University professors plying their trade have been known at times to lie to store managers, restaurant owners, and even the worldwide readership of Wikipedia.
A couple of them have now risked fibbing to a potentially far more problematic lot: thousands of their fellow professors.
The researchers, Katherine L. Milkman of the University of Pennsylvania and Modupe N. Akinola of Columbia University, wanted to find out if people are more likely to act admirably when given more time to do so. And so they sent fake e-mail messages to 6,300 professors nationwide, pretending to be a graduate student seeking a few minutes of the professors’ time.
Ms. Milkman and Ms. Akinola may now have their answer, though perhaps not in the way they intended.
The study “belongs in the trash heap of ill-advised research projects,” Andrew E. Gelman, a professor of statistics and political science at Columbia, fired back to Ms. Milkman and Ms. Akinola after they revealed how and why they had deceived him. He posted his response on his blog, Statistical Modeling, Causal Inference, and Social Science.
Philip H. Daileader, an associate professor of history at the College of William and Mary, wrote back to the two researchers: “Involving colleagues, or any human beings, in a study without their knowledge and their prior consent is unethical.”
Compensation and Apologies
The basic tactic employed by Ms. Milkman, an assistant professor of operations and information management, and Ms. Akinola, an assistant professor of management, is hardly without precedent.
Researchers routinely devise tests in which they or others adopt the guise of job applicants, home buyers, store customers, and many other false personae to test theories about such human behaviors as fraud, racism, and greed.
And some of their targets have protested in the past. One of the most infamous cases, cited by Mr. Gelman in his response to Ms. Milkman and Ms. Akinola, is that of Francis J. Flynn, another Columbia researcher, who wrote to about 240 New York restaurants in 2001 claiming to have contracted food poisoning. Mr. Flynn, now at Stanford University, said he wanted to study how the restaurant owners handled complaints, and ended up being sued by 10 of them.
Nobody is talking about suing Ms. Milkman and Ms. Akinola, and Mr. Gelman readily acknowledges this is a far less serious matter than the one involving Mr. Flynn. But at least a few of the 6,300 professors are complaining loudly and looking for some kind of compensation or response.
Mr. Gelman estimates he is owed $10 for his lost time. Mr. Daileader wants the researchers to know the damage they’ve done to the atmosphere of trust at universities. Corrine McCarthy, an assistant professor of English at George Mason University, feels she’s owed some kind of apology for being falsely led to believe that a student was actually interested in the linguistics studies of a junior researcher like herself.
The professors contacted by Ms. Milkman and Ms. Akinola were divided into two groups, with some told by their fictional graduate student that he or she wanted a 10-minute meeting that same day, and others asked by the fake student for a meeting in a week.
Ms. McCarthy, among those asked by her bogus e-mail sender for an immediate meeting, wrote back saying she would be available during her regular office hours from 10 to 11 a.m. that day. The researchers sent out immediate cancellation messages to those who accepted, explaining what they did and why, but Ms. McCarthy didn’t find that follow-up e-mail message until after sitting in anticipation the full hour.
Ms. Milkman and Ms. Akinola said in their cancellation messages that they hoped to test previous research showing that people “tend to favor doing things they viscerally want to do over what they believe they should do when making decisions for now, while they are more likely to do what they believe they should when making decisions for later.” They also varied the names and genders of their fabricated students, testing what those differences might cause in response rates.
Human-Subject Approvals
Neither Ms. Milkman nor Ms. Akinola responded to requests from The Chronicle for comment. In their follow-up messages to the deceived professors, they said the experiment was approved by the institutional review boards at both Penn and Columbia, and that those boards were prepared to answer any questions about their “rights as a research subject.”
The decision to use deceit in a research experiment is a “really sensitive” matter, said Devah I. Pager, an associate professor of sociology at Princeton University who has used the technique in her exploration of racial discrimination throughout society.
Ms. Pager said she couldn’t assess the propriety of the Milkman-Akinola experiment, but she said she placed strong emphasis on ensuring trust between faculty members and students. “It’s not the same as the type of trust between an employer and its employees or its customers,” she said.
Others, both critical and supportive of the Milkman-Akinola experiment, also suggested at least the possibility of allowing differences between deceiving professors and deceiving most other members of society.
Sandra M. Sanford, director of the Office of Research Subject Protections at George Mason, said she disagreed with a suggestion by Ms. McCarthy that Ms. Milkman and Ms. Akinola should have obtained prior consent from the institutional review board at every university where they contacted a professor. “It’s not possible” to get permission from hundreds of universities, for the sake of perhaps only a handful of professors at each institution, Ms. Sanford said.
Ms. Sanford said her review panel, however, would have expected Ms. Milkman and Ms. Akinola to seek its permission if it appeared they were specifically interested in George Mason professors. In an earlier unrelated case, she said, the George Mason review panel saw no need for its researchers to gain the approval of stores when the researchers proposed sending purported job applicants into the stores testing whether their success was affected by wearing clothes of particular cultural or religious affiliations.
One key factor in the panel’s approval in that case, Ms. Sanford said, is that the study did not pursue a single store or chain of stores. The board also regarded the store managers collectively as a single entity at each store, she said, rather than individuals deserving any human-subject protection. She said she believed the university professors contacted for the Milkman-Akinola study, by contrast, should have been regarded as individuals.
Differences and Regrets
T. Mills Kelly, an associate professor of history at George Mason with his own controversial teaching practices, said Ms. Sanford’s review panel probably would not have approved the Milkman-Akinola request if it came from George Mason professors, saying the board “is really touchy about anything like that.”
Mr. Kelly has gained attention for experiments such as having his class post to Wikipedia the fictional tale of a pirate who stalked the Chesapeake Bay in the 1870s, to help the students gain a skeptical attitude toward the reliability of historical accounts. Mr. Kelly never sought review-board permission for that exercise, feeling it didn’t technically involve human subjects. The Milkman-Akinola method differed in that they sent their lie directly to a few thousand professors, he said, rather than let an unknown number of people find it on the Internet.
“There’s a difference,” Mr. Kelly said, “between push and pull.”
The executive director of human research protections at Penn, Yvonne K. Higgins, said she looks forward to having her institutional review board examine all the feedback it receives from the experiment. “We’ll talk about how we could improve this particular study, how we might do things differently in the future,” she said.
And as for Mr. Gelman, it’s taken far less than a week—the time frame set out by Ms. Milkman and Ms. Akinola to measure differences in behavior—to realize his mistake in complaining publicly. Mr. Gelman said he now just wishes the two researchers had never sent the second e-mail message explaining that the graduate student seeking advice was a fabrication.
“Then I would have never thought about it because it would just be one of various students who didn’t respond,” he said. “Every time a student asks for help and I offer them and they don’t respond, it hurts a little bit, but it happens a lot.”
Referring to the sarcastic response he posted to his blog, he conceded: “I probably overreacted.”