Pygmalion in the Classroom: Revisited

Whether you’re looking generally at Education and Methods in Context or specifically at teacher expectations as an “Inside School” factor in differential achievement, a useful study to have in your locker is Rosenthal and Jacobson’s “Pygmalion in the Classroom” (1965) experiment. Accessible examples of experiments are quite rare in sociology and “Pygmalion” can be cited as either a field or laboratory experiment, depending upon how tightly you define the latter (which is an interesting evaluative point in itself).


Rosenthal and Jacobson’s research focused on the so-called Pygmalion Effect, named after the mythical Greek sculptor who fell in love with a statue he’d created and which was given life by Aphrodite the goddess of love. A more familiar contemporary version is George Bernard Shaw’s play Pygmalian (1913), in which Henri Higgins, a professor of phonetics, wagers a colleague that he could transform an uncouth working-class flower-seller (Eliza Doolittle) into someone able to pass as a Duchess in upper-class English society, simply by teaching her how to speak, dress and behave in ways expected by the latter.

The underlying ideas here are those of transformation and improvement through certain types of interventions. In the Shaw play, for example, these interventions involved an upper-class professor teaching a lower-class woman how to successfully pass herself off as a member of “polite society”, despite her lack of education and “class”.

Shaw’s basic argument here was that differences in social class were not somehow the result of “natural differences” but rather the outcome of cultural choices and by managing the expectations of significant players – in this imaginary instance the ability to convince Eliza she could develop ways of speaking and acting that would allow her to pass as a member of the upper-class – you create a powerful expectancy effect that brings about the desired transformations and improvements.

This relates to Rosenthal and Jacobson’s study in the sense that it was designed to test the idea that differences in educational achievement were not simply based on “natural differences” in “intelligence”, measured through success or otherwise in school examinations, but were actually based on the expectations teachers had for their pupils. In other words, they wanted to test whether or not changing the way teachers saw the abilities of their pupils would bring about a change in the measured intelligence of those pupils.

In order to achieve this Rosenthal and Jacobson, just like their fictional counterpart Henri Higgins, introduced a series of transformations and improvements into selected classrooms through a range of interventions, most-notably those that involved “telling teachers that certain children could be expected to be “growth spurters” (i.e. academic improvers) based on the students’ results on the Harvard Test of lnflected Acquisition” – a respectable-sounding “scientific test” whose existence was entirely fabricated by the researchers.

In the Classroom.

As we know “the test was non-existent and those children designated as “‘spurters'” were chosen at random” and what Rosenthal and Jacobson hoped to determine by this “was the degree (if any) to which changes in teacher expectation produce changes in student achievement” – something, we further know, that broadly occurred in relation to both gender and ethnicity: those children who had been randomly identified to teachers as “special children” on the basis of a (non-existent) IQ test did indeed see an increase in their IQ level a few months later.

The problem here is that while Rosenthal and Jacobson quite-reasonably conclude that “The results of the experiment…provide further evidence that one person’s expectations of another’s behavior (sic) may come to serve as a self-fulfilling prophecy. When teachers expected that certain children would show greater intellectual development, those children did show greater intellectual development”, this is often uncritically expressed as the idea that if teachers expect their students to achieve academically they will (and, by extension, if they don’t, they won’t).

The problem here, however, is that the research reality is somewhat more nuanced…


The conclusion usually drawn here is that teacher expectations raised achievement across the board in those students who were “predicted to be spurters”. But this is not the case, as Rosenthal and Jacobson note, because “each group profited more from teachers’ prophecies in the area of intellectual functioning in which they were already a little advantaged”.

This is an important qualification because it argues expectancy effects are not universal, as evidenced by Rosenthal and Jacobson’s pre-and-post intervention conclusions:

(Before “special children” were identified)  
Boys had higher verbal IQ than girls.
Girls had higher reasoning IQ than boys.  
(After “special children” were identified)  
Boys increased their verbal IQ
Girls increased their reasoning IQ.  

The progress made by boys and girls wasn’t uniform (universal) across the different subtypes of “intelligence” supposedly being tested:

  • in measures of verbal IQ those boys who began with a higher verbal IQ gained the most.
  • in measures of reasoning IQ, those girls who began with a higher reasoning IQ gained the most.
  • In other words, it was only where the results of the pre-test conformed to general teacher expectations about ability that there were significant post-test increases in IQ.

    The significance of this more-granular breakdown of experimental results is that it suggests is teachers hold graded expectations of different groups, such as boys and girls, rather than necessarily holding an overall picture of the abilities of such broad groups. Where boys, for example, began with a higher baseline set of teacher beliefs about their verbal ability this was the area in which they improved following Rosenthal and Jacobson’s intervention, with the reverse being true for girls and reasoning ability.

    What this demonstrates, therefore, is that expectancy effects are neither uniform nor universal and we can’t simply conclude that “positive labelling”, in-and-of-itself, is a key variable in increased levels of attainment. Rather, the teacher’s underlying beliefs about student ability coupled with an acceptance of the “scientific study’s” ability to identify “bloomer children” seem to be critical variables.

    If, for example, the underlying belief is that girls are better at subject X than boys, the Pygmalion Effect improves female achievement in this specific area. If teachers have no such underlying belief, there is no such effect.

    This is significant because conventional readings of Pygmalion in the Classroom tend to emphasise the idea that teacher expectations (or labelling if you prefer) are the crucial variable in the sense that if teachers have a positive view of their pupil’ abilities (whatever this may actually mean in reality) their pupils will display higher levels of academic achievement (with, as we’ve suggested, the reverse being true by implication).

    Rosenthal and Jacobson were, however, keen to point-out that this simple reading was incorrect when they noted the results of the same basic experiment carried-out at schools with different student intakes.

    In broad terms, while Rosenthal and Jacobson’s original experiment (at “Oak School”) involved pupils drawn from a “lower-class community” with a substantial proportion of “minority-group members”, subsequent experiments at “two elementary schools located in a small Midwestern town” drew their pupils from a mainly “middle-class community” with few minority-group members.

    In the latter case, a similar identification of “special children” not only found “No expectancy advantage for either boys or girls as measured by total IQ or verbal IQ” but also a reversal of the “reasoning results” found at the original (Oak) school:

    (Before “special children” were identified)  
    Boys had higher verbal IQ than girls.
    Girls had higher reasoning IQ than boys.  
    (After “special children” were identified)  
    Boys increased their reasoning IQ
    Girls did not increase their reasoning IQ.  

    Now it was the boys who showed the benefits of favorable (sic) teacher expectations. Those who had been expected to bloom gained over 16 IQ points compared to the less than 9 gained by control-group boys. Among the girls it was the control-group children who gained about 15 IQ points while those of the experimental group gained just over 5”.

    The significance of this reading, therefore, is that while we may be reasonably sure teacher expectations play some role in student achievement, this is neither universal – applicable to “every school in every situation, regardless of intake, class and ethic background and so forth” – nor uniform: expectation outcomes depend to some unknown extent on the initial expectations teachers have about their school and pupils, allied to the social and intellectual qualities of the students involved.

    Stay Updated

    Enter your email to be notified when we post something new:

    Archived Posts