Hacker News Reader

Comment by diN0bot | Hacker News Reader

lots of people here are saying they scored lower than what they expected, and that maybe other people cheated. that could be it, but it could also be that hacker news folks tend to be overconfident. this would match the stereotype of this group being mainly male nerd entreprenuers, which could score worse on things like this but perceive themselves to score much higher (a feeling not a fact backed by studies that i can remember). who knows; just voicing this thought since no one has mentioned it yet.

View on HN

I got just under the median on this test, but I scored around the 99th percentile in the SAT verbal, and I feel it's reasonable to say the two test approximately the same things. It seems unlikely that I've slipped so far in just two years :)

I never really considered Hacker News to be full of overconfident people. If anything, to me being part of HN is a humbling experience. It reminds me that there are so many people out there that are smarter or more experienced than me, and I've seen other people say similar things.

I'm pretty much in the same boat. This scored me below median, but I scored in the 99th percentile on the SAT verbal as well (although that was a decade ago for me).

I think one contributing factor is the structure of the questions. If you ask me if I KNOW at least one definition for the word mawkish for example, I'll choose no I don't know a definition of it. I do however have a good enough feel for the word, that I can almost guarantee I can get a SAT style analogy with it correct, or if you gave me a multiple choice selection of definitions I can probably pick out the right one. I don't consider either of those skills equivalent to actually knowing a definition of a word.

Yeah, on the more difficult words, it becomes a test of your ability to tell yourself you know what they mean.

Same with me (SAT)... It says I have below the median vocabulary. They should have at least had some sort of base standard to compare the voluntary data with so that they could measure their sample drift.

Hacker News is _very_ full of over-confident people, believe me.

During a discussion of 'whether open source contributions were being overly important to job seekers' a while back, a surprising number of HN commenters automatically put themselves in the role of employer, peering dubiously over their glasses at me. Many of these commenters were hilariously under-qualified to be taking on that kind of role with respect to anyone.

I don't think SAT verbal is nearly as intensely focused on obscure vocab. I didn't do it, but did do a GRE verbal back in 1994 to get into grad school. For what it's worth, my score on this and my GRE verbal were quite consistent.

"Many of these commenters were hilariously under-qualified to be taking on that kind of role with respect to anyone."

You inferred this based on user profiles?

Correct. Most of the folks in question had filled in enough to be able to be straightforwardly findable in terms of linkedin, blogs, etc. It wasn't rocket science.

> Hacker News is _very_ full of over-confident people, believe me.

Heh.

i scored just over 90th percentile and i promise i didn't lie. it seems to me that a lot of the words were very old-fashioned - they were the kind of words i learnt from context while reading books by people like dickens. they're not words i would use in normal conversation, or when writing.

anyway, what i'm saying is that i suspect sat verbal is testing something quite different - isn't it much more aimed at ability to use the language rather than whether you can recognise obscure words?

[what surprised me the most was how graded it was - on both tests there was a pretty clear cut-off point where the words became unknown. i didn't expect it to be that ordered.]

I doubt that many people are cheating, but the "percentiles" are basically meaningless because the survey obviously self-selects for 1) people who are on the Internet, 2) people who are on Internet sites where this survey will get reposted 2) people who suspect they have pretty good vocabularies (who is going to take a non-mandatory 'test' that doesn't even earn them any Facebook Credits to inform them they aren't so bright), etc.

End result: substantial inflation of scores relative to what you'd see if you gave the same survey to the general population (and assuming everyone is relatively honest).

That's possible, even likely. Another explanation would be that tests like this tend to be biased towards people with very evenly distributed interests, preferrably with a literary, classical bias or background. Someone who knows 10,000 words from one or two specialzed disciplines will always score worse than someone who diligently read through all the high school text books and topped it off with some gutenberg.org.

(a feeling not a fact backed by studies that i can remember)

Maybe you're thinking of the Dunning-Kruger effect? (http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect)

Reply on news.ycombinator.com