Psychometric testing / Abstract reasoning as software developer measurement?

[Necuno]

First time that I've encountered these as part of an interview for a software development position. However my question is how relevant are these?

For example:

Your task here is to look at the series of five figures, work out the rule which links them all. Then choose which of the four following figures obey that rule and identify the one which does not.

 

[cguy]

C is different

I suppose they have a place, but I would prefer to work through a more relevant puzzle or coding question myself. The only issue with my approach is that it isn't really standardized. My take is that it's fine for screening for a company, but for a specific position, I would rather have a position targeted test/interview - usually set by the hiring team and manager.

 

[Syzygy78]

a) line from vertices b) quadrilateral from vertices c) triangle from vertices d) line from vertices e) quadrilateral from vertices

what does the outlined block and plus sign represent?

 

[Knyro]

a) line from vertices b) quadrilateral from vertices c) triangle from vertices d) line from vertices e) quadrilateral from vertices

what does the outlined block and plus sign represent?

Nothing, this is the kind of thing you'd see on an IQ test. Four of the blocks have something in common and one breaks the rule. Figure out which one does not belong.

Also went with c

 

[cguy]

SPOILER They all have one block, one plus sign and an even number of circles, except c, which has an odd number. I also considered shapes implied by the circles/vertices and also whether or not the plus sign and blocks are inside or outside the shapes, however, the layout is too ambiguous for this to be what they were looking for.

 

[Syzygy78]

SPOILER They all have one block, one plus sign and an even number of circles, except c, which has an odd number. I also considered shapes implied by the circles/vertices and also whether or not the plus sign and blocks are inside or outside the shapes, however, the layout is too ambiguous for this to be what they were looking for.

you are right on the money, well done ^^ that satisfies the rule they were looking for

 

[HavocXphere]

First time that I've encountered these as part of an interview for a software development position.

That seems surprising. IT seems well suited to psychometric testing. They're expensive though (usually outsourced) - couple thousand rand per person for the 6 hour version so they're not done for all positions/candidates.

However my question is how relevant are these?

These tests (stuff like Raven's Progressive Matrices etc) are intended to measure general intelligence.

They're also pretty eerie sometimes. e.g. The last one I did looked a bit like a solitaire game (computer based) & I also walked out there thinking wtf was that about. They let me see the result afterwards for that particular test - stuff like impulse control based on timings between actions etc. So just because you can't see the relevance doesn't mean its not there.

Source: My employer keeps a couple professional psychometricians on the payroll so we got poked & prodded by them quite a bit.

 

[rwenzori]

Psychometric testing is just rubbish. They just measure your ability to do psychometric tests. The only ones fun to do are those personality ones, that are often so easy to direct to what you would like to show.

 

[Necuno]

I would however like to know how they are going to tell me if I'm going to write for example a good or bad webApi using dependency injection or a SPA with something like breeze and angular. Perhaps it tells me that you might be better at DDD, recursive or just plain algorithms.

This strikes me more of something to make someone feel warm and fuzzy within and has little to nothing to do with actual development skill. Of course every one is a john carmack

 

[shauntir]

SPOILER They all have one block, one plus sign and an even number of circles, except c, which has an odd number. I also considered shapes implied by the circles/vertices and also whether or not the plus sign and blocks are inside or outside the shapes, however, the layout is too ambiguous for this to be what they were looking for.

Took me about 20 seconds to see that as well

OP, I don't think they are that relevant to be honest. I'd rather they give me a problem and ask me to define an abstract solution in which they can determine my thought process.

The only thing I can think of is they are testing your pattern recognition which could help when building code architecture...

 

[scudsucker]

Or, more commonly, testing your ability for anti-pattern recognition when trying to fix the pile of **** that has been handed to you by the previous developer....

 

[Necuno]

Or, more commonly, testing your ability for anti-pattern recognition when trying to fix the pile of **** that has been handed to you by the previous developer....

In a small way yes, but if the crux of the test is pattern recognition then it fails at implementation and experience. With anti patterns you have to understand and study patterns its not simple a matter of just picking C out of A,B,C and E.

 

[HideInLight]

I don't know, the picture has no context. Just because there's an odd number of circles in C, does it really mean anything?

How exactly would a,d be linked to b,e?
ex: There are 2 circles for each square in b,e, but 1 circle for a,d.
That's an odd number of circles for each square/plus in b,e and a equal number of circles foe each square/plus in a,d

If the number of symbols is looked at, you have two pairs of identical images, and you would need a 3rd one to actually see a real pattern. But I guess that's why it's called abstract, open to interpretation...

although you could look at it as a regular expression:
PS(CC)*

 

[Syzygy78]

After some consideration, I have decided I'm not much of a fan for these types of questions. I mean honestly, now that I have started searching for these patterns and looking at examples the more similar they become and therefore much easier to solve. Its just like when you are first introduced to alpha-beta pruning but after a few examples it becomes somewhat trivial in nature.

 

[smallearth]

A could also be seen as different, as the + sign appears in the right of the square, whereas in B,C,D,E it is always on the left.

 

[HideInLight]

A could also be seen as different, as the + sign appears in the right of the square, whereas in B,C,D,E it is always on the left.

That's actually a pretty good observation.
I think you might of actually got it right, this is an abstract test, not a logical one.
Or at least the most right in a certain category, these tests normally have different scores for different answers.

 

[smallearth]

That's actually a pretty good observation.
I think you might of actually got it right, this is an abstract test, not a logical one.
Or at least the most right in a certain category, these tests normally have different scores for different answers.

That's what I thought too. I sat and stared for about 2 min at the damn thing...

 

[Syzygy78]

you could also argue that A is different given the fact that it is the only block that has a black sphere to the left of a + sign

 

[SauRoNZA]

Every time I've gone for psychometric testing it's been through a small company and I never got the job or ended up declining it for some or other bull****.

So either I'm crazy or these small companies take it way too seriously.

 

[cguy]

A could also be seen as different, as the + sign appears in the right of the square, whereas in B,C,D,E it is always on the left.

you could also argue that A is different given the fact that it is the only block that has a black sphere to the left of a + sign

In machine learning, we would typically call these types of conclusions a result of "overfit". It could conceivably be a rule, but there are several other possible rules of similar strength (ie, all spatial rules), e.g., only in C can you draw a line separating the spheres from the other entities. Which means that it is unlikely for a particular spatial rule to be the one of interest. There is also two with 2 spheres and 2 with four spheres, so min or max is ambiguous. The actual rule that becomes apparent is actually still a little ambiguous: either an odd number is unusual, or specifically 3 is unusual - either way it is C though.