Are companies really legally disallowed from using IQ tests in the USA?
This is definitely not the case in India - most companies that come to an engineering college campus for mass hiring follow this process:
1. Initial screen on grades
2. Written test for for those past the first filter (might be technical, or an "IQ" test)
3. In-person interview
Of course, keeping in mind some of the people that go through this process successfully, the bar is laughably low for all of these. (And of course, mass hiring companies get lazy and re-use questions, so there's quite a thriving industry built around 'question banks' for steps 2 and 3 above.)
Are companies really legally disallowed from using IQ tests in the USA?
Griggs v. Duke Power Company, 1971. US Supreme Court found that a particular use of IQ tests in hiring practices caused a disproportionate impact on African American employees. "Disproportionate impact" can make a facially neutral policy illegal under various US civil rights laws.
This is not a blanket ban on IQ testing in employment, but corporations being risk-averse, most of them don't really do it much any more.
(This is a very happy outcome for universities, since it gives them a virtual monopoly on discriminating on the basis of intelligence. Since that is really useful to do, all a university has to do is maintain its reputation as being a mostly reliable discriminator, and the actual contents of what it teaches are virtually irrelevant.)
The weird thing is that there was a case in New Hampshire a few years ago where firefighters had a test, and the city threw out the results of the test in making hiring decisions because...they thought the test would be unfairly biased against minorities. The Supreme Court heard this case and ruled that the test should have been applied, which at least to me seems to be the opposite of the Griggs case.
That wasn't an IQ test. It was a direct test of firefighting knowledge. While the test was never released to the public, a couple of questions were.
They were mostly stuff like: "You are entering a burning factory which appears to be made primarily out of concrete. What sort of saw blade do you attach to your buzzsaws? a) Steel b) diamond coated steel c)..."
The court ruled that the test results should have been applied simply because the de-facto procedure of:
1) Promote people based on a test, BUT
2) If black people don't do well, ignore the test.
was racist in favor of blacks.
(Another question famously related to NYC geography, since the test was lazily copied from NYC.)
If I recall correctly, the fire department in question was afraid that if it promoted strictly based on the test, then at least one black firefighter who hadn’t been promoted would sue, and the department didn’t want to defend the test in court. So they tried to dodge the problem by bailing on the test, and then got sued from the other direction.
After Ricci was decided, a black firefighter sued the city over the test [1]. His theory was that weighting promotions via 60% objective written test/40% subjective oral test was racist since it had a disparate impact on blacks. He lost.
Well, yeah, it’s a legitimate fear, but they should have handled that fear by being ready to defend their testing system in court before the tests were administered, and having the courage of their convictions.
If you could show that passing the test was an important part of performing in the job (i.e. those that did not pass the test would be worse, or unable to perform the job) then it would pass the Griggs finding.
The main issue with Griggs was that the court found that people who failed the test were no worse (at their job) than those that passed it. And, so, decided (rightly) that the test was not there for performance/ability reasons, but for other reasons.
>Griggs v. Duke Power Company, 1971. US Supreme Court found that a particular use of IQ tests in hiring practices caused a disproportionate impact on African American employees.
Did the US Supreme Court really find that African Americans are less intelligent than people of other races?
At the risk of replying seriously to a flippant comment, it is a stretch to say that the IQ test is a rock solid method to identify intelligence.
It's already common knowledge that African Americans, as a group, score poorly on IQ tests as opposed to other groups.
I personally find it strange that people feel it is all that important. After all, what if it was provable that a racial group was less intelligent than other groups? How could one reasonably and humanely find that information actionable?
Well, there are certain mission critical kinds of areas -- like medicine or bridge engineering -- where this knowledge is actionable.
Certainly people here on Hacker News don't need to be convinced of the value of finding elite engineers, or the very real wealth they can generate from book learning.
The problem is the doublethink that allows the IQ elite -- graduates of highly selective institutions like Harvard and Stanford -- to pretend that there is no such thing as an IQ elite, while simultaneously selecting their peers (coworkers, marriage partners, fellow students, etc.) from that very same ostensibly nonexistent IQ elite.
The contrast is whiplash inducing. One minute IQ doesn't matter, the next minute everyone is talking about how all these "idiots" can't solve FizzBuzz... Very big blind spot here, which many Asian countries don't share in the same way.
I took an IQ test in fifth grade. I will never forget one of the questions on the “general knowledge” section: “Who discovered America?”
Aha, fifth-grade-me thought, a trick question! “The Indians,” I said confidently.
The woman administering the test rephrased the question: “Who is generally credited with discovering America?”
Even though I scored well enough to qualify for the “IQ elite”, that experience gave me a livelong skepticism about how seriously to take IQ tests. I don’t need to ask for someone’s IQ scores before deciding whether or not to work, hang out, or have sex with them.
The comparison with FizzBuzz is not apposite, because nobody is claiming that FizzBuzz is a measure of general intelligence or even a measure of general programming aptitude. If candidate A solved the FizzBuzz problem twice as fast as candidate B, no sane hiring manager would conclude that candidate A is going to be twice as good at programming.
I'd be extremely surprised if you aren't misremembering. IQ tests are never supposed to have sections like "general knowledge", "history", "celebrity gossip", etc. "Who discovered America" is a textbook example of a non g-loaded question.
If such a section was used on a test they described as an "IQ test", it was almost certainly there only for statistical validation purposes.
There have been a number of studies done which ask obviously knowledge based questions, obviously g-loaded questions, and unknown questions. They then compare the statistical distribution of answers to the unknown questions to the obviously bad questions ("who discovered America") and the obviously good questions ("you want to carry a fox and two goats across the river in a 2 seated boat, but if you leave the fox alone with the goat the fox eats the goat..." [1]). If you aren't misremembering, I suspect you were participating in such a study.
[1] This is unbiased because all information is given. You don't need to know what a fox or goat is, or have ever seen a boat. It could be rephrased "you want to transport a X and two Y's from location A to location B. But if you leave an X and a Y alone, you fail."
I don’t think I am misremembering and I don’t think I was part of a study. This would have been around 1980, so the test was probably the WISC-R, which was published in 1974; maybe they got better about eliminating loaded questions in the WISC-III (which wasn’t published until 1991).
When I Google “who discovered america? wisc-r” I get several hits mentioning this test question.
Not to mention that the answer which gives you more points of IQ, hence meaning you're "smarter", isn't even correct. You can start with the native americans, go through to the vikings, and I read a book recently that made a very convincing case that the Chinese sailed to South America in 1421-1423 before they shut themselves off from the world for 400 years.
>This is unbiased because all information is given.
In this case, though, it is biased because some people will have heard the puzzle before, and having heard it before is tied to culture rather than to intelligence.
But even if you’re coming up with new questions all the time, a child who has experience solving this kind of problem will be at an advantage over a child who has never tried it.
Not only that, but questions can include a lot of implicit assumptions. One example I saw from a test was something along the lines of, "You go to a store. The sign says it's closed for lunch. It's currently 12:55, and you have an appointment at 1:30. What do you do?"
The correct answer was "Wait 10 minutes, then leave if it's not open." But people with lower incomes were disproportionately likely to say you should just walk away and come back some other time. This is because people in low-income neighborhoods have been taught that loitering in an empty lot leads to bad things.
I don't know about IQ tests, but this is certainly not true for the SAT. SAT prep only raises scores by about 30 pts (on the old 1600 pt test), at least according to studies not funded by Kaplan.
In those mission critical areas you mention, how does the knowledge of average IQ of the race/group the candidate belongs to give you any more information than the actual IQ of the candidate?
You're conflating different things. "These idiots can't solve FizzBuzz" isn't a complaint about performance on IQ tests — it's a complaint about their skill as a programmer. The only way that proves anything is if we assume that IQ scores are strongly correlated with programming ability, which begs the question.
There is plenty of evidence that, at least in the US, that simply thinking "black people don't do well on this test" or "women aren't good at math" lowers the test scores of both black people and women. Since these are both "common knowledge," that is a very good reason to exclude those tests. This is called "stereotype threat".
The good news is that there's also emerging research showing how test scores for blacks & women can be raised by exposing them to certain facts/examples that reduce the effect of "stereotype threat".
How could one reasonably and humanely find that information actionable?
It completely debunks the theory of disparate impact, assuming you believe IQ drives general problem solving ability. This information would be useful in rejecting a huge number of baseless lawsuits.
It also suggests that racial gaps in education are here to stay, and nothing can be done about them. Therefore, we should accept them and move on.
No it doesn't. It suggests the fact that one component of racial gaps in education may be immutable, and that when we identify the extent of that immutable genetic gap, we may be able to avoid wasting effort trying to eliminate it directly.
I don't accept for a moment the notion that "blacks" are "less intelligent" than "whites", or even that we can with any useful precision define what these terms are, but the way you've summarized the points seems to illustrate well the reason people have such a hard time discussing this issue.
"After all, what if it was provable that a racial group was less intelligent than other groups? How could one reasonably and humanely find that information actionable?"
I didn't feel the need to carefully explain the exact policy implications of this fact since the hypothetical didn't give enough info to work from. I also don't feel the need to couch a conversation on this topic in terms of "I'm not racist" "scare" "quotes" and genuflections towards diversity since I think most HN readers can distinguish a positive claim from a normative one.
I don't think you're a racist but I disagree with your conclusion and further think you took it to an extreme that inhibits reasonable discussion. Apparently in the hypothetical.
Our politics clearly couldn't be more different, but I didn't set out to offend you. I'm sorry.
For what it's worth, you are absolutely right that my hypothetical claim is stronger than most reasonable claims I'd make in reality. I wasn't aiming for completeness, merely trying to illustrate the fact that there are ways one can find such information reasonably actionable.
Patricks summary was just the basics of the case. The background is that Duke Power originally segregated employees, only employing African-Americans in the labor sector. Then the civil rights laws came in and they introduced requirements, including an IQ test, that didn't violate the laws, but had the same effect.
''Did the US Supreme Court really find that African Americans are less intelligent than people of other races?''
Simple answer is no, they didn't.
The court basically found that because African-Americans had suffered a lower standard of education (their wording: ''Because they are Negroes, petitioners have long received inferior education in segregated schools, and this Court expressly recognized these differences in Gaston County v. United States, 395 U.S. 285 (1969)'') and that DP knew this the whole point of the new scheme was to continue segregation.
Whether that would be the same now, I don't know. On the other hand I could completely believe that companies are still worried by it, and so avoid IQ tests etc.
[I attempted to reply to this, got a weird error code from HN, tried again, saw two copies of my reply both marked as "dead", and am now very confused. I'm trying again in different words in case there's some dupe-checker thing at work.]
> The court basically found [...] that DP knew this the whole point of the new scheme was to continue segregation.
No, not quite. The WP article (which incidentally I couldn't get at with your URL, but http://en.wikipedia.org/wiki/Griggs_v_Duke_Power_Co worked) quotes the judgement as saying that they didn't question the company's motive, but that its policies were illegally discriminatory regardless of motive.
Yeh you are right on both points (HN mungs the period on the end of the URL... grr).
The WP article was a bit of a mess so I've been rewriting/sourcing it - and, yes, the court carefully states that it didn't believe it was deliberate discrimination. That part of my comment was incorrect. :)
what if it was provable that a racial group was less intelligent than other groups
This is provable. Suppose that all racial groups were equally intelligent by some metric. Then the average value of that metric for each racial group would be the same. Assuming a sufficient resolution for the value range and millions of people taking the test, the odds of any racial groups average value being equal to any others is astronomically small.
Clearly I'm engaged in hand-waving. But proving something is getting others to agree with you, and I think what I said is very agreeable, without me delving into more formality.
No, what the Supreme Court found in that case at that time is that employment test, in the newly desegregating south, had not been shown to have relevance to occupational requirements. The Court did leave open the possibility that a test that was of the nature of an IQ test could be used for hiring, but only if the company hiring showed rigorously a relationship to job requirements. If that is shown, the disparate impact in something that is a bona-fide occupational qualification wouldn't matter.
"Disparate impact" reasoning in equal protection cases has a long history, by the way. One of the earliest cases, Yick Wo v. Hopkins, 118 U.S. 356 (1886),
held that a local law that regulated laundries in San Francisco violated the federal Constitution because most of the laundry businesses banned by the law were owned by Chinese immigrants, while most laundries that fit the law's requirements were owned by people of "white" backgrounds. (Knowing the time and place of this case, it's rather plain that the law was designed to discriminate in that manner.)
Context matters. But some of the comments above are true, that today many companies use school credentials as proxies for intelligence, and give higher education institutions a lot of incentive to operate as diploma mills. But there ARE companies in the United States that use various tests (the Wonderlic test is one example) that are essentially IQ tests as part of the hiring process.
The current claim in the mainstream literature on hiring procedures is that the two most effective screening procedures for job applicants are
1) work-sample tests
or
2) IQ-like tests (these two are tied for first place),
with all other procedures such as personal interviews, resume reviews, diploma requirements, personal references, etc., etc. ranking lower in practical usefulness. IQ-like tests are also quite inexpensive (in the form such as the Wonderlic test) and thus fare well for organizations that use them. That is the current claim in the literature, across a broad array of occupations in a broad array of industries.
After edit: Here's a link to a reasonably good, current bibliography on IQ-testing issues, including some of the issues about legal requirements and trade-offs in efficacy of different ways to rate applicants for jobs.
I would suspect the court found that the particular form of IQ test used was biased against poorer people, or would be biased agaist people from certain cultures.
People used to believe that, yes, based on IQ tests which measured "innate" intelligence. We now know that IQ problem-solving abilities are influenced by one's childhood development.
I've heard from Indian students at my university that grade screening in ridiculous in India, is that true? When applying coop employment here in North America many of them are shocked that so many companies don't have a minimum gpa cutoff and even the companies that require a minimum GPA to consider you have a reasonably bar 3-3.3 at worst. I've heard stories about people needing to get straight As or some ridiculous benchmark like that in India. Sounds like a toxic environment for the student, no wonder there are so few startups coming out of there despite a booming IT industry. What college kid would spend their time on a side project when they are being held to such ridiculously high academic standard.
Yes this is applied by some companies in India, however the reasoning behind it needs some explanation. Personally in my college there were around 350 CSE and IT students. The top companies(in India) say your Amazon, MS and Google have a lengthy interview procedure and cannot possibly go through this will all the students. So they screen grades.Note that, this happens typically only in the second tier instituions and not the premier institutions which do not have the problem of a huge student strength. For this reason typically students from other majors(electrical,chemical etc.) are also not allowed to sit for programming interviews.
Sadly many students will ask - "give my code away? For free!?"
This will also probably lead to "Heck, the code is free, let me copy it."
Also on the point of getting interviewed - most companies just want someone who can learn, and has good enough grades and a workable grasp of English.
Once you pass the tests, you are sent to their training centers where they will teach you whatever skills you required. The Tata training campus and is course is considered to be the equivalent of a 2 year Comp Sci. education in India.
Having a good grasp of code, is not necessarily an asset to get you into those jobs.If you want to strike out on your own and jump ahead of the curve, then I can see the merits of the suggestion.
Effectively for an undergrad 2 things matter - Your grades and your pedigree (IIT, Engineering students are the most valuable, and this combines with the reputation of institution. Commerce students are next in line, and arts students last.)
Basically these are proxies for ability discovery. Its pretty brutal.
I had to do an IQ test when applying to a hedge fund in LA: Madison Tyler Holdings / http://www.ewtcareers.com/ . I'm not sure if they'd talked to legal counsel about it, and I've never seen it anywhere else.
This is definitely not the case in India - most companies that come to an engineering college campus for mass hiring follow this process:
1. Initial screen on grades 2. Written test for for those past the first filter (might be technical, or an "IQ" test) 3. In-person interview
Of course, keeping in mind some of the people that go through this process successfully, the bar is laughably low for all of these. (And of course, mass hiring companies get lazy and re-use questions, so there's quite a thriving industry built around 'question banks' for steps 2 and 3 above.)