Marc Hauser

After examining Harvard's records and conducting its own analysis, the Office of Research Integrity found several instances of fabrication and falsification. A voluntary three-year agreement provided, notably, for supervision of any Public Health Service-supported research, institutional certification and exclusion from PHS advisory or peer review roles.

State of the evidence

Settlement

Attribution

Directly implicated

The ORI found that results and methodology in an experiment reported in Proceedings of the Royal Society B had been falsely described, including numbers of responses and the existence of video recordings for every subject. Hauser accepted responsibility for the specified falsifications.

State of the evidence

Documented

Attribution

Directly implicated

In a 2007 Science paper, the ORI identified a false methodological statement that all tested animals could be individually identified; in reality, only approximately half could, making repeated testing of some animals possible.

State of the evidence

Documented

Attribution

Directly implicated

After his leave, a majority of the psychology department opposed his return to teaching and his scheduled classes were cancelled. Hauser subsequently resigned from Harvard with effect from 1 August 2011.

State of the evidence

Documented

Attribution

Directly implicated

The Office of Research Integrity concluded that Marc Hauser published fabricated data in Figure 2 of a 2002 paper on rule learning in tamarins: the animals had never been exposed to one of the conditions shown, making approximately half the graph fabricated. The paper was retracted in 2010.

State of the evidence

Documented

Attribution

Directly implicated

The ORI concluded that Hauser had falsified the coding of monkeys' responses in two unpublished experiments. After recoding, he acknowledged to collaborators that some data had been incorrectly coded and that the experiment did not support the original hypothesis.

State of the evidence

Documented

Attribution

Directly implicated

In an unpublished playback experiment, the ORI found that Hauser's recoding of 201 trials diverged from another coder on 36 occasions and that 29 of those changes favoured the predicted outcome, bringing the result to the threshold of statistical significance.

State of the evidence

Documented

Attribution

Directly implicated

Harvard announced that its internal investigation had found Hauser solely responsible for eight instances of scientific misconduct concerning, among other matters, data acquisition, analysis and retention, and the reporting of methods and results. He took academic leave in this context.

State of the evidence

Documented

Attribution

Directly implicated

For a manuscript on grammatical pattern learning submitted successively to Cognition, Science and Nature, the ORI found that Hauser had falsely described the coding as blind, fabricated inter-rater reliability measures and wrongly reported a result of sixteen animals out of sixteen.

State of the evidence

Documented

Attribution

Directly implicated