Editor’s note: This article was initially published in The Daily Gazette, Swarthmore’s online, daily newspaper founded in Fall 1996. As of Fall 2018, the DG has merged with The Phoenix. See the about page to read more about the DG.
Unlike the psychology experiments that Swat students are regularly invited to participate in, the rewards earned from participating in economics experiments actually depend on the decisions that you make in them. Make certain choices, and you could earn rewards ranging beyond the scope of Swat’s pay scale.
The researcher behind these experiments is a recent addition to Swarthmore’s Economics department. Assistant Professor of Economics David Huffman, whose background includes a Ph.D. in Economics from UC Berkeley and work at private research institute IZA, uses lab experiments to conduct economic research. Although this methodology is familiar to researchers in the natural sciences, it is relatively new for the field of economics and groundbreaking for Swarthmore.
In the past, economic research has most often been conducted empirically, where numeric data is collected and analyzed to explain how the observations are correlated to each other. “With empirical research, you get correlations and don’t know why they happened—but you can try to make a good case for them,” explained Jeff Weaver ’10, Huffman’s research assistant on this project. The correlations themselves may provide incomplete information, because “too often, the correlation it makes could reflect the influence of some unobserved variables,” argued Associate Professor of Economics Tom Dee, who has also conducted a variety of social experiments where research is conducted in the field rather than in a lab setting.
Many of economics’ foundations are based on mathematical proofs that make assumptions about human behavior. The experimental paradigm allows economists to test those assumptions.
To conduct experiments, Huffman recreates a situation similar to a market place where buyers and sellers exist. Students sit in front of computers in Trotter 301, dividers set between them to ensure that they are unaware of other players. Each player is randomly paired up with another player in the room and makes decisions in the constructed market situation.
One example offered by Weaver is a game of splitting $100 between two people. Economic theory, which assumes actors will behave rationally, predicts that the first player will not split any amount of the money to maximize his payoff of $100. Yet, the observed behavior yields a more equitable outcome with the first player offering half the amount to the second player. With this game performed in a natural setting, it is difficult to explain precisely the outcome. Does it imply that people instinctively value fairness? Or did the first player simply choose a fair outcome to project a reputation?
“A well constructed experiment can precisely point to motives,” Huffman said. “A lab setting takes away some factors that are unobservable in empirical research, often categorized into social pressure.” Since this lab structure does not allow students to interact with each other personally, variables associated with social pressure, like concern for reputation, are controlled.
The sample used for most of Huffman’s experiments consists predominantly of students, although anyone in the Swarthmore community is invited to participate. Huffman explained that “students are convenient, because they are right there and have lower incomes, so they don’t need to be paid as much … Students are smart, so it is easy to explain [the rules] to them.”
Yet, both Huffman and Dee noted that use of students may also be a drawback for the validity of extrapolations of the outcomes of these experiments to other contexts. “The kind of subjects in lab experiments may not be representative in real world … it’s odd to infer something about strategic behavior of multinational firms from 20-year-old students in a lab,” Dee said. Huffman could not reveal the nature of the experiments, but he reassured that “for these types of experiments, it’s okay that we look at students.”
Huffman will be running experiments throughout the year and will choose a sample from a database of participants.. Anyone in the Swarthmore community is welcome to participate and can register at