Error, Misconduct and Fraud in Scientific Research

 

Parameters:

(1) source of the problem

(2) attitude towards misbehavior

(3) type of problem

 

Source of the problem (reflecting any intent by the researcher)

(1) error (unintentional - accidental, ignorance)

(2) misconduct (some intention - belief action justified or acceptable or not that bad; "venial")

(3) fraud (intentional - known to be wrong but done anyway)

 

Attitudes toward misbehavior:

(1) would never do intentionally

(2) would do if "minor" or if "everyone else does it"

(3) would do without hesitation if undetected and for profit

Action	Error	Misconduct	Fraud
wrong observation	common	occasional	unlikely (see falsification)
wrong analysis	common	common	occasional
undeclared conflict of interest	occasional	common	occasional
publication bias	occasional	common	occasional
undeserved or unrecognized authorship	occasional	common	unlikely
suppression of data	unlikely	common	occasional
duplicate submission for publication	unlikely	common	occasional
plagiarism	unlikely	unlikely	almost always
falsification	unlikely	unlikely	almost always
fabrication	unlikely	unlikely	almost always

 

where:

• A person may choose the wrong statistical method because s/he are unfamiliar with the method. A statistician may keep changing methods until s/he gets the desired answer.

• Falsification involves recording false data for an experiment that took place.

• Fabrication involves making up data for an experiment that did not occur.

• Just because a person has the same idea as someone else does not mean that there is plagiarism.

 

Ways to reduce problems:

(1) Planning an experiment well and in advance can significantly reduce accidental errors.

(2) Careful records of experiments should be kept that cannot be altered without detection and that can be audited.

(3) People should be trained in good scientific practices and what constitutes appropriate ethical behavior.

(4) Problems of potential conflict of interest or in analysis should be addressed prospectively rather than retrospectively.

(5) Undue pressures should be avoided.

(6) Organizational leaders should provide active oversight and review of research activities. There should be a mechanism for reporting possible misconduct or fraud (whistleblowing) that is taken seriously.

 

Limitations:

• An error in judgment can be morphed into criminal intent during a witch hunt.

• Misconduct may be overlooked if politically expedient.