Amidst the scandals and snafus of the past few weeks, there was one ethical cow patty you might have missed. It was one of the most blatant examples of scientific fraud in the modern era.
Diederik Stapel was a Danish social psychologist with a talent for producing research that grabbed some headlines. The topics weren't exactly paradigm-shifters, but included the sorts of buzz words that turned into easy gee-whiz features:
Receive news alerts
People who live in messy neighborhood are more likely to suffer stereotyping. People who cooperate are more likely to find similarities with each other than those who compete. Powerful people are more judgmental of the moral failings of others, while being more likely to excuse their own flaws.
Turns out that over his most of his career, Stapel simply made up a lot of his data. A preliminary report by Tilberg University, where Stapel worked, concluded that Stapel had faked dozens of studies.
"The fact is that the fraud with data has been on a large scale and has persisted for a lengthy period, so that people, and in particular young researchers entrusted to him, have been affected profoundly at the start of their careers. This conduct is deplorable, and has done great harm to science, and the field of social psychology in particular. To the best of our knowledge, misconduct of this kind by a full professor in his position is unprecedented."
Science fraud is usually more subtle.
A researcher ignores outliers and data that contradict his hypothesis. Or deliberately misinterprets the results of some else's work. Or jiggers with the scale and appearance of a chart.
According to the Tilberg report, Stapel did some of those, too. But in other cases, he simply invented the data.
In those cases, he never performed his experiments. He devised them, it is true. But then he created datasets based on zero actual work. And then he pawned off the data on other researchers working in the very field his "data" somehow applied to.
How bad is this in the realm of science? If Bernie Madoff had been a scientist, he would have been Diederick Stapel.
I had three reactions when I heard about Stapel, who has admitted his fraud and turned back his Ph.D:
1) Maybe this just proves that scientist suck as badly as other human beings. Depending on your religious or ethical underpinnings, people are either inevitably fallen or simply imperfect. Scientists are no more immune to the temptations and failures than priests, financiers or politicians. So finding fraud this enormous in science should be no more shocking than finding it in any other realm.
A study published in 2009 in the online journal PLoS ONE came up with a whole series of estimates for how often scientific fraud is committed. The meta-analysis looked at 18 studies that were rigorous enough to pass muster with the author of the report, Daniele Fanelli at the University of Edinburgh,
According to the surveys she examined, between 0.3% and 4.9% of scientists admitted they'd ever fabricated or falsified research data. Asked whether they knew of another scientist who faked results in any way, between 5.2% and 33.3% said they had.
Fanelli figured that the real numbers are higher than the results of the surveys, which depended, after all, on the honesty of the scientists who participated. (Despite a page or two of higher math, the study is reasonably accessible to a lay reader.)
And even an outright fraud like Stapel is hardly unique in the history of science. Ask Mr. Google about "Piltdown Man" for a fascinating look at an example from the early 1900s.
So feh on scientists?
2) Maybe this just says that social science sucks? After all, a lot of the experiments done in that realm seem squishy even if the data is unquestionably true. Psychology students are surely the most analyzed set of humans on the planet because they're often the subjects of psychology experiments. And often the results of the experiments are wrapped in such impenetrable jargon that it's impossible to decide if they bear any resemblance to reality.
Consider the title of one of Patel's papers: ""I, we, and the effects of others on me: How self-construal moderates social comparison effects." What the what?
But I know about other kinds of social science experiments that have been done repeatedly and with rigor.
My favorite is about something called the Knobe Effect, named after the Yale philosophy professor who came up with it. All things being equal, people are more likely to assign blame than to offer credit. The experiments have been done by many researchers and on lots of different kinds of folks across the world. (I offer a much more detailed explanation of the Knobe Effect here.)
So I'm not going to simply dump on social science because of this one terrible fraud. Which leads me to my third reaction:
3) Science, even on its worst day, is a self-correcting system.
Here's a difference between science and most religions: Religions are generally considered revealed absolute truth by their believers. God or the equivalent delivers the goods to the founder or founders. And the rest of the religion's history is about applying the revealed truths to everyday life.
That's not to say there aren't disagreements about those interpretations. But if the disagreements are important enough, like the use of the world filioque in the Nicene Creed, you end up with two religions.
No real scientist, on the other hand, claims to have the whole and unquestionable truth about anything.
Our best scientific theories become closer and closer approximations of reality. And they get closer and closer because scientists keep trying new experiments to shave the errors.
Let me backtrack on my comparison of Stapel to Madoff. The fraudulence is same but not the scale.
Stapel got away with his frauds for so long because his claims were relatively modest. If he'd made more dramatic claims, other scientists would have stepped in to duplicate his experiments to see if they could replicate his results. And assuming they failed, they would have caught Stapel in his lies.
Screwups and frauds in science get caught that way all through history. "Polywater" was a screwup that nobody could replicate. "Cold fusion" turned out to be partly a screwup and maybe a bit of fraud. In both cases, these involved claims about "How Things Work" that were dramatically different than the accepted understanding. So many scientists jumped in to see if they could duplicate the results.
Read Full Article »