’s design, such as the Obama cutout.

Trick-or-treater Khadija Hussain, 12, who wore a “crazy cat lady” costume, said she found the experiment “interesting” despite having participated in it last year. She donated all five Tootsie Rolls and three additional candies she had collected during the night.

Researchers recorded each child’s age and whether their costume was homemade, and then judged their costumes on originality. Ted Barnett, a Project Associate at IPA, who was in charge of scoring costumes, said he had no set rubric for judging the costumes, but felt a bias towards homemade costumes.

Some costumes included a lemon tree, a goth bumblebee, a decapitated body and many young Spidermen.

The purpose of rating the originality of costumes was to assess whether originality correlates with a willingness to donate, SPI member Zachary Groff ’13 said. He hypothesized that kids with common costumes would be more likely to “follow the leader”, or put candy in the fuller bowl.

These “Halloween experiments” began in 2007, when the experiment focused on the correlation between common costumes and aversion to uncertainty. In 2008, in honor of the presidential elections, the porch was adorned with McCain decorations on one side and Obama decorations on the other. That experiment measured trick-or-treaters’ preferences between the candidates against their willingness to move away from the side of their preferred candidate if offered more candy on the opposite side.

Last year, experimenters explored children’s preferences between domestic and international charities. Groff estimated that 400 trick-or-treaters participated last year.

Yale Law School professor Claire Priest ’94 LAW ’01 GRD ’03 said she has taken her two sons to the Karlan house for the past three Halloweens.

“I think it’s great,” she said. “It’s a wonderful way to collect data.”

The data will be analyzed by IPA and eventually published.

Students for Proven Impact, previously known as the Proven Impact Club, was founded at Yale in 2010 and promotes development work through scientific assessments and quantitative research.