Rosenthal & Jacobson (1966)

Teachers' expectancies: determinants of pupils I.Q. gains

It had previously been demonstrated that experimenters could influence the results of their experiment. this is known as "experimenter bias". It was thought that this effect could describe a broader concept which has become known as the "self fulfilling prophecy". This means that when a person is labelled as being a particular type of person then often that person will change their behaviour to fit the label. For example, if a teacher calls a pupil "the class clown", the pupil could display more clownish behaviour.


In this field experiment, all the children in a primary school in America are used as subjects. In the school there are six grades, which are equivalent to the six years of a British primary school. Each grade, or year, was split into three streams (above average, average, and below average).


The experimenters told the teachers at the school that they were going to administer an intelligence test that would determine which children would be academic "bloomers". These children would stand the greatest chance of becoming academically bright in the future. Flanagan's Tests Of General Ability (T. O. G. A.; Flanagan, 1960) was administered to all of the children. 20% of children in each of the 18 classes were chosen at random and labelled as bloomers. Their classroom teachers were told that these children were bloomers and therefore stood a good chance of becoming quite academic, when in fact, on average, the children would have been no different in academic ability than the rest of their classmates.


After eight months the test was administered again to all of the children and the IQ gains were calculated. To check for experiment bias a blind judge, or independent researcher, without knowledge of which children had been labelled as bloomers, tested some of the children for a third time.


It was found that the children who had been labelled bloomers had significantly higher gains in IQ (p=.02, One-tailed). the greatest gains were seen in the youngest children, grades one and two.

IQ gain control subjects experimental subjects x p
ten points 49 79 4.75 .02
twenty points 19 47 5.59 .01
thirty points 5 21 3.47 .04

There have been several criticisms of this experiment. The IQ test had not been standardised for the age range of children which was used. This means the test may not have been valid for children, particularly the youngest. Teachers may not have taken much notice of the list of children labelled as bloomers. Attempts to replicate this study have not been that successful.


A natural experiment by Seaver (1973) has found some evidence to support the above results. In Seaver's experiment brothers and sisters who were taught by the same teachers were compared with brothers and sisters taught by different teachers. If the self fulfilling prophecy is true then teachers who had taught bright older brothers or sisters might well expect the younger brothers and sisters to be bright as well. This indeed was found to be the case. Similar findings were found for less bright siblings. As a control for any possible genetic link, the siblings taught by different teachers did not display similar levels of attainment.