Hacker News Reader

Comment by bsder | Hacker News Reader

The problem is that generally people use these kinds of articles to extrapolate from group to individual--and that's generally not a good idea.

Individuals break "typical" all the time.

If I'm interviewing you for a job, I shouldn't apply "group stereotype" as I'm specifically looking for your individual traits that break the norm. I don't want the "average" from your group--I probably want someone far from it.

So the arrow from individual to aggregate group is generally strong. The reverse arrow from aggregate group to individual, however, has nowhere near the same strength.

View on HN

I think that as a policy matter (what interviewers should do or should even be allowed to do), a stronger claim can be made. Suppose there are some groups A1 and A2 (men vs women, different races, etc — there could be A3, etc). Then I would argue that, regardless of whether the groups have substantial differences affecting qualification for the job, group membership or perceived group membership should not be allowed as a substitute for individual qualification. So, as an extreme example, if a job requires heavy lifting and short or contained hair, then an applicant should not be preferred because they’re male.

Obviously there are nuances here. If the groups are actually job qualifications (hiring a singer based on vocal range, for example), then none of this applies. There are surely plenty of gray areas, too. Gender doesn’t fit cleanly into a pair of groups. Sometimes it might be very hard to measure actual aptitude, and proxies of various quality are needed. College admissions are an example of the latter — you can’t measure how well someone can do, so admissions departments use proxies of varying quality and social acceptability.

As for gender differences, if I’m hiring someone, I believe that it doesn’t matter whether one gender is likely to be more qualified for any reason — if I’m hiring, I have an obligation to consider the applicant as an individual regardless of gender.

> if I’m hiring, I have an obligation to consider the applicant as an individual regardless of gender.

Since it's illegal to hire based on the results of an IQ test in the US, we have to come up with other proxies for it. So, we use stereotypes that are associated intelligent individuals as filtering criteria. Gender provides one of those stereotypes - in both directions. Men have traditionally not been considered for jobs like nursing or caretaking. Women have traditionally not been considered for jobs which require physical strength or analytical skills.

But back to your original premise: when you're evaluating hundreds, or thousands, of applications for a single opening, you're going to fall back on stereotypes, because it's too much work to do otherwise. Gap in employment longer than a few months? Skills have expired. Lots of job hopping after 6-18 months? Unreliable. Only has worked for companies known for their physical labor? Not capable of working in an office. A short job history with no college? Not smart enough.

Even though every single one of those is potentially wrong, it's also potentially correct. If you need to weed 100 applications down to 10 for phone interviews in the next hour, can you really afford to give each of the 100 applicants the benefit of the doubt?

EDIT: Removed "intelligence", replaced with "the results of an IQ test", which is what I meant originally. Apologies.

> Since it's illegal to hire based on intelligence in the US

Source? This article says it can be legal. https://www.hiresuccess.com/resources/guide-to-employment-te...

> when you're evaluating hundreds, or thousands, of applications for a single opening, you're going to fall back on stereotypes.

You personally might. But that's not what everyone does, not what the most successful people in recruiting do, and definitely not what I do. Why not use far stronger signals? For example:

1. Do they have the exact skills you are looking for or only some of them? 2. Aside from listing the skills, do they give concrete examples of using those skills? 3. How much experience do they have in their role? 4. Do they have experience in our specific industry? 5. If written communication is part of the job, how well is their resume written?

You can do a quick scan of the above in 10 or 20 seconds.

Thousands of resumes? Scan the first hundred that arrived and skip the rest. For most companies, your job isn't to find the one best person out of thousands of applicants, which would be a huge cost. It's to find someone qualified for the job.

Why use stereotypes when you have much stronger signals already available and it's just as quick to scan for those better signals?

> Source

See sibling comment.

> Why not use far stronger signals

These all rely on assumptions that an exhibited trait - a well polished resume that is tailored to the job description - have the non-visible traits sought.

Stereotypes work the same way. They're an assumption of someone's internal traits based on an outward trait, and are typically accurate when applied to groups (much as your resume filtering likely is mostly accurate, with a few outliers).

Stereotypes are just typically considered to be negative.

The factors I listed have nothing to do with a polished resume. It's comparing information with a strong signal.

Stereotypes are actually an extremely poor quality signal when hiring. If you use them, you're doing more work than you need to find a qualified person. Test it out and find out for yourself like I did on the advice of someone who does it for a living.

>Since it's illegal to hire based on intelligence in the US, we have to come up with other proxies for it.

Where are you getting this from? Mental disability is a protected class but not the same as intelligence.

I should have worded it better - it's effectively illegal.

http://en.wikipedia.org/wiki/Griggs_v._Duke_Power_Co

TL;DR: It's hard to defend an IQ test against a discrimination lawsuit.

Unless there's an explicit business need to concretely identify someone's IQ, a business is at risk of being called out as using the IQ test to discriminate against some protected group, since that protected group is potentially disadvantaged while taking the IQ test by any one of dozens of (legitimate) reasons.

It's not effectively illegal. That's a mis-characterization of both current law and that lawsuit in particular.

The court ruled that "the company's employment requirements did not pertain to applicants' ability to perform the job". And it was more specific than that. If such tests disparately impact ethnic minority groups, businesses must demonstrate that such tests are "reasonably related" to the job for which the test is required.

That does not make IQ tests effectively illegal. Demonstrate that anyone with an IQ lower than X cannot perform the job in question, and you're good. The actual "problem" is that other tests are far better at determining if you will do well in a job role. So companies use the better tests. If there were a job for which IQ was highly related to being able to do the job, then companies would surely use IQ as the test for that job.

To give a concrete example, why would I give a programmer an IQ test when I can give them a programming test? I'll take it further: an extremely high IQ is a deficit for many developer roles. Most of the genius programmers I personally know write code no one else can maintain, and are often condescending - usually without even knowing it, which hurts team moral. I love those guys and they often inspire me, but many of them would make bad employees if the job required writing average business programs on a large team. A programming test would reveal that when you see their code is very clever, and also includes an internal compiler so they can write the solution in a high level language they invented... There is good reason my very smartest programming friends are often unemployed. When a genius programmer is called for though, boy do they shine.

So it's not a case of companies avoiding the use of a good test due the law and replacing that good test with less effective proxies, as you first suggested. The real effect of that ruling and the law is that companies use tests that can be proven to demonstrate whether or not you can perform the job. That's better for everyone.

It is almost certainly legal to hire based on assessments of intelligence. If I insisted in a job interview that 4+9=12, the moon is simply a sky-reflection of the lights from Manhattan, and that I can’t stand to be in the same room as tall people since they affect gravity in disturbing ways, you would draw conclusions about my intelligence. Would you then avoid hiring me?

IQ tests might be illegal under some circumstances, but I would point out that they are not a reliable measure of intelligence anyways, making this all a bit of a moot point.

Interesting read, as was https://en.wikipedia.org/wiki/Ricci_v._DeStefano

I'm glad I don't have to worry about hiring. Seems like employers are caught in a catch-22.

> generally people use these kinds of articles to extrapolate from group to individual

As the article quotes: "People might be more reasonable than you think"

Or in other words: Citation needed.

Please check out Lee Jussim's work on Stereotype Accuracy[1]. His research shows that not only are stereotypes highly accurate (according to him one of the strongest effects in social psychology), but people only use them when they have no other information available. As soon as they obtain information about the individual in front of them, that information takes precedence over the stereotype.

[1] http://www.spsp.org/news-center/blog/stereotype-accuracy-res...

The article seems to say that the difference here is strong enough that you can do that:

> To put this number in context, a D= 2.10 means a classification accuracy of 85%. In other words, their data suggests that the probability that a randomly picked individual will be correctly classified as male or female based on knowledge of their global personality profile is 85% (after correcting for the unreliability of the personality tests).

The article suggests that you can do the opposite of that (use traits of the individual to reasonably accurately infer that they're male or female).

The parent's concern is that, given you know someone is male or female, you will make further assumptions about them based on their group membership rather than their individual traits, something that is not supported by the article.

> I am a strong believer that individual differences are more important than sex differences.

But the only case where you would rely on group membership is if the individual traits are unknown. Obviously generalization will never be as accurate as individual knowledge. Why would anybody be worried that people would act otherwise?

Assuming things about individuals based on group membership is a key part of racism, sexism, etc. Why would anybody be worried? Because people do it all the time.

Stereotypes exist because there are statistical patterns in groups of people. Those stereotypes are useful cognitive tools while they remain accurate. It certainly isn't fair to people in those groups who don't fit the stereotype, but as long as the stereotype accurately describes the majority, there's nothing to be done about it.

Would you run up to a lion or a tiger in the wild and try to hug and kiss it? I'm guessing you have a stereotype of big cats as dangerous predators, and you'd steer well clear, even though there are countless friendly lions and tigers in captivity that have never hurt anyone.

Are you saying that with a high accuracy, you'd feel comfortable treating an individual according to their group?

I don't want to appear too argumentative; it's just not exactly clear what you mean when you write

> .. difference here is strong enough that you can do that

Do you not do that? When you are first approaching someone you don't know, do you not take their expressed gender into account in how you make the initial greeting? Would you offer a woman the exact same handshake you'd offer a guy?

How would you approach someone who doesn't fit the norms - for example a male bodybuilder in drag? Does your brain hiccup for a moment when considering how to greet them? Mine does.

I'm sure I do act a little differently, but I try very hard not to.

Especially if I work with them or have some influence over their future.

This is a poor excuse, and I do not buy that people “generally” do this. It’s plainly obvious what people mean when they discuss averages and trends, and this hypersensitivity and paranoia that they might also hold those beliefs as universally applicable to every member of a group, and thus must be silenced, is not something we should accept.

If you put it in historical context it's neither hypersensitive nor paranoid. People did and do hold beliefs as universally applicable to every member of a group. Women were not allowed to vote in the USA less than 100 years ago. Only in the 1960s did it became illegal to request a specific gender in job advertisements. And it was only in 1974 that it became illegal to discriminate using gender in credit matters, including being able to obtain and use credit cards.

Plenty of people still use average and trends as an excuse to claim a gender (or race) is not as capable at some skill or career. It's not even that hard to find. Probably pretty easy actually in the US states that require creationism to be taught in public school.

> and thus must be silenced

Where did the parent comment say discussing averages must be silenced?

Those examples are not indications that people hold beliefs as universally applicable, especially those that were 100 years ago.

Also, most auto insurance still discriminates against men, especially young unmarried men, and we know for certain that’s not because every actuary with authority at every insurance company is under the impression that the increased risk is applicable to every single member. Likewise for life insurance that asks about age and smoking.

In case it needs to be said, there is a major difference between believing traits are universally applicable (which nobody believes) and believing that because something is true on average, that this is valid justification to discriminate against the whole group.

If someone makes the claim that a given trait is true on average, and therefore discrimination against a group is justified, only then should the discussion focus on that. Otherwise, it’s logically inconsistent, hypocritical, and even a little ironic to suggest or even suspect that individual’s line of inquiry is in any way connected to them suggesting or even suspecting a diminished capability from you (or any other specific person) as an individual.

> Plenty of people still use average and trends as an excuse to claim a gender (or race) is not as capable at some skill or career. It's not even that hard to find.

I’ve never seen this claim being made personally, and I don’t believe it happens with any significant prevalence, at least in the US. I could be wrong of course. I have seen many cases of people claiming that someone said it, only for the truth to come out that they actually did not.

> Plenty of people still use average and trends as an excuse to claim a gender (or race) is not as capable at some skill or career. It's not even that hard to find. Probably pretty easy actually in the US states that require creationism to be taught in public school.

Hah, you can even find this in neighbouring comments in this very thread.

> The problem is that generally people use these kinds of articles to extrapolate from group to individual--and that's generally not a good idea.

Conversely, people assume that any studies or discussions that even tangentially touch upon this are meant to encourage extrapolation from group to individual (even if they explicitly state the opposite, this is dismissed as lip service).

Reply on news.ycombinator.com