Athene Donald's Blog

Reflections on working at the physics/biology interface, being a senior woman scientist, and anything else I feel strongly about

Stereotype Threat, Underperformance and Diversity

Posted by Athene Donald on November 18, 2010

Some months ago I came across a review in the THE for a book with the strange title of Whistling Vivaldi, by Claude Steele and it finally came to the top of my reading list recently. Far too late to write any kind of review, I will instead write a commentary on why it may be relevant to the vexed question of women in science, and any minority’s performance more generally, under certain exam conditions.

Steele’s research has identified a phenomenon he termed ‘stereotype threat’ which is defined as “a situational predicament felt in situations where one can be judged by, treated in terms of, or self-fulfill negative stereotypes about one’s group”. In other words, when taking a test, for instance, the candidate is not simply performing according to their innate ability: this ability is moderated by the impact of contingent and situational factors deriving from generally held beliefs that people have about the performance and abilities of the candidate’s grouping – as defined by race, sex, age or whatever.

This work grew out of a study Steele made of minority (in general black) students in top-ranked US universities where they seemed to underperform on the basis of their entering SAT scores. What he discovered, by carefully constructed laboratory studies, was that blacks perform significantly less well if they are reminded in some subtle way before the exam that blacks aren’t expected to be intellectually as strong as whites. If the test is introduced as one where intellectual strength is not being tested, though it may be an identical test, the blacks perform as well as the whites. Likewise, stress the fact that girls aren’t so good at maths before a test (using of course girls and boys thought to be equally strong) and lo and behold the girls perform less well than the boys. Tell them the test has nothing to do with maths ability but is exploring how a task is tackled, or some other neutral issue, and the differences go away. A whole host of different factors were studied under test conditions and in ‘real life’ and all the evidence points to this being an important factor. When a test really matters – and the effect doesn’t seem to come into play when the test is easy or people don’t care – then being reminded of a negative stereotyping can wreak havoc with your results. Even if this reminder of the stereotype is separated in time from the actual test there still seems to be damage done.  However, give the same test but without the same contingencies, so implying that the negative stereotype is irrelevant to the test, and the underperformance vanishes.

So, how does that play out in the world? If girls are consistently told by their peers, in the media or by their teachers that girls don’t and can’t do maths or science, the evidence is that this will lodge in their sub-conscious to the extent that it will cause anxiety during tests, so lowering their performance. This effect then becomes self-reinforcing; having done badly in one test they will ‘know’ that because they are a girl they will do badly in the next – and so they do, until it becomes a self-fulfilling prophesy and they walk away from the subjects. This could be one factor that reduces the number of girls doing physics and maths at A Level. No one makes similar comments about biology so the girls’ numbers, on this argument, would hold up – as indeed they do.

Melinda commented on a previous post that she thought there was a difference in writing style between men and women. I am not sure how true that is in the standard style of science journals, but what about essays in arts subjects? If girls are told they write less well in boys, does that translate into weaker exam performance?  Unfortunately there are indications that sometimes students are told precisely this: ‘write more like men’ is the message  – what an unhelpful piece of advice! I have no idea what it means, and probably the students don’t either.   If a teacher wants to recommend a change in style, they need to be able to state precisely what it is they are looking for.  For years, within my university as I believe in many, the percentage of women getting firsts in subjects such as History and English is smaller than men. Is there any connection? It is something that is clearly worth pursuing, because these are subjects where there is not an initial shortage of bright women entering the university and yet they appear not to thrive.

Since I started writing this (it’s been long in gestation), Imran Khan has written a stimulating and provocative piece about the whiteness of science in the UK.  He says

‘Thousands of people are being deterred from careers they would excel in, which is a loss to those individuals. But it also means a loss to society, and the economy. We’re talking about thousands of people who could be making advances, and might be excelling in their field, if it wasn’t due to the discrimination which seems to be built-in to our science and engineering establishment.’

Given that much of the evidence Steele cites in his book deals with young blacks in high schools and colleges in the US, one must wonder if stereotype threat is contributing to the paucity of young blacks entering higher education, and science in particular here. A number small, this approach would suggest, not just because of familial expectations and socioeconomic factors, but because they feel additionally challenged by the perceived stereotypical labels such as ‘lazy’ and ‘stupid’. Again, it would be very interesting to see this issue pursued as people try to disentangle the multiple factors which contribute to the comparative under-performance – and consequent under-representation – of certain groups of people.  I am no psychologist or sociologist, so I do not know if much research is being done in the UK on these topics, but it seems to me it could be important to check how relevant this phenomenon is on this side of the Atlantic. The more so as the evidence from Steele and colleagues is that quite simple steps can drastically improve the situation.  It is intriguing because, both the original threat leading to the under-performance, and the counter steps which seem to work, seem so slight it is hard to see that their impact can be as great as the evidence presented suggests it is.


7 Responses to “Stereotype Threat, Underperformance and Diversity”

  1. Imran Khan said

    What an intriguing experiment – I hadn’t come across the research before. Though I suppose it conceptually makes sense that lowering expectations leads to lowering standards.

    I should mention that I wasn’t trying to make a statement specifically about ethnicity in my piece – and indeed the title came from the team over at the paper.

    If I’d had more space in the article, I would have made the point about the feedback mechanism built into schools. As an example, there aren’t many female physics teachers, which means there aren’t many role models for girls who want to go into physics.

    So it’s interesting you mention the feedback mechanism here too – of low expectations leading to low standards leading to low expectations…

    To me this makes the case for positive discrimination. Or as a certain mentor of mine puts it, ‘reversing negative discrimination’. Enabling and encouraging is important and useful, but when you’ve got feedback loops like this at some point you need to step in and do something more radical.

    For instance, everyone knows about the presence and effect of the publication gap for women with families in science. Many schemes try and counter its effects downstream, but – and correct me if I’m ignorant here – I don’t know of any that try and deal with the problem in a more proximate way.

    e.g… actually getting the gap taken into account when grants and posts are awarded. I’m not enough of an expert to know how it’d work in practice, but I’d be amazed if it couldn’t be figured out.

  2. Thanks Imran, a variety of interesting points there to address.

    1 Feedback mechanisms and role models: the hard evidence, as opposed to anecdote, that I have seen regarding the importance of role models, is not that strong. It is often cited as a cause of few women physicists or whatever, but it is hard to find the proof to support it (I looked when I was writing up my draft for the recent Gender and Science meeting). I also think the role models don’t need to be in the immediate vicinity: my belief is much more important than a female teacher is a female physicist in the media e.g Emilia Fox playing a potentially Nobel Prize winning physicist in some lab drama would be much more effective!

    2 Positive discrimination has been illegal, so I assume you meant positive action, and of course I agree. There are many things that can be done to level an initially uneven playing field. Of course, in the context of university admissions this is regarded as social engineering and has been reviled and lobbied against by various parties. But in other contexts there is much that can be done. For instance, in academia job recruitment can either be done simply by waiting and seeing who applies – which may then rely on an ‘old boys’ network of some sort – or actively seeking out well-qualified minority potential applicants. I knew one head of a physical science department who used to approach all female URF’s in the appropriate field to enquire if they might be interested. I believe it never translated into applications let alone appointments, but at least he tried. However, if the trouble starts at schools and during university, in lowering outcomes for minorities, then any such actions are too late.

    3 As regards the ‘publication gap’ you refer to for women who have taken maternity leave, I think there are some very explicit things that have been done. For instance, in the RAE it was possible to declare what was (if I recall correctly) termed ‘special circumstances’ which included time out for children, as well as other factors (sick leave, heavy administrative burden etc). This meant the panels could – and did – factor in if fewer than 4 publications were submitted. My memory is in physics this practically never actually needed to be taken into account, because for most in my field 4 papers over the relevant period was not too heavy a burden. However, having said that, there were departments whose submissions gave no information about any of the permitted special circumstances (including naming early career researchers) and this was definitely an unwise, and incomprehensible strategy which weakened their submissions.

    4 For promotions panels too I have seen gaps in publications explicitly discussed, explained and then appropriately factored in. Here the problem is different. It is clear that even now some heads of departments, or others in a position of seniority, advise women NOT to declare such periods of absence (I discussed this previously Clearly they feel having children is an inconvenient fact that needs to be suppressed, and the women themselves lack confidence to override this advice. Then the promotion panel etc is left in a quandary. If they have no first hand knowledge they cannot attribute the reduced publication rate to the correct explanation and the case may then look weak. This fear of admitting to a simple fact of children is the sort of cultural factor that must be eradicated.

    5 So, many panels do deal with this specific issue in an informed way when they can. One of the key things people like myself can do is to keep reminding people that providing mechanisms whereby ‘special circumstances’ can be taken into account should and must be made straightforward. The RAE got that right. Other committees must do the same.

    However, these points do not address the early stage problems that I think the Steele’s experiments highlight, which happen either at school or university. I would be really interested to know if research is being done on this side of the Atlantic to explore the effect, or if teachers are taught about it and enabled to implement what appear to be quite minor actions to help overcome the stereotype threat. I would also be keen to know if this effect is disadvantaging other minorities, such as the example of black students being admitted to universities such as Imperial that you alluded to in your article, and what steps are and could be being taken to improve the situation.

  3. A recent very relevant paper has just been brought to my attention: The Effects of Gender Stereotypic and Counter-Stereotypic Textbook Images on Science Performance which apparently shows that girls perform less well in chemistry comprehension assessments when shown photographs of male scientists in the reading material provided, compared with when the photos are of female scientists. The authors conclude that ‘textbook images do have some impact on student’s retention of lesson material, and thus may play a role in maintaining the gender gap.’ Their research is couched within the framework of stereotype threat and they suggest that ‘research should investigate the influence of diverse role models presented in textbooks as a way of improving performance of multiple stereotyped groups, not just women.’ Very interesting.

  4. The AAUW report ‘Why so Few? Women in Science, Technology, Engineering and Mathematics’ (American Association of University Women, February, 2010, has material about stereotype threat, in particular, about the interaction between stereotype threat and beliefs about innate ability vs a growth mindset.

    The Recruitment Project at the University of Cambridge (2003-2005) did find that actively seeking applications from women increased the number of female applicants for lecturer positions. Unfortunately, we did not have enough data to see whether this resulted in more women being appointed. My own view is that it would be more effective to make sure that women were encouraged to apply for prestigious fellowships, such as URFs. in the first place.

  5. Ursula Martin said

    I am finding it a little hard to believe that this writing exercise, where students spend a few minutes in two sessions at the start of the year writing about their values and why they find them important, has the benefit claimed for it in closing the gender gap in university physics, but worth a look

    • Yes I saw this blog article too and found it surprising. However Steele indicated similar results too in his book, and I don’t think it was the same study. He called it self-affirmation, and implied it was a trivial thing to do which had big results. Perhaps, people like you and me who have survived and thrived in academia, are not the ones who would be put off in the first place by the stereotype threat, or so easily reassured by such small writing exercises. But I have no reason to doubt that for some people such simple things do make a difference – because the evidence seems to be there and one has to believe it!

  6. Ruth Wilson said

    Hi Athene,
    I will have to read this book – thanks for writing about it here.

    There are a number of resources and comments relating to stereotype threat here: , which people may find useful.

    Re your last comment, once you introduce action to overcome or forestall stereotype threat, you are working on the stereotyper (whether this is the system/organisation/attitudes of individiuals) as well as the person under threat, so the writing exercise may have an impact at different levels. That is bourne out by our experience at the UKRC ( – work has to go on with employers, policy makers, educators and individual women to bring about change.

    Best wishes,
    Ruth Wilson, the UKRC

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s