Did blind orchestra auditions benefit women?
statmodeling.stat.columbia.edu
statmodeling.stat.columbia.edu
One such comment: "Table 4 presents the first results comparing success in blind auditions vs non-blind auditions. . . . this table unambigiously shows that men are doing comparatively better in blind auditions than in non-blind auditions. The exact opposite of what is claimed."
I had a look at the linked paper. And it is true that the table shows that women perform worse in blind auditions. However, the paper does not claim that this table shows that women perform better. Instead the paper elaborates on the options of why women perform worse by arguing the following:
"One interpretation of this result is that the adoption of the screen lowered the average quality of female auditionees in the blind auditions. Only if we can hold quality constant can we identify the true impact of the screen."
The paper goes on to explain how they discovered that during blind auditions there were a lot more under-qualified women and that this was skewing the data. They discovered this because they had the names of all the participants and saw that some women would participate in both blind and non-blind auditions while others would only participate in blind ones.
The following was the papers conclusion on this matter: "When we limit the sample to those who auditioned both with and without a screen, the success rate for women competing in blind auditions is almost always higher than in those that were not blind."
Worse, the authors acknowledge that blind auditions were part of a larger effort to recruit more women, but fail to account for other aspects of that effort. They simply attribute the entire effect to the blind auditions. But it seems likely that orchestras that were actively trying to recruit more women (like the NYPhil, which the paper claims increased the percentage of women in the orchestra from least in the nation to most) would have hired more women with or without blind auditions.
Original paper: https://www.nber.org/papers/w5903.pdf
The entire point is that there is no objective measure of playing ability, which allows conscious and subconscious biases to affect audition decisions, which is why blind auditions make sense for choosing orchestra performers, but is not required (let alone practical) for choosing sprinters.
We can easily measure the eventual decision of the audition, but the audition process itself is much more difficult to scrutinize, unlike your example of a sprint race. That's why your original challenge to compare this audition situation to a sprint race does not apply.
There's something of a repeating theme lately. Study of blind interviews/tests in various fields => women found to underperform despite experimental controls and expectations => commentary and authors grasp at straws to explain the result with anything but the fact that perhaps men and women don't have the same performance envelope in every walk of life.
How many more of such studies and straw-grasping explanations will we have to go through before someone considers occam's razor?
> perhaps men and women don't have the same performance envelope in every walk of life
I believe what they said is right. At the same time, I also believe that the worst performing men perform worse than the worst performing women - males have a wider spread of ability.
I don't follow. If GP's comment had ended with the bit you quoted then it might just be a straw-grasping explanation, but the rest of the comment says the study authors tested whether the explanation was true, and found that it was. That is, they suspected their result was due to sampling bias, found a bias in the data, and found that removing the bias reversed the result. If that's an accurate description of what happened, it doesn't sound much like grasping at straws.
Failing to reject the null hypothesis but then saying 'if <alternative hypothesis picked with knowledge of result> is true then <result> would make sense' is not science. This claim is useless non-information after the results are known. This is a completely new claim that must be validated with a new study designed around it. Maybe this alternative hypothesis is true, maybe it isn't. Hindsight of the original experiment's data can never shed light on that.
Trying to pass off a revisionist hypothesis with advance knowledge of results, is very much grasping at straws. Reduced to the absurd, its as if someone predicted a coin flip would land on heads, and then revised their reasoning after seeing it land on tails and told you that their method of predicting coin flips was right all along except for that little <sampling bias/whatever>.
I hope this is clear and makes sense, because this is one of the fundamental pillars of the scientific method. You state a hypothesis, then you test, then you arrive at a conclusion. You don't start with a conclusion and then revise your hypothesis to match it after knowing the outcome of the test.
That coin-flip prediction example seems incomplete to me. Here's how I would extend it to relate to the topic at hand: you ask a researcher to predict the distribution of heads and tails for a specific coin, which is secretly a trick coin with heads on both sides. The scientist trains a prediction algorithm on thousands of coin-flips performed on many randomly selected coins in the wild. Upon testing your coin, they find that 100% of the outcomes are heads, statistically very different from the training data (as confirmed by the appropriate statistical methods). You seem to advocate an uncritical reporting of this result, with no allowance for discussing possible explanations beyond the strict study flow of "predict 50% ± X% heads, flip coin N times, observe 100% heads, report result, end of story". A responsible scientist would discuss possible explanations for the anomalous result, including biased instrumentation, biased training data, biased sampling, uncontrolled variables, and so on.
Uncritically collecting data, grinding through statistical tests, and reporting p-values/effect sizes/confidence intervals is not science. Researchers are fully expected to interpret their results in the greater context of their field, while of course maintaining statistical rigour, which is necessary but not sufficient to do good science.
I think you're reading too much into the "first they found X but then they found Y" narrative in the top comment of this thread. There's nothing like that in the actual paper - the authors start with one hypothesis, gather the data, do a bunch of regressions and whatnot, and report all the results. Then their overall conclusion is that the balance of the evidence supports their hypothesis, but they note all the caveats. There's no revisionism or changing of hypotheses or whatever, that I can see.
In theory, such a person should be better suited to such evaluation too: it's not uncommon to have psychology majors in HR departments, for instance.
And finally, once you find someone is detrimenal to the team (the orchestra), you can still fire them (I know it might be harder).
The more I do interviews and get assigned the "culture & fit" asessment role, the more depressed I get. It's really just code for the bias "looks and acts like me => good"
Certainly a segment of the entertainment industry earns money by catering to the worst aspects of human nature, but that's not the whole industry.
Also: Give me the best orchestra. I don’t go to the concert to gaze at a bunch of beuatiful penguins and ladies in dresses. I go there to listen to music. If they are good I don’t give a damn about their looks.
An orchestra must give the audience what the audience expects. Is just an expensive entertainment act.
It not just entertainment for a lot of people, unless you consider all cultural events "expensive entertainment acts".
Of course the justification fits to the sex differences as well. You should hire the gender that makes music performance more enjoyable for your audience.
In fact, maybe the goal should be to find how much gender bias is customer preference try to achieve that.
https://www.youtube.com/watch?v=6PyYcnXQZJY
Not exactly a classical orchestra, but it serves well to illustrate the point.
Jazz is more amicable to diversity than classical music on a fundamental level, because it's based on improvisation and unique personal expression. If you're physically unable to play the same way as everyone else, you're pretty much useless to a classical orchestra but you may well become a pioneer in jazz.
Some obstacles to diversity can't be overcome with blind auditions and quotas and sensitivity training. Some environments are lacking in diversity because they are structurally incompatible with diversity; they can't have diversity retrofitted on to them, but need to be reformed from the ground up.
Why should being based on improvisation and unique personal expression be the be-all-and-end-all of analysis? Not everybody wants to play / hear improvised music.
Classical music is dying, because the audience is literally dying of old age. Jazz or folk can sustain itself as a niche pursuit, because a small ensemble can profitably play to a small audience; a symphony orchestra has an enormous wage bill and has to fill a large hall week in and week out.
Classical music can renew itself and guarantee a sustained future, but it can't do it without fundamental structural reform. If classical musicians want a secure future for their art form, they need to seriously reflect on what classical music is, what it could be and what it needs to be. It's not enough to offer cheap tickets to under-30s, it's not enough to have the odd pops concert in the park - lip service will not abate an existential threat.
https://www.artsjournal.com/sandow/2011/03/age_of_the_audien...
It seems to me that tenure was meant to solve this, but it doesn’t. Academics are groomed and selected to be career climbers, willing to sacrifice anything to please the gods of their establishment. 1000 years later, still a monastery.
By far the better approach than removing information by having blind / limited information applications is to have MORE information, e.g. https://marginalrevolution.com/marginalrevolution/2019/02/re...
> We conduct four randomized field experiments among 1,801 hosts on Airbnb by creating fictitious guest accounts and sending accommodation requests to them. We find that requests from guests with African American-sounding names are 19.2 percentage points less likely to be accepted than those with white-sounding names. However, a positive review posted on a guest’s page significantly reduces discrimination: When guest accounts receive a positive review, the acceptance rates of guest accounts with white-sounding and African American-sounding names are statistically indistinguishable.
That is: you want your team to succeed, but you'll accept a little less productivity if the team is fun, or there's less drama, or you feel happy being with your co-workers.
>considering diversity needs //
What are these needs?
I've struggled with this issue recently - in my kids school all the teaching staff are female. All the appointed governors are female too. So the people choosing the staff are females used to working in a female-only environment. There's no [visible or reported] effort to bias towards employing any men. I think not having any men in teaching roles, only the janitor is a man, is a bad thing for all the kids in the school. But, I don't really agree with "positive discrimination" as it's just discrimination and a woman who fails to get the job solely because of her sex would, IMO, be being unfairly discriminated against.
My sense is that the unfair discrimination is perhaps necessary, but it sits very uneasily with me.
To my mind this is the same sort of consideration, choosing a worse candidate because on a broader perspective, considering characteristics of the system rather than only the character, there's a benefit.
tl;dr systemic optimisation rather than local optimisation.
The systematic interpretation of this law is that it is legal to moderately discriminate in the favor of women when they apply for a position (school, work) that has a male majority, while it is illegal to discriminate in the favor of men when they apply for a position that has a female majority.
Examples of male-majority situations where this is applied is CEO or board positions in publicly-owned companies or engineering school, while examples of female-majority situations where this cannot be applied is veterinarian or medical school.
I know this from film sets: I happend to be in crews, that were so diverse, that the only overlap in the end was that all of us somehow liked film. These were incredibly interesting and motivating crews, with a ton of creative ideas, willingness to go the extra length etc.
The worst crews I saw, were the ones where you had departments that had creepily aligned departments, that defined themselves by drawing a border to other departments. E.g. a light crew consisting completely out of white “dudes” all with the same ideas, jokes and taste, similar age etc. They did good work, but once they outnumbered the rest, the open discourse/exchange of ideas on set suffered.
But there are other ways to approach, and possibly fix, this problem. One would be increasing the reach, e.g. you could investigate if your job ads reach mostly women and instead advertise on some male-heavy media as well. Another would be fixing the underlying biases and misconceptions of the job (e.g. that teaching is a female job or that men that want to spend time with children are likely pedophiles) - though that one is obviously harder to do, you need society-wide influence. Or maybe you'd figure out that it's just the way it is, and there is a huge gender (or some other characteristic) imbalance in the whole pipeline (i.e. not just the applicants for the job, but already when people are applying to universities for the appropriate field of study) (in which case positive discrimination would mean you'd have to lower the bar to an unacceptable degree), which might be OK as well.
It's also odd that talk about the value of different perspectives and diversity can disappear as soon as some women create women-only workspaces for themselves.
It's examples like this that make me unfriendly to feminism. Equality of opportunity is a worthy goal, but either you support genuine diversity or you don't.
From what I understand, the suggestion is that if one person is obviously more qualified than another, you pick the most qualified person regardless of other factors, but if there are multiple people where it's unclear who is better, you pick the person who's from a marginalized group.
If we had some kind of ultra-precise scientific way to measure a person's level of qualification, the point would be moot; it would be highly unlikely that two people would measure exactly the same, so we would never be in a situation where we choose based on other factors than qualifications. It's precisely because it's impossible to determine exactly how qualified someone is (probably by the very nature of what we mean by "qualified") that the idea has any merit at all.
Presumably the best person for the job would be the one that maximizes the company's profit. Suppose it's easier to hire the female candidate that's better than all male candidates (hard as that might be to imagine for many on this site) if you currently have a woman working on the team. The existing team of eight high-performing men just isn't as appealing as a team of seven high-performing men and one not quite as high-performing woman.
I have heard from multiple women, that they will absolutely not join a team if they are the only woman on it.
You would miss out on a lot of talent that way.
Diverse teams gather more information which leads to better insight of edge cases, and ultimately to better products.
So in this case I do not see at all that software from "diverse" teams has gotten better.
I fact I think that overall software has gotten worse in the past decade. There is no structure, quality or documentation. The only thing that matters is that "something visible happens" so it resembles a product.
If men and women are fundamentally different then there are good reasons to choose men or women only in a given situation (along normal stereotypes: women are better caregivers, men are better manual labourers).
Recruiting in this way is illegal in a lot of places.
>they're all going to ask the same questions if they all have the same perspectives, biases, etc. //
Then shoudln't you choose people based on their ability to consider the specific perspectives you need?
One of the things about people is the ability to adopt the perspective of others readily.
Not sure about what "modern feminism" is about, but the goal is fairness, which means treating women as women and not requiring male characteristics. Basically, detaching personality traits more common in one population (eg gender) from actual job requirements.
Different genders/races/cultural backgrounds/educational backgrounds are connected to having different experiences in life. Of course you can imagine what e.g. sexism looks like, but when it happens to you every other day as a woman you will have it much more present than a man. Same thing with other aspects, people who have different racial/educational/cultural backgrounds bring different experiences with them and they might offer them precisely at the point where you didn’t think about it, not because you are a bad person, but because emulating all other perspectives your self is exhausting.
The variance between individuals of both genders and within races/nationalities/etc is big enough for you to look on the individual level. When you say “women are statistically on average better caregivers” and hire only women, not only are you missing out on very talented men, but the functionality of your team is reduced as well. E.g. because in some situation a male caregiver might be a better fit for a given client. Not because they are better or worse at their job, but because beeing male brings a whole set of experiences, cultural expectations etc with itself. The same apparently goes for other traits your employee might have.
Most jobs require teams to react to new and unpredictable challanges and having a broad range of backgrounds, helps them to always have a team member whi knows how to deal with it, because of their background ans life experience.
When you hire a team where you have a monoculture, it is hit or miss. It might work well for a certain job at a certain time, but if anything changes, your team might be less resilent, flexible and able to solve problems compared to a more diverse team.
Apprently there is, otherwise why claim that any gender imbalance in the workplace must be the result of discrimination? I agree with you that diverse points of view can be useful; but diverse points of view imply (average) psychological and cultural differences, diverse attitudes and interests, and these might well be the reason of the gender imbalance. To use a steereotype (don't kill me for this) if female software developers can bring value to a team because on average they have a better eye for UX or design, then it is also probable that more females will be in UX or design than males, creating a gender imbalance in the number of candidates for software development positions.
I don't think Feminists claim any gender imbalance in the workplace must be the result of discrimination. At least the ones I know usually hold much more balanced views here and the subset of them that is for a fixed quota of women in certain fields/levels usually also say, that they really don't think of this as a good solution, but as they unsucessfully atempted to change things using voluntary solutions for decades, they are now in favour of such solutions.
I am a European, so our feminist debate might differ from the one in the US. But you shouldn't listen to the ones screaming the loudest anywhere and think they are representative. If we'd judge white christians based on that it would mean the world would be filled with lunatics, when in fact it is much more nuanced.
> [...] then it is also probable that more females will be in UX or design than males [...]
I think nobody rational would argue agaisnt your thesis. The point feminists usually tend to make about this is, these bilogical tendencies are the only one thing that decides careers – there is also the question of work culture and sociatal expectations, etc.
If you do a little bit of research you will be fascinated how many women helped in the dawn of computation. This was, because before electronic computers, the word "computer" literally meant humans who did the calculations and these humans were predominantly female maths graduates in the dawn of the computer age. The job of a female computer was not well paid, only a fraction of what the male deciding mathematicians earned. But there weren't enough skilled people here, so women filled the position.
When computers became bigger, it started to become more male and the focus shifted from maths specialists to special degrees, the quality of software declined with the quality of education, more men started to work in the fields salaries increased, until at some point it was a male dominated field, with slight shifts in the recent years.
It is conviniently easy to say women are better at the thing they do at a specific point in history (which for some reason is always paid worse) and so they should just stick to it. E.g. today many women work in textiles and fashion, this was the polar opposite 150 years ago, yet today every unreflected person would swear that there is something inherently feminine about working in the textile industry. So IMO biological factors make a difference, but it has significantly more impact whether you live in a society that expects certain genders to fill certain roles. There will always be men or women that go against these roles, but wouldn't it be the best to live in a society where men are not expected to do/be/embody X and women not expected to do/be/embody Y? This would mean everybody could be closer to their true selves, and we end up with the best people for a given job.
In a practical setting the sensible and civilized stance to take on this is not to focus too much on gender and other features when deciding if somebody has the skill, while focusing on it when it is about the "systemic" question, whether somebody should be part of a team or not, with the goal of including people with different backgrounds.
This is just an assertion currently, and I am not inclined to believe it is true. I would believe it for more "people" and/or culture oriented problems, such as catering, but for the majority of software development problems, diversity in people is nothing compared to diversity of software development experience. Which, ironically if we take as given most developers are white males, we maximise on average by only hiring white males!
The idea that this kind of a priori judgement is not only acceptable but laudible has come as a bit of a surprise.
Not least because it would seem to justify rejecting female applicants for a role where a 'male perspective' is considered important.
All that time I've wasted writing code...
Following from that, it may be in your interest as a hiring person to bootstrap some diversity to remove that "unhealthy" idea early on.
One advantage I could imagine is mixed teams making the company more attractive for men to work at, making it easier to attract talented developers.
Run a company that women see as a good place to work, where they won't be harassed and where they're treated fairly. Once the company gets a reputation for being a good place for women to work you'll get a lot more applications from women, and the problem of balance will resolve itself because you can hire on merit if you have a good selection of candidates.
The hard part will be firing people who do things that give the company a bad reputation even if they're great workers.
Just imagine a 20 person company with 5 women who are being fairly treated: how do they get better gender balance? Even if these five persons were outspoken, how many people could they reach? Not to mention that a true character of a company is only revealed in a crisis (harrasment, known unequal pay, role progression...). This company will never get the breadth of candidates Facebook might to be able to rely on statistics. Without actively pursuing that as a goal, balance never happens.
Positive gender discrimination is a societal tool, a tool to fix unfair social treatment of genders at an early age ("here are pink dolls for you dear, your brother can go play games on the computer" just as a blunt example, but it's a lot more subtle in general).
But this isn’t soley about race, nationality or gender, you have to find a good mix also in terms of personalities, skillsets etc. I always found system theory to be a useful abstraction when thinking about this. System theory asks the question which function a member of the system (the team) will take, given their personality etc. You cannot decide this from top down, it will just happen. Like in a family, where you can visit after 10 years and instantly, without beeing aware of it you put yourself into a certain position.
The question is: which function does a team/family member fulfill for the system?
A systemic view means the answer to this question depends on the constellation of the whole team and not just on individual members.
So my answer would be: make a team where everybody gets to align their personality with the function they will have within that social construct, make everybody a valuable part to the whole and make everybody realize, they are better off as a whole.
A good experienced team is like magic just because everybody precisely knows what their function within it is, and having a different background is part of these functions as well.
To my mind, this is the key point. It may be likely that a team with e.g. diverse skills/personalities will end up being gender diverse. But to assume that it has to be gender diverse, or that you can achieve a diversely skilled team by enforcing gender diversity seems to lapse into lazy gender stereotyping about how e.g. women's skills/perspectives are limited by their gender.
And by background I surely mean gender, but also things like educational background, nationality, personality etc.
Nobody should brute force for gender diversity just for the symbolic victory. Diversity done right has its benefits, but like all paradigms you have to use your brains when implementing it. Refusing to do so and then after a half assed attempt declaring it doesn’t work, is what you sadly witness quite often.
I know multiple women in tech and every single one of them had their gender used against them many times during their careers. Hiring men and women like in regular fields won’t hurt tech at all. I don’t really get why it has to be a big thing at all.
If you want someone who can contribute to a team in a certain way, then hire for that.