Nice one, Deloitte
Posted: October 12, 2015 Filed under: Education | Tags: advocacy, community, education, educational research, equality, equity, higher education, resources, thinking 3 CommentsGiven how much research there is to tell us that intrinsic bias is having a terrifying effect on hiring decisions and opportunity, it’s great to have some good news to report when a company decides to address the issue.
Deloitte U.K are going to start using a school-blind approach in its hiring process (WaPo link), which will potentially benefit thousands of school- and college-leavers who would have been perceived to come from “less suitable” educational pedigrees. They are also planning to work out how students performed relative to their peers in a given school, rather than against broad district or national levels. Deloitte are now part of a growing group of companies who have moved beyond “Am I getting the best people” to understanding that their biassed perception of best may be preventing them from seeing some candidates at all. They could be picking their “best” from a much broader range of better, if only they’d address those biases.
This is certainly not going to end bias overnight (there are so many other issues to address) but it’s a great start.
Nick Falkner 
I have been watching the bias first hand. Not contrary to the research you linked ( because this other one is for a different type of position ), a new PNAS research shows that the bias is completely the other way around when hiring for academic and tenure track positions:
http://www.pnas.org/content/early/2015/04/08/1418878112.abstract
So even though there is a high bias in STEM employment against females in industry, there seems to be a 2:1 bias in favor of females for academic positions.
In the PNAS research you linked, the job is a “lab manager” position. I’d guess that the bias depends on the job, more than it depends on the employer. I wonder if this is a results of academic positions being seen as “female jobs” or the academia being more aware of the bias against females and actively trying to fix it.
Which one do you think is the reason?
LikeLike
The Williams paper is very interesting, contrasting as it does with the ‘technical’ orientation of the previous one.
I think the reason that there is such a difference is that it is very hard to run an experiment like this without people realising that it’s not real. Yes, people are now aware of bias and many are seeking to change it, but I am concerned that what people report that they would do and what they actually would do may not match up.
The whole point of unconscious bias is that you are not aware of it. Of the 68% of people who chose not to respond to Williams and Ceci’s survey request, some of them would have been totally aware of their bias and thus not responded. However, when you confront an unconsciously biassed person and ask them if they are biassed against “x”, they will tell you ‘no’ (and be able to explain it) while still carrying out actions biassed against “x”.
Basically, no, some of your best friends aren’t “x”, and this is why it’s what people do in real situations that’s important.
It is not so much that I think Williams and Ceci did a bad job (I think it’s a pretty amazing study) but I’m not sure that we will ever be able to answer this difficult question outside of empirical evidence collected from real hiring decisions over a very long period. And we have that. And it’s called “not very many women in STEM”.
Let’s start by noting that the 2:1 is a little bit misleading because it makes it sound that we should be brimming with women. It’s important to remember that the number of women in the application pool is almost always smaller than the number of men in this area. This commentary
http://sciencecareers.sciencemag.org/career_magazine/previous_issues/articles/2015_04_29/caredit.a1500108
(which goes to great lengths to NOT undermine the paper and not hostile) notes that, for UCD over the last 5 years, ” For Math and Physical Sciences, for example, women were 15.2% of the applicant pool, 31.4% of the short list invited for an interview, and 26.7% of the faculty actually hired.” So, yes, women making it out of the pool in this case are coming out at the predicted rate from the paper but they are still making up a small number of the pool and only a quarter of hires. And that is in a generally progressive state like California.
We are also, again, faced by the rather blunt realities of faculties such as my own, where the representation of women is incredibly low. If Williams and Ceci have the right of it, why are we not seeing this everywhere? Why are women not doubling in number? Why are women less than 25% of my academic peers in my discipline at my Uni? (And, let us note, that this is a vast improvement on what it was ten years ago.)
I do have a concern with this study that it was framed in a way that did not quite hit the “gold standard” of the original PNAS paper I mentioned for two reasons:
1. The first is, even with the extremely rigorous and laudable attempts that the authors made to produce an authentic task, it was still very hard to disguise the hypothetical nature of the task and that this may lead to artificially high ‘expected’ answers rather than what would happen. We don’t hire scientists based on narratives where I work and I’m not sure how widespread it is elsewhere.
In the original paper I cited, this statement is key “Of importance, participants believed they were evaluating a real student who would subsequently receive the faculty participants’ ratings as feedback to help their career development.” This is why I hold that study in such high regard – because people thought that their actions mattered, it was more than asking what they thought they would do.
2. The second is the validation experiments. The first was conducted in a school of Psychology, which is not really that good a fit for the STEM environment. In 2005, 72% of PhDs and PhD entering the workforce in psychology were women (according to the APA) and gender is a very central theme in Psychology. The second validation experiment, full CV with Engineering faculty, is precisely the right approach to address my concerns for experiment 1 BUT, again, as far as I can tell, there was almost way to hide the hypothetical nature of such a ranking exercise because of the nature of an academic CV. I agree with the authors’ conclusions that the CV and the narrative were equivalent, but it doesn’t address my core concern.
Ultimately, the best validation of something like this is to present the data back to the institution and say “Ok, is this how you actually hired?” Now we know that UCD matched this in at least one area but that’s not enough for a study on such a scale, where the authors have done so much to make it sweepingly large n for just such reasons of validity. And I think this lack of tying it back to “real decisions” weakens it significantly against the original paper.
(If I have missed a validation against actual hiring decisions, then I heartily apologise and will retract. Let me know, please!)
We know that there is a terribly leaky pipeline for women into STEM and inside the career, especially for women who are primary carers or wished to combine family and a career. Williams and Ceci have excellent work in this paper, which identifies bias against marital and child status rather than straight gender. Single, childless women are preferred by both men and women – and this is a telling note. But you’ve read the paper so you know that there’s a lot of great stuff in here as well as the things I’m not so sure about.
You know what? I actually really hope that, despite my concerns, this paper is totally predictive of real-world behaviour and we ARE hiring twice as many women because it will reflect that everything we’ve been saying about bias is finally being used to fight it. But I think there’s still a lot of difference between what people say and what people do and I am fearful that the design used here was one where it was possible for people to realise what was going on and give the right answer.
But I really, really hope I’m wrong.
But, when it gets down to it, until qualified women are making up roughly 50% of pool candidates and then hires as well, we’ve still got a long way to go.
LikeLike
And thank you very much for the comment!
LikeLike