Devereux, George. (1967). From anxiety to method in the behavioral sciences. The Hague: Mouton & Co.

The scientific study of man:

• Is impeded by the anxiety arousing overlap between subject and observer,
• Which requires an analysis of the nature and locus of the partition between the two;
• Must compensate for the partialness of communication between subject and observer on the conscious level; but
• Must avoid the temptation to compensate for the completeness of communication between subject and observer on the unconscious level,
• Which arouses anxiety and therefore also countertransference reactions,
• Distorting the perception and interpretation of the data, and
• Producing countertransference resistances masquerading as methodology, which cause further sui generis distortions.
• Since the existence of the observer, his observational activities and his anxieties (even in self observation) produce distortions which it is not only technically but also logically impossible to eliminate,
• Any effective behavioral science methodology must treat these disturbances as the most significant and characteristic data of behavioral science research, and
• Must use the subjectivity inherent in all observation as the royal road to an authentic, rather than fictitious, objectivity,
• Which must be defined in terms of what is really possible, rather than in terms of what 'should be.'
• When ignored, or warded off by means of countertransference resistances masquerading as methodology, these 'disturbances' become sources of uncontrolled and uncontrollable error, although
• When treated as basic and characteristic data of behavioral science they are more valid and more productive of insight than any other type of datum.
• In short, behavioral science data arouse anxieties, which are warded off by a countertransference inspired pseudo-methodology; this maneuver is responsible for nearly all the defects of behavioral science. (pp. xvi-xvii)