Wednesday, September 05, 2012
The strange case of Marc Hauser
As you may recall, Hauser was the well-known Harvard psychology professor and author of several popular books who resigned after he was accused of research misconduct. The National Institutes of Health has now completed its investigation, and yep, he did it.
In fact, it sounds worse than the accusations which have previously been made public. What we knew about was that he coded observations of monkeys without establishing intercoder reliability, and without being blinded to the stimulus, and that the results were not replicable when others viewed the videotapes. That is definitely unacceptably sloppy research, but may not have constituted deliberate fraud -- he could have been fooling himself. The investigation finds that he also outright fabricated some findings.
(I'll digress for a brief comment on research methods. In social psychology, sociolinguistics, and sociological research, we often want to classify behaviors, such as categories of speech acts, orienting toward a stimulus, categories of gesture or body language. Identifying these requires human judgment. In order for us to believe that the coding decisions really do reflect some underlying reality, and not just the idiosyncratic or possibly biased, we want to see the same data coded by multiple people and that there is a high level of agreement. That's called intercoder or interrater reliability. Second, if the point of the study is to show that a particular behavior results from a particular stimulus, we want the coder not to know what the stimulus condition is, to prevent conscious or unconscious influence on the results. Without both of those conditions, nobody should believe the findings.)
Okay, he's not the first person to commit scientific fraud, that's for sure. But what is, as far as I know unique about this case, and also strange, is the apparent lack of motive. Usually these sorts of cases involve post-doctoral fellows or junior investigators trying to win grants or out-compete their peers and get ahead. Hauser already had a large body of apparently sound work and he was a tenured professor at Harvard. And his prior work, reputation, and program of research certainly did not depend on a specific species of monkey being able to parse syllables or syntax, or whatever it was he was specifically looking at. A negative finding in this particular experiment would not have somehow caused him any major problems.
Why risk everything for so little, really nothing at all? People are hard to figure.