Recruitment processes often ask to take logical aptitude tests. Here's one such example of a test question:


The goal is to replace the question mark with the image that fits.

I went with Image D, since I couldn't figure out what else it ought to be. Yet I got it wrong.

Why do recruitment process use these tests? What exactly are these tests supposed to measure? And what the hell is the correct answer in the above?!?

  • 2
    That seems more like IQ than aptitude test, why do you think its not the case? – Tymoteusz Paul Jun 20 at 21:03
  • 1
    @TymoteuszPaul -- IQ tests supposedly are aptitude tests. The intent of an IQ test (aka an aptitude test) is to measure a person's inherent intelligence as opposed to something the person has learned (aka achievement tests). The problem is that just as multiple guess achievement test scores can be improved by learning how to game such achievement tests, IQ test scores can be improved significantly by learning how to game IQ tests. IQ tests do not measure aptitude. What IQ tests do measure is how well one does on IQ tests. – David Hammen Jun 21 at 5:35
  • @DavidHammen IQ tests cannot be "gamed". And those pattern recognition tests are only a part of an IQ test. What can be "trained" (instead of gamed") though is sitting in a room under testing conditions and understand what the tasks asks of you. The same way that having been in many interviews will help keep your cool in the next interview, also it's with different people, different questions for a different job. – nvoigt Jun 21 at 7:27
  • never seen a test like this, is it industry specific for this sort of foolery? – Kilisi Jun 21 at 9:18
  • 2
    @nvoigt - The intent of IQ tests is to test for an intrinsic quality called "intelligence", whatever that is. This intrinsic quality is not supposed to be something that is teachable. But every IQ test developed to date is teachable, and some of the teaching involves learning to eliminate some of the answers so that guessing can be rewarded. That's gaming. – David Hammen Jun 21 at 22:21

It is a pattern recognition assessment

Companies get a lot of applications, so their solution is to whittle down the applications with a barrage of tests with the theoretical goal of finding the smartest people. Companies much prefer to throw away good candidates than to hire poor candidates.

The correct answer is E. You got the wrong pattern for the black ones. It is one clockwise turn, two clockwise turns, three clockwise turns, etc.

| improve this answer | |
  • 5
    @JoeStrazzere - Whether IQ tests measure anything other than the ability to perform well on IQ tests is highly debated. But they are cheap, easy, and fast. – David Hammen Jun 21 at 5:32
  • 1
    But it's B. You got the wrong pattern for the white ones. – gnasher729 Jun 21 at 10:42
  • 8
    @gnasher729 You will have to explain your reasoning. The white ones turn counter-clockwise by one when assuming a sequence from left to right. Therefore the answer is E and Matthew is correct in their assessment. – Koenigsberg Jun 21 at 14:45
  • 1
    Pattern recognition is a very important skill for new hires who will be tasked with reverse engineering a load of questionably written legacy code put together by developers who have long since left the company :-) – DanK Jun 23 at 18:37

I agree with Matthew Gaiser's reasoning for how you are supposed to solve it.

When these questions are set naively there is a trick that sometimes works if you can't find all the patterns. The test setter wants to give you a set of answers that ensure you will get it wrong if you are wrong about the pattern for any one of the features. That means they are tempted to supply, for each feature considered separately, more correct than incorrect answers.

In this case, 4 out of 5 answers have the same line angle. 3 out of 5 have the same white tag position. 4 out of 5 have the same black tag position. E is in the majority in every case.

If I had worked out the patterns for the line and white tag, but not the black tags, I would have picked E over D because of the majority rule.

I don't think this would work on a serious, validated IQ test - the test setters would know that sort of trick and make sure it does not work, at least across a complete test.

| improve this answer | |

Why do recruitment process use these tests?

Companies are inundated with resumes from completely unqualified candidates, perhaps by a factor a hundred to one, or more. As an extreme, NASA receives several thousand submittals for every single astronaut candidate opening. Something needs to be done to filter the completely unqualified candidates from the perhaps qualified candidates.

The formal interview process is very expensive. Hiring someone is even more expensive than is the formal interview process, and firing / laying off someone is more expensive yet. This means that an initial filtering process that happens to reject someone who would have been a perfect fit is okay.

Erroneous hirings are a huge mistake. This makes organizations go out of their way to avoid such mistakes. It is far better to simply reject people who might be bad fits as early as possible in the hiring process. The cost of a false positive (hiring someone who shouldn't have been hired) is far greater than is the cost of a false negative (rejecting someone who shouldn't have been rejected).

As much as I do not like superficial or seemingly irrelevant automated tests, I do understand why they are used. At every step from the initial screening to the final hiring, it is far better to reject someone who is good than it is to accept someone who is not good. An automated initial screening is cheap and fast. That it might reject someone who would otherwise be perfect is irrelevant. The intent is to reject as many as possible who might not be good. False negatives are acceptable due to the huge number of candidates, the vast majority of whom should not have applied in the first place.

| improve this answer | |

I don't have enough reputation to post this as a comment; it's pretty long and not directly an answer:

These pattern recognition tests function well when used naively, but they break down completely once a participant knows something more advanced about math. If you care about this, for every finite sequence S of length N and for every new element E, there is a finite sequence S' of length N+1 that starts with the S and ends with E, i.e. S' = (S, E). You can basically append whatever you want at the end and still be mathematically sound.

Keeping in mind that these tests are nonsense once you see through them, I actually went with D as well when trying to play along. My reasoning is the following: (I'm going from right to left in this)

  • The black onces have positions (l, l, t, b). If you encode that in numbers, you could for example get (0, 0, 1, 3). The differences between them are (0, 1, 2).
  • The white onces have position (r, b, l, t). If you encode that in numbers, you could get (0, 1, 2, 3). The differences between them are (1, 1, 1).

My reasoning now was the following: It is obvious (...), that the next white position will be 1 away from the last one, which is 4 (or 0 if you're calculating in a clock fashion), so it should be on the right. The next difference for the black ones they wanted to get is 3, so it would be on the right (giving E as the answer). I, however, didn't feel that way (not trying to actively break the test); I was feeling like the sequence of positions should be left. Why? In my calculations, 0 ~ 4 (they are at the same position, as is 12 am and 12 pm on an analogue clock), so I was seeing the positions more like (0, 4, 1, 3) with (0+4 = 1+3). And I felt like repeating should be the way to go. That would give me the sequence (0, 0, 1, 3, 0, 0, 1, 3). Of course, this could also be (0, 0, 1, 3, 2, 2, 3, 1), but that's just the point.

As a conclusion:

People using math that don't understand math to try to get rid of people not understanding math is always amusing when knowing something about math ;)

| improve this answer | |
  • So what is your point? Do you suggest to tell the interviewer "I know math and all answers are correct, so you have to count those exercises as solved correctly for me"? Do you have experience with this approach? – guest Jun 22 at 6:01
  • No. My point is that when handed such a test, pick something that seems "right" to you. If there's an after-test-interview and they ask you why you picked the "wrong" one, explain your reasoning (but that approach is not intrinsic to my solution, that's the way I'd always handle such questions). If you feel like there's no such interview and you don't want to lower your chances, write your reasoning as notes on the sheet so that they can see them while correcting them. If there's no space for such comments (e.g. tests done electronically) and they remove you from the process, good for you. – SonneXo Jun 22 at 6:08
  • As you just dodged a company with (in my eyes) ridiculous tests (see David's and Joe's comments at the question) that do only care for a "perfect solution". Has it worked for me in the past? Yes, but not formulated as your suggestion. But I'm not inclined to extrapolate my (little) experience. "It worked for me, so it works for you" is a nonsensical inference rule. That's why I've written that this is not strictly an answer. That's no one-fits-all recipe andit doesn't answer all of OP's questions. – SonneXo Jun 22 at 6:10
  • In effect, you are using real number polynomial fitting. You fit a polynomial of order N to real numbers derived from the N points, and then use it to project the next point. Polynomial fit is not always the best way of analyzing something, especially when you are dealing with arithmetic modulo 4, not real number arithmetic. – Patricia Shanahan Jul 10 at 8:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .