In 1963, Robert Rosenthal and Kermit L. Fode supervised an animal study performed by students in order to gain further experience in duplicating experimental findings and, in addition, to introduce the students to the field of animal research and to help them overcome any fears that they may have with regard to working with rats.
The students were told that this experiment was a repetition of work done on maze-bright and maze-dull rats. Many studies have shown that continuous inbreeding of rats that do well on a maze leads to successive generations of rats that do considerably better than ‘normal’ rats. Furthermore, these studies have shown that continuous inbreeding of rats that do badly on a maze leads to successive generations of rats that do considerably worse than ‘normal’ rats. Thus, generations of maze-bright rats do much better than generations of maze-dull rats.
Half of the students were then assigned a group of maze-bright rats and the other half worked with maze-dull rats. The maze-bright rats were expected to show some evidence of learning during the first day of running and that performance should rapidly increase thereafter. In contrast, students assigned the maze-dull rats should find on average very little evidence of learning in these rats.
And indeed, the expected results were later confirmed in the experiment:

In the elevated T maze experiment (the two arms were interchangeable and one was painted white while the other was painted black), a correct response was recorded when the animals went to the darker platform (in order that the animals do not simply learn a position response, the position of the darker platform was varied throughout each day’s run.)

Even if you haven’t heard about this study you might have guessed by now:
The study subjects were not the rats but rather the psychology students unknowingly became subjects themselves!
In reality, all rats were standard lab rats which did not actually differ significantly in their inherent attributes, and were randomly assigned to the “dull” and “bright” conditions. However, the results showed that the rats labeled as “bright” learned the mazes more quickly than those labeled as “dull.” Apparently, the students had influenced the performance of their rats, depending on what they had been told. Importantly, these students were not cheating or purposefully slanting their results. The influences they exerted on their animals were unintentional and unconscious.
Rosenthal is now well known for his research on experimenter expectancy effects, the influence that a researcher can exert on the outcome of a research investigation. As a result of such research, the threat of experimenter bias to scientific research has been well established. These experiments demonstrate how important it is to use careful procedures (such as blinding and randomization methods, in which the experimenters who come in contact with the subjects are unaware of the hypotheses of the study) to avoid most of these expectancy effects.