Gender Equality

New study: If you need quality, you need affirmative action

New research demonstrates that when affirmative action programs are used, the quality of the applicants increases.

Affirmative action is often criticized as giving unfair advantages. Different people are evaluated by different criteria, which inevitably lowers the quality of the selected group, is the claim.

Diversity achieved through intervention is quality-compromising diversity, says the critic.

The logic behind these claims is not hard to understand, but it may be wrong.

Imagine that 100 students are going to be admitted to a university. If the historical trend is that 70 of them are men and 30 of them are women, and if affirmative action is implemented to increase the number of women to 40, the claim of detractors would be that 10 men of higher quality are being left aside to bring in 10 women that otherwise would not have been selected.

One basic problem with this logic that I’ll leave aside here is the dubious assertion that a process resulting in 70 men and 30 women is fair; I’ve discussed that elsewhere, in Equality targets as a leadership tool and in a post I wrote for, Engaging CEOs in gender diversity.

There’s a more subtle problem with the claim that affirmative action compromises quality, and two recent articles in Science show that this claim is wrong.

In Ready, Steady, Compete, Marie Claire Villeval focuses on gender differences in competitions. This can be seen in sports, where ‘boys tend to outperform girls when racing against someone else, but not when running alone.‘ In other words, competition changes the relative performance, either enhancing the performance of boys or reducing the performance of girls.

If girls are not motivated by competition — if they in fact avoid it — then reducing competition might have a surprisingly different effect than compromising quality.

What if women — even highly qualified women — opt out when they perceive too much competition? What if reducing competition increases the willingness of women to participate?

When the level of competition is reduced, the hypothesis might go, high-performing women are increasingly likely to enter the competition. When they then win, it need not be at the cost of a higher-performing man; that man might only have won against a weaker pool.

A second Science article tests this hypothesis. In Affirmative action policies promote women and do not harm efficiency in the laboratory, Loukas Balafoutas and Matthias Sutter run 360 subjects through four different repetitions of an addition task, in which they solve as many math problems as they can in three minutes.

The first time they do it, they are rewarded for each correct calculation. The second time they do it, they are groups of six — three men and three women — and only the two best performers are rewarded. The third time they do it, they can choose if they want to do it individually — and be rewarded for each correct answer — or in a competition — and be rewarded more if they are one of two winners. The fourth time they all do it in a competition again, like in the second round.

Affirmative action is introduced in the third and fourth rounds. In the third round, before they choose whether they want to do the task individually or in a competition, the women are divided into five groups and given different information about the competition. In the fourth round, everyone competes, and again there are these five different groups and models.

  1. Group one is the control group; their competition is just like that in round 2.
  2. Group two has quotas added to the competition: there will be two winners, as in round 2, but one of them must be a woman. In practice, this means that the best performing woman will always win, even if that means a better performing man is prevented from winning.
  3. Group 3 experiences weak preferential treatment: when a man and a woman have the same score, the woman wins, and the equally well performing man may not. (Remember that there are two winners in each group. If a man and a woman tie for best, they both win in Group 3. But if a man and a woman tie for second best, then the woman joins the best performer as one of the two winners.)
  4. Group 4 experiences strong preferential treatment: when a woman’s score is just slightly less than a man’s, the woman still wins, and the man may not. (If the man was best and the woman next best, they both still win. If the man came in second and the woman was third, then she will win over him, if her score was very close to his.)
  5. Group 5 has a requirement that at least one woman is among the two winners, but the scores are not manipulated. If the result of the competition gives no woman among the winners, then the competition is repeated until one is. (This could be like a requirement to re-do a hiring or promotion process if no women are on the short-list.)

What do we learn from this study?

In the third round, when subjects choose if they want to be rewarded for individual performance or for winning a competition, the number of men choosing competition is twice the number of women doing so in group one, the control group, where there is no affirmative action.

But when there is affirmative action, the number of women choosing to participate in the competition increases; this is most dramatic for the weak and strong preferential treatment seen in groups 3 and 4.

In the control group, with no affirmative action, only 30% of the women chose competition over individual evaluation; with strong preferential treatment, 70% do.

Think about what this means: when they can choose, women are significantly more likely to enter into a competition when the possibility of affirmative action is in place. Not just weaker women; highly qualified women, too.

The impact of affirmative action on the combined talent of the group of winners could go in two directions. Affirmative action could lower the collective talent of the winners if better qualified men are passed over by worse qualified women.

But affirmative action could also increase the overall talent of the group of winners if better qualified women now enter the competition.

These women could then join the group of winners based on their performance alone; the affirmative action measure draws them into the competition, but gender-balanced results in the competition are achieved without actually intervening to change any results.

The large increase in competition entry by strong female performers shows the potential of policy interventions to improve the quality of participants. It is also encouraging to observe that strong male performers do not respond to policy interventions in a negative way.

The research shows that the average ability of the group of winners is higher with some forms of affirmative action. And in this particular study, the authors note that ‘hardly any better-qualified men were passed over as a result of interventions.’ For example, in group 5, where the competition is repeated until there is a woman among the winners, it was in fact never necessary to repeat the competition.

Finally, after the four rounds of doing this task, the group was given a task that measured cooperation. The groups that had completed round four with affirmative action showed no less cooperation than those in the control group, where there was no affirmative action. Furthermore, the winners and losers in the groups with affirmative action did not differ from one another in terms of how cooperative they were either. In short, the presence of affirmative action in a competition within a group did not negatively affect the ability of that group to subsequently perform cooperatively.

The claim that affirmative action, if implemented, necessarily lowers the quality of the selected group, is illogical. Indeed, the evidence from this study makes it clear that affirmative action for women as a policy can raise the overall quality of the winners without being unfair to the men.

What do these results mean for national and local policies? What do they mean for universities? I’d like to know your answers to these questions. I’ll be writing more soon with mine.

Photo courtesy of the Nordic Council of Ministries

My interest in moving universities towards balance encompasses gender equality, the communication of scientific results, promoting research-based education and leadership development more generally. Read more



  • unnamed says:

    Thanks for sharing A lot of interesting thoughts — some of which I think are great, some of which I believe unfairly simplify a very complex situation. Allow me a little digression here…

    I believe we ignore the psychological implications and durability of our sexually-dimorphic evolutionary adaptations at our peril. In other words– men and women are, statistically, very different! And that’s not only okay, but this state of affairs has been a powerful advantage in our evolutionary past.

    I think the dominant thought in many circles nowadays is that on large scales, equality of opportunity necessarily implies equality of outcome, and that when we *don’t* have equality of outcome in every possible case, this necessarily implies inequality in opportunity. (E.g., if only 12% of engineers are women, this implies there’s *necessarily* something unfairly sexist about the training of engineers.’)

    I don’t agree with these assumptions — and I hope people don’t think I’m a bad person because of it. I think occam’s razor suggests evolutionary-driven differences in average physiology between the sexes is a significant factor in why men and women do different things in our society. Mostly, I think they just WANT to do different things (though I believe aptitudes differ as well). This isn’t a free pass for all the gender-based injustices in society. Society is often unfair to women. Society is often unfair to men. We can do better. But learning all we can about how men and women differ biologically is a necessary first step to fully understanding the problem.

    I may have recommended this a few years ago, but is kinda where I’m coming from with ‘men and women are different, and that’s okay’.

    Thanks for reading and sharing!

  • Curt Rice says:

    Thanks for these comments. I think there’s some basic stuff we disagree on, and it would be interesting to see if we can agree on what those specific points are. Here’s a few candidates; tell me where I’ve misunderstood you.

    — “Men and women are statistically very different.” In this post, for example, I suggest that men and women are different in terms of how they react to competitive situations. I don’t know yet if that difference is grounded in culture or in biochemistry, but either way, if this is the kind of thing you have in mind, we might agree (ignoring for the moment exactly what “very” means here). But if you’re thinking about intelligence or the ability to lead people or the ability to do a good job, then I don’t think there are significant differences.

    — Your “opportunity” vs “outcome” point is important, and we may have some points of agreement here. It does seem that men and women students are attracted to different kinds of studies — at least at first glance. But I know of an experiment (in Brussels) where courses like “Bridgebuilding 101” got new names like “Moving people to the other side of the river 101” which in turn attracted more women students, and indeed which changed the way engineering profs talked to their students. So, the picture is complex, and I don’t think we should just conclude that women don’t like engineering. They don’t choose it.

    — Now, is that a problem? Well, if you think intelligence is roughly equally distributed across men and women (although IQ scores give women a slight advantage), then it is a problem. We don’t want to be drawing from the bottom of the barrel when we build skyscrapers.

    — Regarding what people want, I spend much time talking about how Associate Professors become Full Professors. It turns out very differently for men and women. And the idea that this is because women don’t want to be full professors is just not plausible. Ask the ones you know.

    — Occam’s Razor for me would suggest that in fields that don’t directly appeal to things like physical strength, we should by default expect equal numbers of men and women. Deviation from that has to be explained. Part of it might be explained by what people want, and I agree that this is unproblematic. But a lot of it isn’t. A lot of women try to get into certain jobs or industries and fail because there are structures in place that impact women differently than men. And if we want to focus on quality, we need to focus on removing those structures.


    • Mike Johnson says:

      Curt- thanks for the thoughtful reply. This topic is as important as it is divisive, and I’m always happy to find places where we can create a little common ground.

      There’s lots of supporting science that men and women are different (I do recommend the link to Baumeister’s speech on the topic). Interested and good at different things, from the standpoint of statistical distributions.

      Your point about “Bridgebuilding 101” is definitely interesting. Frankly, if such a simple change can produce more quality engineers as well as help one gender (while not harming the other) I’m all for it! If you can find a link to that I’d love to read it. On the other hand, if women are on average simply less interested in certain fields, and find them less intrinsically rewarding, it might not be doing them any favors to e.g., encourage them to pursue engineering instead of nursing. I think there’s a necessary tension in modern society between making female representation in all prestigious fields equal to (or greater than) male representation in those fields, and letting women do what makes them the happiest. We may disagree on this.

      >>– Now, is that a problem? Well, if you think intelligence is roughly equally distributed across men and women (although IQ scores give women a slight advantage), then it is a problem. We don’t want to be drawing from the bottom of the barrel when we build skyscrapers.

      I do think men and women are roughly equally intelligent on average, with young girls outscoring young boys (I don’t think this is due to sexism). One big question here is whether men have a visual-spacial edge over women, on average (just as women seem to have a verbal intelligence edge over men, on average) that would make them better at building skyscrapers? Most research suggests so. I can find some links if you’re interested.

      Perhaps a larger issue, however, is whether the male IQ curve is flatter. In fact, you can take almost any quality or trait, and there will be more male variance in that trait. Height, weight, IQ, physical disability — males depart from the average much more drastically than females. (Think of a normal distribution or bell curve — this is the statistical female distribution of a trait — now press down on the top and flatten it out, and that’s the male distribution of that trait. Does that make sense?)

      We look at the top of society, and say “we’re not getting equal outcomes between genders, so we must have inequalities in opportunity!” –Men seem to have an advantage going from Associate Professors to Full Professors, CEOs are predominantly male, all 44 of the recipients of the Fields Medal (essentially the Nobel Prize for math) have been male, among many other examples.

      It doesn’t sound fair. But if we take a flatter trait distribution curve for males into account, some of the apparent unfairness goes away. To quote Baumeister here,

      >>Men go to extremes more than women. Stereotypes are sustained by confirmation bias. Want to think men are better than women? Then look at the top, the heroes, the inventors, the philanthropists, and so on. Want to think women are better than men? Then look at the bottom, the criminals, the junkies, the losers.

      The statistics on how many males vs females suffer from disabilities, drop out of school, get locked up, commit suicide, and general differences in life expectancy and outcome are quite sobering. As Baumeister notes, we look at the top and think “oh how easy it is for men, and it must be our culture’s fault that it’s this way!” – the people making these comments tend to avoid looking at the bottom. (I’m not saying your posts are without nuance here, just speaking in generalities.)

      Practically speaking, I think we’re on much the same page: society is systemically unfair to women in some ways, and holds them back, and we owe it to them and ourselves to fix it. (As you mention, “We don’t want to be drawing from the bottom of the barrel when we build skyscrapers.”) Society is also systemically unfair to men in some ways, and exploits them, and we owe it to them and ourselves to fix it, for similar reasons. But we should also be very careful not to blame culture for issues primarily driven by biology. Therein lies failed policies.

      My core point here is that men and women have different career and life outcomes, and sexism undoubtedly plays a part, but biology plays a very significant part too, and skews many of the statistics we try to use.

      Thanks for reading & commenting-

  • Duncan McDonald says:

    I have been following the debate in the UK media regarding introducing quotas for women in top boardroom posts and initially saw this article as a possible alternative to the traditional selection process.

    Please correct me if I am being too simplistic. In my hypothetical scenario affirmative action is applied to the recruitment process for a boardroom position and assumes that there are two places available with 100 candidates applying. Only one or two women are on the list of candidates however, with affirmative action one of them is guaranteed to win at least one of the places. Does this even suggest a level playing field? How can we then be recruiting the best of the bunch? What we do get is the best candidate from each gender. But what if a woman wins the first round outright? Your model makes no mention of this and assumes that women will always be beaten.

    When we talk of gender equality it is invariably loaded towards improving the lot of just one gender. If the solution is as good as the proposer argues then surely it would only take time before the argument turns on its head. What about female dominant boardrooms will the same recruitment process in favour of men apply or would this encroachment of men into a female dominated environment be deemed unfair?

    The main argument seems to be that women do not apply for posts because they dislike competition. Remove the competition and all is fair again! In a competitive world this seems a perfect solution for those who do not win but I fail to see the fairness! I wonder if Jacquie Smith agrees, after all she had the opportunity to contest the Labour party leadership but chose not to? A personal choice or was it fear of competition? I cannot see how we can avoid stigmatising women as second rate if any sort of unfair advantage is afforded to them. They are better than that and do not deserve spurious attempts to artificially push them up the ladder.

    • Curt Rice says:

      Thanks for this comment. Let me try to clarify my own position and my understanding of the research I reported on, in light of your comments and questions.

      We’ve had quotas in the boardroom here in Norway for about 10 years, and the requirement is that 40% of the boards of publicly traded companies are women (and men, for that matter, i.e. balance is the goal). So, I don’t think people actually apply for board positions; they are directly recruited. The rules then force those doing the recruiting to look at little more broadly to find qualified women. And they do.

      But that’s somewhat independent of your hypothetical scenario, so let me try to respond to that. If there are two positions available and if there is a quota stating that at least one of them must go to a woman, then you are right that the result is to take the best two overall, unless that ends up being two men, in which case, one of them is replaced by the best woman. To simplify things, let’s say that in practice a quota like this will give a spot to the best man and to the best woman. The research I was reporting on — when extended to a workplace scenario — suggests that the statement that at least one woman will be hired for the two positions has the effect of getting more women to apply, including more women with higher level qualifications. The effect was then that it was not actually necessary to modify the results, i.e. a woman did win outright in the first round.

      Regarding the men vs women issue, you are onto something very important that is going to be increasingly relevant in coming years. As noted above, the boardroom quotas in Norway apply to both sexes, so that would allay your concern for that particular case. But if part of the argument for gender balance is that the groups perform better when there are approximately equal numbers of men and women, then you’re quite right that one has to be concerned with both. If on the other hand, the problem leading to imbalance is that the selection system favors men for reasons unrelated to quality — such as gender differences about the effects of competition, as noted in the blog entry — then it could be the case that the coming predominance of women actually reflects a corrected system that really does select based on quality. For example, in Europe today, about 70% of medical students are women. Is that a problem? Do we need to give admission to more young men based on other criteria? In Norway, we have a “point” system for admission, so the discussion today is whether boys should get points just for being boys. Tough question.

      I wouldn’t agree that removing or modifying competition gives an unfair advantage. Indeed, the point of the article I reported on, which was published in Science, is that making the competition less stiff increases the quality of the resulting group. That’s quite a stunning finding.

      When ladders are built in ways that structurally favor men over women, there’s nothing spurious or unfair about trying to build a better ladder.

  • 2ndnin says:

    Questions to ask though:

    Do the men know that they are competing against affirmative action policies?

    Does it make a difference if no actual policy change is implemented (so we tell women they get an advantage then don’t actually implement it)?

    Also we have to consider numbers, in this study we have a 50/50 split so the chance of getting a winner should be fairly average, however in something like boardrooms where there are 98 male applicants and 2 female applicants does this have a more significant effect?

    Finally on bridge building 101 – the change from bridge building to getting people across the river changed the focus of the course from mechanical structure to people focus. From anecdotal evidence male identified people tend to focus more on the why does this work, and female identified on how can I use this. We need to actually study if this is true – and if so why.

  • Christine Delphy says:

    It is amazing to me the extent to which US scholars are in the thrall of biologistic reductionnism. But it is not surprising when the conclusion of one of them is that men too are “exploited”. The genders aree symmetrical; women are not exploited AS WOMEN. The French too use biology but they don’t rely all that much on “genes” in order to explain away sexism, racism and the rest of a hierarchical society.

    • Curt Rice says:

      Thanks for your comment, Christine. If I’ve understood you correctly, your claim is that women are not exploited as women. I think I understand you to mean that it isn’t the biological differences that are the basis of the exploitation. But, if biological differences correlate with cultural differences, then how could we tease these things apart? Women are exploited, and that’s something worth trying to change. And obviously some extreme cases are clearly based on biology (rape, for example), whereas discrimination in the workplace might be based on more culturally grounded discrimination. But, either way, there are problems that need to be fixed here. No?

      • Christine Delphy says:

        Sorry that sentence , “the genders are symmetrical, etc” should have been in brackets, it’s what follows from the writer’s contention that men are exploited too. Of course women are exploited as women !

      • Christine Delphy says:

        And rape is not due to any biological factor. Men are raped too. Humiliated by being treated “as a woman”. I am a constructionnist and d not think anything in pour society is biological in origin. Then the oppression of course uses teh body and therefore its idiosyncrasies.

        • Curt Rice says:

          My mention of rape as an example was in an attempt to understand your original comment, which at that point I thought was claiming that there’s nothing biological in the oppression of women. I was suggesting that the oppression of rape is in part of function of biology. Whether or not men are raped is irrelevant to this claim. It’s also not incompatible with non-biological factors. But, I guess I’m not in your “constructionist” camp, in the sense that I think biology is incredibly important in society, in category construction, in interactions. If bodies facilitate oppression, as you seem to suggest in your last sentence, then it seems to me that you’re also saying that biology is part of the story. But, if you want it, I give you the last word here 🙂 Thank for the discussion.

  • Karl Lichteman says:

    Very interesting post – thank you.

    I am wondering how this is actually transformed into the practical world. If I understand you correctly women will perform better if they feel it is less of a competition.
    Thus affirmative action in a selection process should lead to higher results for women and thus increase gender balance. However – this is only good for them, if we assume the competition stops after the selection, but in most real-life applications this is just the first competition of an endless amount of competitions that have to be tackled i.e. with your peers at Uni (for the next scholarship, or publication), with your colleagues etc.

    Instead of reducing the amount of competition in the selection process – which would possibly lead to highly-qualified candidates being accepted only to realise that they do not cope that well with the competition in their position afterwards – should we not rather work on increasing the ability of handling competition for instance in school? Considering that we are living in a more and more connected world, where most people will be able to communicate in a common language I assume that competition for resources, jobs etc. is increasing and not decreasing.

  • alex says:

    So women are more willing to compete when given an advantage? Shocking. I bet men would too. And I bet both men and women would be less likely to compete if they know the other gender gets free points.

    If women don’t perform well in a competitive environment then they shouldn’t be placed in a competitive environment at the expense of men who do perform well in a competitive environment.

    • Curt Rice says:

      I think most of your hypotheses can be confirmed. For example, your idea that women are less likely to compete if they know that men get free points is the explanation in the published research for the low participation of women. They know the deck is stacked against them, so they opt out.
      But I think you might have perceived the point of the research differently than I did. I think the idea is that the affirmative action measures send a signal that the employer cares about gender equality issues, and the extremely interesting result is that this attracts the participation of more highly qualified women, i.e. it changes the nature of the applicant pool, most often to such a degree that the affirmative action measure actually doesn’t have to be implemented. That is a shocking result, indeed.

10 Trackbacks


I encourage you to republish this article online and in print, under the following conditions.

  • You have to credit the author.
  • If you’re republishing online, you must use our page view counter and link to its appearance here (included in the bottom of the HTML code), and include links from the story. In short, this means you should grab the html code below the post and use all of it.
  • Unless otherwise noted, all my pieces here have a Creative Commons Attribution licence -- CC BY 4.0 -- and you must follow the (extremely minimal) conditions of that license.
  • Keeping all this in mind, please take this work and spread it wherever it suits you to do so!