The Pygmalion of Greek myth was a sculptor. He carved an ivory statue – Galatea – so beautiful and realistic that he fell in love with it. At the altar of Aphrodite, the goddess of love, he wished that she would come to life, as he had no interest in other women. When he returned to the statue, she came to life and they went on to marry and have children. His name has been lent to the Pygmalion Effect – the more positive the expectation placed upon someone, the better they perform.
The Pygmalion Effect
This is also eponymously called the Rosenthal effect, after Robert Rosenthal, who conducted a number of experiments into self-fulfilling prophecies. In 1968, he, alongside Lenore Jacobson, published “Pygmalion in the Classroom”, incorporating a study at Oak School in California. In the experiment they randomly identified around 20% of the students as having particularly high potential for the coming year and they then informed the teachers about these particularly talented individuals. The results showed that, for younger pupils, being labelled as high potential delivered a statistically significant improvement in IQ increase above and beyond that achieved by ‘normal’ students. Thus the first evidence for the impact of expectations on performance was found and a new area of research was born.
This study, however, created just as many questions as answers – why were only the younger children affected? Does that mean expectations play no role for adults? Could it be due to teachers giving more support to those students (and thus less to the rest), meaning the effect can’t be used to benefit everyone in a group? Was any effect due to the teacher’s expectations or was it due to the students identifying themselves as a strong performer?
Research at around the same time (1968 and 1970) by Major Wilburn Schrank looked into these questions, to varying extents. He used similar treatments to the experiment above, but Schrank’s experiment was conducted at the US Air Force Academy Preparatory School (ages 17 and over) and labelled whole ‘sections’ (i.e. classes) as more or less talented. In the 1968 study the teachers were told the labels were authentic, whereas in the 1970 study they knew it was random, while everything else was kept the same. In the 1968 study, the same effect was seen as for the younger pupils at Oak School, showing that the impact of expectations continued beyond youth and that whole groups could benefit (One major caveat – there was no control for how much effort the teachers invested in each group outside of the classroom, such as lesson planning). In the 1970 version there was no effect, revealing that teachers’ expectations were the key variable rather than the students’ beliefs. This still only looked, however, at a narrow age range and a very specific work environment.
In 2000, both Nicole Kierein & Michael Gold and Brian McNatt conducted meta-analyses of the Pygmalion Effect within work organisations, providing us with some more thorough answers. Kierein & Gold looked at 13 different studies within organisations and found that there was a significant effect (for the statisticians out there, the overall effect size was a d value of 0.81 – for the non-statisticians, Cohen said anything over 0.8 represented a “large” effect), while McNatt looked at 17 studies, finding an effect size of 1.13.
So overall we can be confident that, when looking across a whole population, there really is a Pygmalion Effect. It is not, however – and as ever – that simple…
The Opposite – Golem’s Emergence
Every silver lining has its cloud, and the Pygmalion Effect’s counterpoint is the Golem Effect – if we have negative expectations of someone then they are dragged down by them. In fact, both meta-analyses found that the Golem Effect exceeded the effect size of its Pygmalion brother (to reassure readers that psychologists aren’t too evil, the Golem Effect isn’t tested by randomly designating people as incompetent. Researchers ask managers what their expectations are for their staff, and then actively tell them that their expectations should be average rather than low for a randomly selected portion of those designated ‘low expectation’. Therefore the ‘control’ group is the one that experiences the Golem Effect and the ‘treatment’ group is de-Golem’ed).
Therefore restraining the negative is at least as important as maximising the positive. This is particularly important if expectations are part of a zero sum game (i.e. if designating some as ‘high expectation’ means others become ‘low expectation’). This can happen naturally; we baseline our expectations based on the people around us and then judge people relative to the baseline. If you suddenly start working with an exceptional employee, then you might reasonably wonder why the others aren’t as good – even though before working with the exceptional employee you thought they were all fine.
That’s one reason why holding expectations for a group can be useful, particularly if it’s the group who’s performance is most important to you (and, equally, the group which is most influenced by your expectations). By placing expectations on a group you avoid the need to compare members within that group to define your baseline.
It turns out that the Pygmalion Effect, while prevalent, varies in size depending on who you are and the environment you’re in. The meta-analyses found three specific moderators on effect size (I really can’t emphasise enough that it is only possible to identify moderators that we record, by definition, so we’re unlikely to see anything too subtle. Only the very basics about participants are normally recorded – there are a number of other potential, but unproven, moderators).
The first moderator they identified is gender. Both analyses found that men were more strongly influenced by expectations than women. It seems likely that this is a societal and cultural effect – and these analyses are now 15 years old and relate to research older than that. Senior levels within organisations were more male dominated then and men had increased opportunities; this may have motivated men towards trying to fulfil expectations or the men in leadership positions may have invested disproportionately more in ‘high expectation’ men – the reasons aren’t clear.
More recent research (2008) by Gloria Natanovich and Dov Eden showed that the gender of the ‘expecter’ was not, at least in a specific environment, an important variable, while also suggesting that ‘expectee’ gender didn’t matter either. As workplace culture becomes increasingly balanced, these results suggests that biases in Pygmalion effect will gradually fade away. In the interim, however, we need to be aware of how workplace biases can play a role.
Secondly, effect size is influenced by initial expectations and/or performance. The lower the initial expectations, the larger the impact of the Pygmalion Effect. This follows logically – the greater the scope for change in a variable, the larger any effect should be.
Thirdly, workplace environment has a significant influence. In particular, both analyses found a stronger impact in military environments than in other workplaces. Perhaps this is due to the hierarchical environment in the military leading to people investing more in living up to their manager’s expectations, in comparison to the less ingrained and less linear (you’re not as strongly ‘owned’ by any one person) relationship in non-military organisations.
This reflects a broader concept – which stretches far beyond the Pygmalion Effect – that I believe could use a lot more attention; analysis of the differences in the impact of psychological phenomena on different groups (here I am simply defining a group as a number of people who share any specific trait). It is both easy and useful to have a global analysis; it provides more straightforward actions, while only needing random samples from the total population to identify an effect. It’s also controversial and complicated to start to look at actions that would be targeted at specific segments. It could, however, help both employees and employers and it is, in my opinion, something we should explore.
As a final twist in the Pygmalion tale, we need to think about assessment of the quality of someone’s performance. In general terms, our perception of performance follows this equation (this applies equally to products as it does to people):
- Perceived Performance = Actual Performance – Expectations
Therefore the higher our expectations, the lower our perception of the quality of the same actual performance (for example, if I bought a watch for £10 and it broke after a couple of years then I’d think that was fair enough. If I bought a watch for £500 and it lasted a couple of years, then I’d feel ripped off).
The Pygmalion Effect means that raising expectations of employees (or students) can lead to increased actual performance, as we know. But if the increase in expectations is larger than the increase in performance, then we can still end up perceiving a decrease in performance. Therefore we are disappointed in delivery, probably express this to others and suppress future performance.
We need to be aware of this and do what we can to act to correct against it. Using objective metrics enables us to make direct comparisons, but mostly we just need to reflect on this when thinking about how individuals (and groups) are performing.
By understanding our expectations, and how to use them, we can improve performance, increase individual motivation and appropriately recognise people for their successes, making this a topic worth exploring – after all, Pygmalion did, eventually, get the girl.