Unconscious Bias and the Impact on Women entering Science

This is text of the talk I gave at the Howthelightgetsin Festival at Hay at the weekend. The talk was misleadingly entitled ‘Saving Science’ by the organisers, trailed as how women can ‘save’ science, but it is really about how unconscious bias hinders women during their careers, and why we need to do a better job of nurturing talent amongst this half of the population.  The video of the talk can be found here.

What’s the first thing you think when you look at me standing here. Perhaps it’s you thought speakers would have more dress sense. Perhaps it’s that I don’t look like a scientist. But you will certainly have formed some snap judgement as to whether I am what you expect, which implies you have some expectations given what you have read about me that provoked you into buying tickets for this event (unless it was merely the very inaccurate title of ‘Saving Science, not the title I supplied!). We all make rapid evaluations all the time based on what we expect, and that in turn is based on preconceptions and bias.

This starts at birth: the way we handle and talk to newborns and babies is determined by their gender. You may have heard the story last week of a couple who are refusing to state what sex their baby – called Storm – is, to try to overcome this. Whether this will work or just cause confusion in the adults around as well as the growing child is of course unclear. So what do you expect a scientist to look like? Stereotypical images are often portrayed as mad scientists in white lab coats; more realistically one could consider leading scientists such as Paul Nurse or Martin Rees and of course, Brian Cox who is much on our TV screens. But almost certainly your first thoughts of a scientist will be male.

Can you even name a female scientist? Most of the population can’t, as revealed by a survey published last year, although there are some rather familiar examples. Historically one could cite Marie Curie and Rosalind Franklin; currently Susan Greenfield and Jocelyn Bell Burnell are both very visible. The evidence is that most people implicitly think scientists are male, probably white.  This is an example of our unconscious bias.

Project Implicit at Harvard looks at many of the places in which unconscious bias lurks and tests the associations one makes. In terms of the underlying assumptions people make about science, 26% of people taking the test have been shown to have a strong association of male gender and science, and 28% a moderate association. And even I, when I do the test (which I’ve done several times) fall into exactly the same trap. Practically no one operates the other way round, associating women more strongly with science than men.

In itself that may not matter, but it does if this preconception starts spilling out into how job applications are viewed and interviews are carried out.  Let’s look at how this might impact on hiring processes, unless you are well aware of all this. It is possible to explore how men and women react to identical CV’s simply by changing the apparent name at the top of the application (for a science job). One such study (by Steinpreis et al ) showed that both men and women were more likely to vote to hire a male job applicant than a female job applicant with an identical record. So this is not about facts, it is about reaction to facts and we don’t collectively react in the same way to men and women.

Similarly, studies show there are different ways of writing letters of reference for men and women. A study by Madera et al showed that men were more likely to be described with ‘agentic’ terms and women with ‘communal’. This means that women are more likely to be described as affectionate, kind and nurturing whereas men were described as ambitious, dominant, and self-confident. Which are you more likely to hire?  But writing like this is almost certainly not intentional, it is just the way referees think about the individuals they’re writing about.

Both these two previous factors will affect who gets invited to interview. At interview itself a new set of complications come into play, exemplified by the statement ‘men are assertive, women are aggressive’.  A strong woman may well actually be marked down for doing things that would be approved of in a man. And even if an offer is made it is clear that women fall foul of attitudes to negotiation. Again there are studies which show this.  There has been a particularly interesting study from Stanford which gave female professors training in negotiation so that they could interact better with the administration when salary levels and resources were set. But the administrators weren’t trained to prepare for this and reacted extremely badly. Half the number of negotiations was successful for women when compared with men. Studies of videos of women asking for a pay rise show that men tend to penalise them, but not the men doing exactly the same thing. But, a lower salary at the outset of a career can propagate throughout its entire span to mean lifetime earnings are significantly lower for women.

None of this means that any of the actions are done deliberately – just because women are in some senses ‘victims’ doesn’t mean the men – or indeed senior women – are aggressors. We are simply all coloured by our perceptions. But we ought to be able to do better to make sure that ostensibly even playing fields really are even.

What about how well girls perform in tests? Conventional wisdom has it that girls are better at exams than boys. But there are exceptions, and stereotype threat is one factor in this. Stereotype threat is an idea due to Claude Steele, the black American psychologist. He first identified it in the context of black college students undertaking exams, noting that because black students are expected to do badly they tend to underperform, whatever their innate abilities. He has written an extremely interesting book on his – and other people’s studies – to this effect, Whistling Vivaldi. The same effects are found for girls and maths/physics tests.  ‘Everyone knows’ girls can’t do maths, so they feel the burden of this message and consequently underperform.  Set a maths test when girls are told it is actually to look at how students carry out problem-solving in general and they will do significantly better than when they are told it is to test their maths ability.  A similar effect is observed if senior citizens are given tests to do with memory. Depending on whether that is spelt out or if it is described in some other way, they will perform worse or better respectively. People appear to be much more suggestive than perhaps is anticipated.

The ‘cure’ for this, or at least a tactic that ameliorates the suggestion, is to have some sort of self-affirmation exercise in advance of the test, and this then apparently overrides the anxiety associated with the test. One recent study by Miyake in Colorado showed that getting students, girls in particular, to write about things they cared greatly about for a few minutes at the start of the year was sufficient to improve their physics test marks substantially throughout the entire year.

So far, these are fairly general findings about gendered differences and unconscious bias. So let me know turn to the specific issue of women in science.  In some senses I am working backwards through life, so let me now turn to the early years before any decisions are being made about careers or higher education.  As I mentioned earlier there is evidence that the very way adults interact with newborns is coloured by their gender, including the faces they make at the child and the way they handle it, boys being subjected to more rough and tumble from a very early age.  Everything about the way society interacts with children collectively seems to be pretty gendered.

For instance, let’s take advertisements for toys and the recent word cloud analysis. This showed that the two words used most heavily in advertisements for boys’ toys were battle and power, for girls the corresponding words were love and magic. Need I say more! Any girl who thinks she is going to sail through life on the basis of magic rather than power is in for a nasty shock.  What about subliminal messages – is it simply irony to dress your pre-teen child in a t-shirt that proclaims ‘I am too pretty to do maths’? Or if we look at the attempt Mattel has made to make Barbie more ‘modern’ we see that Barbie computer engineer is unrelentingly pink and cute.

What messages do the media convey with so few women science experts being wheeled out on mainstream science programmes? Even if the numbers of women are low so that they have to search out for them, do they have any sort of responsibility to try to find them? Indeed, as the blogosphere has been discussing at some length, where is the female Brian Cox? Where are the role models?  We should be deliberately trying to put out positive messages about women in science. For instance, we can propagate powerful images, as exemplified by those stunning black and white photographs the UKRC has produced each year to provide role models and encouragement to offset some of these negative messages. One can argue endlessly about Nature versus nurture; you can believe the Simon Barron-Cohen analysis that says there are more men at the extreme ends of essentially any parameter, and even that the ‘mean’ for intellectual or maths ability is shifted up for men versus girls, but that is irrelevant when it comes to the individual.  As a society we should be asking what are we doing that is hindering at least some girls fulfilling their potential in science?

The examples I have given above are all part of the reason why women may drop out or be forced out of science to a greater extent than men: people’s impressions, the ways hiring is done and the marks women score in exams being affected by stereotype threat, thereby weakening their applications. This has little to do with the women themselves.

Let me turn briefly to some statistics. Already by the time of starting at university we see massive imbalances in some courses: less than 20% girls on engineering courses, slightly better in computing and physical sciences, but around 70% in veterinary sciences and even higher in sciences allied to medicine.  But, once in academia the girls who enter the subjects may not stay there. For many different reasons they find the environment not to their taste. In chemistry, 30% of researchers starting out at postdoctoral level are women, but currently there are fewer than 10% at professorial level.  Even in biology we have only reached a level of around 15% professors. We have a well-documented leaky pipeline with a massive hole which needs to be plugged if we are to derive maximum benefit from the talented girls who enter university full of hopes and excitement.

The leaks have many causes – the issue of unconscious bias I have already addressed.  But many other causes have been identified. An obvious one is the hurdle of combining an active science career with child-bearing. Until too recently senior women were still heard to say upon occasion that it wasn’t possible to be successful and have a family – for instance one of the women elected to the Royal Society at the same time as me (and the one interviewed by the press) said exactly that upon election in 1999, although two out of the five women elected that year had indeed had children (including me).  This isn’t only a case of working practice (including the practicalities of flexible working) making the combination possible, but also of role models demonstrating it can be done.  At least we aren’t in the position the US seem to be, of expecting women to hide career breaks on their CV’s.

The working environment may also turn women off. It has to be said that science, not always incorrectly, is seen as ultra-competitive and needing long hours at the bench. Neither factor may appeal much to people of either gender, but perhaps disproportionately so to women. It isn’t clear that competition is necessarily helpful – much modern science is interdisciplinary and necessarily needs collaboration and teamwork. This doesn’t always seem to have percolated through people’s mindsets.  Numbers of women are improving, but at a glacial speed.

Why does this matter? Why do I think we need more girls entering the sciences and staying there? There are many answers, ranging from fairness to the individual to the good of society. Let me start with the latter. Many people have argued that diversity means exactly that – not just a diversity of constitution of boards or workforce or whatever, but also a diversity of approaches and mindsets. It has been claimed that companies with a higher proportion of women on their board fare better, so one could produce an economic argument for companies to employ more women scientists. But that does not mean there is a feminine science which is somehow different from men’s. Merely that women may approach problems in different ways and for the toughest problems you want to look at the questions in as many different ways as possible.

Secondly, there is a shortage of skilled workers in STEM, as reported by – for instance – a recent CBI study. If we are discouraging half the population from entering the workforce – or even university – in these areas, we are creating a massive and unnecessary problem for society. Finally, there is the issue of fairness for the individual. Is it reasonable that talented young women are not being assisted to fulfil their potential due to out of date thought processes and unconscious bias? None of this has anything to do with the issue of families and children, or even overt hostility, or harassment.  But, cumulatively, we are disadvantaging half the workforce in ways that are fairly systematic if unintended. I believe these factors matter for us as a society and for the individual. We all need to develop antennae to see when we are reacting in ways that are pre-programmed rather than due to the actual individual in front of us.

So, we as a community must do better. We need to start at the earliest stages – as parents, teachers and friends – to make sure we are not subliminally giving off messages that say girls can’t do science, engineering and maths. That it is OK for them to be interested in clothes or boys (or art or the media or whatever else takes their fancy) AND science, so that those whose interests lie in the scientific direction feel comfortable following their inclinations.  We need to be sure working practices don’t discourage women who do enter the scientific professions. That we support and encourage those who do, where necessary providing mentoring and confidence-building exercises; that we celebrate women who do succeed and ensure that young women can see visible role models. We need to get the best talent performing to the best of their ability.

And each and every one of us needs to scrutinise how we perceive those around us – be they women scientists or male nurses – to check our unconscious biases are not taking over our rational thoughts.





This entry was posted in Equality, Women in Science and tagged , , , . Bookmark the permalink.

8 Responses to Unconscious Bias and the Impact on Women entering Science

  1. MGG says:

    Wow! Very nice!

  2. The Festival organisers have now put up the video of the talk here ,

  3. This is a great summary of the problem, Athene. Was your talk well-received?

    • I certainly got a good range of questions (not shown on the video – by that point the noise from the so-called ‘acoustic tent’ was getting to be deafening so they may have had trouble recording them!). People seemed genuinely interested, and various people caught up with me afterwards to say they’d enjoyed it. But, that was a relatively small (40-50 I guess) number of people, and I have no idea if the video will get much attention. It’s there for people to digest at their leisure….

  4. Steve Caplan says:

    Great talk, Athene. One minor thing that I would note is that when someone says the word “scientist”, I probably do think of women first rather than men. This is because in my graduate program in Israel, 2/3 to 3/4 of the students were female. And now in my lab, at least in the meantime, there is the same ratio of female to male scientists (but reversed in faculty members). The problem, as you note, is going to the next step–beyond student or postdoc. But for me, anyone–including technicians–are scientists.

    I also agree completely that it’s all an issue of perceptions. Although one certainly cannot rule out minor differences in the way each gender might process different types of thinking, I have no doubt (in my non-statistical assumptions) that the differences between individuals and their abilities within each gender by far outweigh any potential differences between genders. And anyone who took even an elementary statistics course can figure out what that means.

  5. Ursula Martin says:

    Great piece Athene! I liked your opening line, and I am resisting plagiarising it for a book review I am taking a long time to write.

    The UK REF (for non UK readers just be glad it’s not you) has a report that mentions bias in citations of papers by women, and advises panels to be mindful of this.
    Similarly there are various papers looking at bias in accepting papers, awarding grants and so on.

    In the UK more and more Universities are using metric data for internal ranking systems/performance management etc. Do you know of any kind of report one one can point university managers to so that they can aspire to do things properly, for example something from the RAEng, RS, ECU etc?


  6. Ursula, I am not aware of any document which promotions panels could be made aware of unfortunately. It’s an interesting idea. I guess it’s up to good chairing and awareness from the Chair that these things happen and matter.

    I have just come across a recent post by @samalsbury at Sciencegirlsrock about bias which reports a relevant study I hadn’t come across before. I can’t access the full text but apparently the study carries out a computer simulation of how men and women in an organisation become distributed at different levels if there are small differences in how they are perceived (and so promoted) according to gender. Even if the difference is as small as 1% advantage to one gender (OK, men) over the other, there is a substantial difference in the proportion of men at the top (65% to 35%). What may seem like a trivial level of ‘bias’ therefore has significant consequences, mirroring how many women feel the ‘real world’ operates.

  7. Hi,
    Totally agree with you, the majority of industries are dominated by males, Having just started Women in Business Module at Coventry university, equality issues are prevalent within business. Women Know the reasons why, however lack of R & D within this area is a major factor. Keep up the good work, TO ALL WOMEN, wherever you are.