Handbook for Research
Section II: Guidelines for the Conduct of Research
Issues in Data Analysis and Interpretation
Self-delusion is an ever-present danger to the creative scientist intent on making important discoveries and laboring under pressures for timely publication of the results. It is self-delusion, rather than the intent to deceive others, that underlies many reports of invalid conclusions. Strong belief in a particular hypothesis, for example, may lead to the ready acceptance of confirmatory results with disregard of the possibility that additional experiments may reveal unexpected discrepancies or throw doubt on the reliability of a new technique used.
Investigators are encouraged to develop experimental designs that will minimize the opportunity for self-deception, e.g., use of blinded samples, adequate number of samples and repetitions, and rigorously defined and evaluated controls. They are also reminded that the irksome trainee who challenges the laboratory's hypothesis may often be the most valuable member of the team, and that frequent oral presentations of data to the group and colleagues favor the timely identification of flaws.