Some people might criticise these numbers, particularly the 70% male ratio. However, to really determine whether a company is being as diverse and open to equality as possible, the same measurements should be applied to the applicants as well as the employees.
If seven out of ten applicants are male then the 70% male ratio doesn't indicate a bias. It might mean that the company isn't as attractive to female applicants, but then again, neither is the discipline behind many of their products (at the moment.)
Well, targeted outreach for applications followed by a blind review process has been shown to be successful in increasing the number of women speakers at conferences. John Oliver was recently in the news for using that approach when looking for writers for his new show (but he has only nine writers, so it’s hard to draw any conclusions from the outcome in his case).
Please don’t pretend that companies can do nothing about who sends the applications. I don’t know what Apple does, but I do know that many companies are doing some kind of outreach already (going to conferences, publishing job adverts in many places, maybe even specifically calling people and asking them whether they would want to apply), so optimizing that with diversity goals in mind would not be something very weird.
Your portrayal of the situation is extremely simplistic and misleading.
You are right; a bare 70% number does not say anything without further analysis. It is hard to separate the effects of lower attractiveness of their jobs to women, the acceptance of women in their workplace, and women's perception of that acceptance.
And of course (and as you know) the reverse is not true, either. The fraction of non-Caucasian members in the KKK probably reflects the fraction of non-Caucasian applicants they get, but that does not imply that hey do not have a bias.
The issue with this line of reasoning is that it avoids complex notions of cyclical reinforcement, as well as lower level, and likely systemic, bias.
That is, there are potential ideological underpinnings that lead to the rather by-the-book statistical analysis.
When Lombroso attempted his measurements of criminality, and leveraged what appeared to be an exhaustive bit of data against the subject, inevitable analysis of the data ensued.
What was missed however, were the vastly more complex subjects of “What is criminality” and “Who defines criminality” for example. In a way, the data became a cyclical reinforcement of the (false) hypothesis, while ignoring the entire sociological dimension.
Very few have stopped to question the metrics and models, and instead only see the data.
In this case, it may be prudent to not stop at the 70% data, and dig beneath it to evaluate the ideology and biases that _may_ have played a role in the symptom. It is seductive to lean on Western Rationalism / Scientism, but we conceivably could do well to question the systems that give rise to the models that yield the data.
When Apple came to my university to run an engineering recruitment event, they flew in 12 employees to speak with us, zero of which were female. When asked by the audience about the presence of female engineers at Apple, they sputtered some non-sensical reply. If not enough women are applying for engineering jobs at Apple, then at least one of the reasons is that the company is not making sufficient effort to try and recruit them.
>'However, to really determine whether a company is being as diverse and open to equality as possible, the same measurements should be applied to the applicants as well as the employees.'
That impresses me as the sort of information no company would want to release in total. There's just too much room for criticism.
Please, stop repeating this tiresome excuse. Absolving a company and blaming the pipeline does not help actually solve the problem.
1) There is a "filter" problem, eg bias in interviewing. This is what you are talking about.
2) There is the "pipeline" problem, eg universities and recruiting practices.
3) There is the self-selection problem, eg women / minorities going into other fields because they don't want to deal with the horseshit.
...plus many others, including how the few women are treated after they get the job.
The only cure is sunlight. You optimize what you measure. If we can get tech companies to publish these numbers every year, then maybe they'll start doing something about it.
blaming the pipeline does not help actually solve the problem.
Sure, but if the hiring process is unbiased, the problem exists elsewhere, so the hiring process cannot be "fixed".
The only thing that could be done at the hiring point is "overcompensate" by hiring many inferior/poorly qualified applicants simply because they are female.
Certainly companies hire "token" minorities purely for PR purposes, but generally speaking it seems unrealistic to expect a large company to routinely pick inferior candidates for the bulk of its workforce.
I really can't fathom why you were voted down on this.
Regardless of where the failure is there is clearly something about the software engineering industry and the pipeline leading to it that is biased against minorities and women, which implies we are failing to optimally place the most talented (or potentially talented) people in the right positions assuming talent if evenly distributed among various ethnicities and sexes). Which means our economy is not as productive as it could be and we're missing out on the increased opportunity for possible break throughs in software development and other related fields with similar bias.
It's clearly a problem. It may not be directly the fault of say Apple or Google but it is naive to see numbers like that and not be concerned.
"Regardless of where the failure is there is clearly something about the software engineering industry and the pipeline leading to it that is biased against minorities"
How can you say that when East and South Asians are so overrepresented?
2. Even if it were true, a lack of bias against a few minority groups would not invalidate the statement that there is bias against other minority groups.
3. You are conflating a very wide and disparate set of ethnic groups in a way that is very misleading (and mathematically similar to Gerrymandering[0]). If we were to break those very broad categories further, we would see that many groups that qualify as "East and South Asian" are still themselves under-represented.
1. How so?
2. True, but the comment I responded to made a different claim.
3. is irrelevant. There are always under-represented groups depending on how you look at the data, but if you claim general bias against minorities when some minorities are doing very well, you have some explaining to do.
No, your the problem. Hysterical social-justice-warrior making problems where there is none. There was a study done not long ago that shows the number of people taking computer science was predominately white and male. And the fact is any company that does not have a workforce along those numbers (pinterest for example) should be considered as having suspect hiring practises.
But hey it's only bigotry against "white" "males" and it hip to be ok with that nowadays, right HN?
Why did you quote "white" and "males"? Is it because you suspect many of them to secretly be black women wearing whiteface?
Anyway, there is a reason that white males are not allowed to talk about bigotry against themselves, and it's not about some conspiratorial hipsterism to sabotage you. It's because if you are a white man born in America complaining about discrimination pretty much everyone else in the world will you see you like this:
I'm not American, I don't live in America, and I can to a degree I can understand all this retardation coming from a country with such a rich history of overt racism. But goddamn the internet != america and these international tech companies having to pander to you delicate tulips is getting old.
Hell, why not go after Samsung for hiring too many Koreans.
It's often companies with retail outlets that are willing to provide this data - I suspect a lot of the diversity comes from the people that work at retail Apple stores.
In a way, it is similar to how the government operates - the diversity of people who call the shots is very different than the diversity of people in DMV or SSA offices, but overall it gives the impression that things are diverse.
If their "tech" and "non-tech" categories are what they sound like, then you're right. Their "tech" category is still 80% male, and 54% + 23% white plus Asian. For such a hip company, they look just as bad as their competitors.
Serious Question(s): How does this make them look bad? Is it unfair to think that this happens to be because of the demographic of people interested and qualified in "tech" positions? Or should we genuinely think these numbers are lopsided for some nefarious reasons?
Might be a part of it at least, but what percentage of employees are retail and what are office? Purely anecdotal, but the Apple store here in Charleston, SC is pretty damned white.
That depends. The data segregates into "tech" and "non-tech", but doesn't seem to define these terms well. If tech is taken to mean mostly programmers and engineers then you would expect these to be clustered around Cupertino, in which case (based on pure demographics) you would expect more hispanics and less whites. If you are including people like geniuses at retail stores then perhaps it would break down differently depending on how these stores are distributed around the US.
That's true, but a 17% difference from expected seems quite large. And if we're talking about people working at Cupertino, the pool of tech workers is significantly more white and asian than the average U.S. population, and I would think most of the Apple employees at Cupertino are pulled from that population.
Apple does have a substantially higher percentage of asians than would be expected from the entire U.S. population (15% vs 5%), probably due to the relatively large number of asian tech workers, but I would think the same effect would lead to a larger-than-expected percentage of white employees.
What is the meaning of "white" in that pie chart? Does a pale-skinned person from Jordan or Lebanon count as white? If not, then what category? "Other"?
What is "hispanic"? Is a white person of Spanish descent "hispanic" or "white"? Wikipedia says that this is "is an ethnonym that denotes a relationship to Spain".
So a Greek or Turk would count as "white", but a Spaniard as "hispanic"? Pop quiz: Ricardo Montalbán: white or hispanic?
The pie chart seems to be mixing "ethnonyms" with skin colors.
Then there is "Asian"; what exactly is that, and does it lump together the people of Indians and of Korea?
About "white": suppose that 70% of applicants are white, but they are largely Slavic, from central and eastern Europe; and suppose that the 70% of the employed whites are Germanic whites. Nope, no discrimination there!
Maybe the "white" category is so large because it includes many ethnic groups: someone from Iceland could qualify, as well as someone from Israel.
Since the Arabic and East Indian people are "caucasoid", any pale-skinned people from those regions can represent themselves as white. A person with albinism of Indian descent is very difficult to distinguish from a European, if at all. (Do a Google image search for "indian albino").
If I was cynical I might consider this to be a feature, simply release another set of statistics a year later with the demographics sliced a little differently and declare victory.
This is exactly part of the seductive nature of “data” in EuroWestern culture.
Knowledge as power, but not in the rather trivial sense of education being empowering, but in the crafting of knowledge. As cited below, well explored by Mr. Foucault.
I think I get it but in some countries age is not a privacy issue, business-wise. Actually, in some places (like Brazil) people expect you to disclose it, eventually. There's not much privacy protection about it like there is in US.
The weirdest is other countries where it'll be on the CV. Along with gender, ethnicity, religion, marital status, number of children(including their ages and names sometime) and a photo.
Its really uncomfortable since we're not supposed to use any of that.
Should you withhold information from your CV since some of it is known by you to not being supposed to be used by the reader of said CV?
Because CV without all what you listed is no longer a CV and is closer to a catalog listing. Should you also replace your name with some alphanumeric code? Because name can tell you a lot to discriminate on.
No not necessarily, though I wouldn't use a CV to apply to a job in the States or that is with a primarily US based organization. I just ignore the info when it comes up, it just catches me by surprise sometime.
I've actual lobbied to anonymize all applications as much as possible. Thankfully for some positions I have to review resumes for I'm unfamiliar with the cultural norms of what names are female and what are male. So in that way they are anonymized!
As with age, people are not expected to state their race in an interview, nor are employers allowed to discriminate based on race, yet this diversity report includes anonymized racial data. They could have easily included anonymized age data.
People need to learn the difference between descriptive and prescriptive ideas. Observating that Apple isn't diverse in a certain way is not the same as saying Apple is the problem. In fact I don't think most people believe that Apple's hiring practices, or any tech company's in particular, is the real issue for any metric.
The US has bizarre classifications that make no sense anywhere else. Why are people from Spain Hispanic but not from Portugal, why are Mexicans Hispanic even if they are descended from Aztec people. The idea that all Asians are the same is equally insane. Even White is a stupid concept that has no meaning. If I came from Spain I'm hispanic but if I came from just over the border in France I'm white. If I come from South Africa and move to the US I'm either African-American or White depending on my ancestry even if it goes back 2 centuries. It's a dumb concept that mostly makes sense to anyone. Yet it persists in these strange classifications that have no basis in any scientific system.
Data like this suggests that a common argument employed by diversity activists is false. Specifically, if companies like Apple and Facebook manage to create products that are wildly popular with women, it suggests that it isn't necessary to have female employees to make products for women.
(Repeat for race/nationality - I don't think a single African person works for Whatsapp. I've also never met an African woman in the past year who doesn't use it daily.)
Is anyone actually arguing it's strictly necessary, though?
Diverse viewpoints can help meeting the needs of diverse populations, but it's obviously possible to make a product that appeals to other demographics than your own.
For a very pedantic/strict interpretation of "necessary." I don't think anyone would argue that it is an absolute necessity to employ women to successfully market a product to women.
Neither of those argue it's impossible to appeal to other demographics. They argue that diverse viewports are generally beneficial, and thus it's something worth caring about.
Since we stopped arguing about absolutes, what kind of numbers are we talking about?
How big is the effect size? Does it suffer from diminishing return? I.e. is it enough to have about n women to hear all ideas worth hearing from the womanly perspective? Or perhaps it is not an amount but a proportion?
If diverse viewpoints helped, then we'd expect a higher proportion of companies with diverse viewpoints to be successful. If that occurred, it would be (weak) evidence in favor of the hypothesis.
In contrast, the relative scarcity of such companies is (weak) evidence against the hypothesis. That's just how probability works.
I'll never understand why anyone would care about the absolute numbers. Look at the relative numbers (relative to applications) and try to find any kind of bias (including preferential treatment for minorities). No bias? Great. Some bias? Not great. Pretty simple really.
Another thing I don't get about absolute numbers is that it seems desirable to have an outcome proportional to society (eg. 0.2% "Native Hawaiian or other Pacific Islander"). Given the fact that we're all equal; I don't see why that would be any better/worse/whatever.
There is no intrinsic reason why these numbers should not be proportional to society in the long run, that's why we compare the ratios.
Not even considering issues of equality and access, diversity as a thing in and of itself is at least good business. Especially in positions of power and decision-making. And that is at least one reason to peer at the discrepancies with open-minded curiosity - not to start a witch hunt, but to make smart shifts towards better business.
"There is no intrinsic reason why these numbers should not be proportional to society in the long run, that's why we compare the ratios."
That's true. Roughly proportional anyway. But there's still gonna be minorities inside that group. I guess what I tried to say earlier is that making sure minoritie's voices are heard is (IMHO) more important than making sure they're represented proportionally to society. Simply because there's always gonna be a minority.
"[...] diversity as a thing in and of itself is at least good business. Especially in positions of power and decision-making."
Don't forget that huge groups like races, genders, etc. are incredibly diverse already. I guess a stereotypical example would be hiring a white man might diversify your staff more than hiring a black women if that man happens to be a vegan (or whatever). Or take the Fox News staff as an example if you like (they sure look diverse on mute). No argument on the second part though. Any representational entity (management, senate, etc.) should represent society accurately.
The rest of your post is a no-brainer. Talking to seemingly different people is always a good idea :)
That's his point exactly. The reason more black kids start playing basketball is because more black people are in the NBA. And therefore more black kids end up in the NBA in the future. It's a cycle. There's no _intrinsic_ reason for that though.
If seven out of ten applicants are male then the 70% male ratio doesn't indicate a bias. It might mean that the company isn't as attractive to female applicants, but then again, neither is the discipline behind many of their products (at the moment.)