The effects of token women on shortlists

Imagine a four-person shortlist that has three women and one man on it. With this shortlist, a woman will be hired only 67% of the time.

If you’ve got two women and two men on the shortlist, a woman will be hired 50% of the time – the odds you would expect if people were making hiring decisions purely based on merit.

What chance do you think a woman has of being hired when there’s one woman (against three men) on a four-candidate shortlist?

According to a recent study looking at academic hiring, there’s statistically no chance she’ll be hired.

Many employers are actively trying to recruit more women to senior positions, and are changing the composition of shortlists as a means of doing so. Some large corporates have recently announced that they’re scrapping all-male shortlists and are asking recruiters to find a more diverse range of candidates.

But as the study above suggests, adding just one woman to a shortlist to prevent it from being all-male may not do the trick. This is because the ratio is still sending the implicit message that a man is more appropriate for the job.

Or possible that people see through the inclusion of a token woman on a shortlist?

25 thoughts on “The effects of token women on shortlists”

  1. “According to a recent study looking at academic hiring”

    Ah well, there you go then. Academics. When did academics ever judge anything on its merits?

  2. The Unused Testicle

    We need a law to ensure more women are hired for all jobs. Then we need a law to ensure you can’t sue anyone.then we need a new law…

  3. No more academic hiring. Let’s see how the girlies do when it comes to mass firing uncompensated and sans pension.

    Nothing less than total equality will be acceptable.

  4. Academic hiring is totally biased. Go look at the typical job advert for a German university. Many of them proudly proclaim something along the lines of: “The University of Katzenelnbogen is an equal opportunities employer. Therefore equally-qualified women candidates will be preferred over men”.

    Don’t even get me started on the “who you know” culture.

  5. “The University of Katzenelnbogen is an equal opportunities employer. Therefore equally-qualified women candidates will be preferred over men”

    Was that using Google translate!

  6. If the token applicant had the same grasp of arithmetic as the writer of the extract, I can see why they didn’t get the job.
    3 of 4 = 67%. ROFL.

  7. Academic jobs are often stitched up in advance anyway, and the application process is only to provide a fig leaf of objectivity.

    It would be interesting to see whether there’s a difference between those jobs that go to the internal candidate (pre-selected) and those that go to an external candidate (where there is more chance of it having been a genuine selection process).

  8. Tim the Coder – I can see the maths too.
    67% is correct.

    You are replacing a male candidate with a female candidate – you aren’t picking based on statistics. You are picking on best person for the job.
    Is the 3rd best woman going to be better than the sole male candidate who made it through the process, the male that didn’t get bumped for a female?
    If ‘normal’ is 4 candidates, two male and two female, what percentage would you assign on a 3 female to one male group?

  9. Martin,

    That’s not what they were saying (replacing the second bloke etc!).

    Either they were being simple (and got the number wrong), or they are calling on some actual evidence (and in which case it’s interesting that it’s 67% and not 75%). I haven’t read it to find out which.

  10. i’m really disappointed in the BIT on this. I’m all for suggesting behaviour based tweaks to achieve a policy goal in lieu of legislation. but….

    But BIT have to be rigorous about the behavioural research because that’s what their specialism supposedly is.

    And this “guidance” has woo written all over it.
    They very much assume the demographic stats of the boadroom are down to hidden biases. But they don’t propose remedies to avoid the bias coming into play,(say anonymous applications) and even then its an unaddressed question whether these will account for the differentials they care about,
    Instead they propose nobbling the hiring process, and why some nobbles don’t work so well (actually not why they don’t work in a factual sense but why they postulate it doesn’t)

    So to the study this is based on then..Some Undergrads were asked to make hiring decisions on an experimental basis.
    No , i don’t think you can recommend anything on this study.
    For One, they use implicit bias tests which are very dodgy (i’m not impressed just because it has Harvard in its name) They use status quo preference of undergrad lab rats who don’t have the same incentives or consequences of the professional selectors they’re supposed to model. But above all its not peer reviewed. Jeesus… BIT are jumping straight from an un-peer reviewed “What i did in the summer with my undergrads” article to make recommendations to CEO selection committees. Get out of here BIT, go back and think what it means to your reputation to be recommending this stuff.

  11. I think what you’re all missing is the the percentages were derived from the study, not maths. Its unclear because in the 50/50 example the maths matches the real life outcomes, but otherwise, according to whatever was studied, if there’s 3 women and 1 bloke a woman gets hired two thirds of the time. Which is about right, given the extra woman will be a makeweight, and thus the real competition is between the two best qualified women and the one man.

  12. According to the article they had a study of N job applications with one woman on theshort-list and no-one ever chose the woman.
    That is unbelievable – unless, as Tim postulates, that the woman was only on the short-list because they had to have a woman on the short-list and not because she deserved it. So what happens if they have a woman who deserves to be on the short-list? Do they add the token woman as well? That would explain why in a 3 women to one man situation two out of three winners are women instead of 3 out of 4. But the two-each situation the choice is 50:50 which doesn’t fit.
    I suppose that there is zero chance of seeing the actual data that the claim to analyse.

  13. “I suppose that there is zero chance of seeing the actual data that the claim to analyse.”

    It seems so.

    The article links here, which says:

    When there is only one woman, she does not stand a chance of being hired, but that changes dramatically when there is more than one. Each added woman in the pool does not increase the probability of hiring a woman, however — the difference between having one and two women seems to be what matters.

    The bit highlighted suggests that the probability for each individual woman doesn’t change – so I suspect they did indeed get the maths wrong. The numerical coincidence of the percentages being 1/2 and 2/3 is just too much of a coincidence when the null should be 1/2 and 3/4. But yes, they don’t give enough data or detail to tell for certain. Highly dubious.

  14. @ NiV
    Yeah, I clicked on the link and could not find any data or link to anything with data – hence my comment.

  15. If the people making the shortlist go on to do the hiring then adding people to that list based on anything outside their original selection have no show.

    My, admittedly small, experience of hiring is that interviews only provide confirmation of the intended pick. (As it should be, picking a position on an irrelevant skill like how well you interview is daft.)

    So adding extra women is a waste of time. As would adding extra men, for that matter.

  16. “I suppose that there is zero chance of seeing the actual data that the claim to analyse.”

    Policy-based evidence, i.e. woo, i.e. political bollocks. Deride and ignore.

  17. Each added woman in the pool does not increase the probability of hiring a woman

    Surely they’ve disproved this in the very next sentence?
    1 women in pool – 0% probability of woman being selected
    3 women in pool – 67% probability of woman being selected

    Surely 3 women is additional women compared to 1 woman, and surely 67% is an increased probability than 0%.

  18. “Surely they’ve disproved this in the very next sentence?”

    It depends what they mean by “a” woman; ‘each’, or ‘any’.

    Suppose you say that the probability of hiring each specific woman in a pool of 4 is 1/4. Then the probabilities of hiring *any* woman from a pool with 0, 1, 2, 3, 4 women in it ought to be 0/4, 1/4, 2/4, 3/4, 4/4. For the cases examined, the probabilities would be 1/4, 2/4, 3/4. Getting 0, 2/4, 3/4 would suggest a “token woman” effect only, with no sexist bias and no change in probability for each additional individual woman as soon as there is more than one. Getting 0, 2/4, 2/3 suggests the probability changes every time.

    I’m guessing. But it’s the only plausible interpretation I can see.

  19. “” Many employers are actively trying to recruit more women to senior positions,” Why?”

    Well, one logical reason is that if you can increase the pool of supply, the price goes down.

  20. @ NiV
    Yes, but WHY are the 2:2 choices not affected by the token woman distortion? I can postulate that if there is one good woman but not a second they stick a junk male comparable to the token woman so that there is just one good applicant of each sex but seems more *intelligently* Macchiavellian than I expect of HR.
    One might then assume that if there are two good women they alternate between two good women against two good men and two good women plus the token woman against one good man.
    No, actually, that latter is assuming that they care about the result – this is from US Academia so PC results matter more than hiring someone who will do the job properly, so it must be that they only include more than one good guy if there is no woman worth hiring.

  21. “Yes, but WHY are the 2:2 choices not affected by the token woman distortion?”

    I have no idea. My impression is that the study itself is trash, and Tim only brought it up to give us another opportunity to have a go at feminists wittering on about sexism. But I guess it passes the time.

Leave a Reply

Your email address will not be published. Required fields are marked *