The Indus-Arabic positional system simplified complex calculations greatly. However, doing accounting calculations in Roman numerals also required quite a lot skill.
Did this hard work mastering Roman numerals inhibit Romans from doing algebra or trigonometry ?
I'd be prepared to say the tests are good for at most two standard deviations either side of the mean. I only have experience of very smart people with no common sense finding the wrong right answers on what were originally badly posed intelligence questions. One that springs to mind as a concrete example is find the next value in numerical sequence.
1 2 4 8 16 .. A. 24 B. 31 C. 32 D. 35 E. 40Anyone with common sense will pick C. 32 as the obviously intended answer but someone with extremely high IQ will pick B. 31 and wait to be asked why they have chosen the wrong answer.
These days with the internet and Wiki it is much easier to double check questions to make sure that no false positive answers are left in by accident - but it wasn't always the case. Some testees will attempt to break the test if they can find spurious unintended "right" answers.
I think that might at least be partially true. Some people with powerful pattern matching abilities can short circuit the test completing an insane number of questions correctly in a fixed time or finishing a fixed number of question test in a ridiculously short time.
There are multiple forms of intelligence being tested. The classic test involves a mixture of the different sorts to get a crude average but you can have people whose scores on a particular part of the IQ test are essentially off scale due to innate pattern matching ability.
Although the conventional ones tend to focus on variations of pattern matching in numbers, visual patterns and words.
The most extreme example I know of was a guy with an IQ of 60 modulo 100 who was also dyslexic with a written vocabulary of only 200 words. He was still a top flight mathematician though due to visualisation.
It is slightly curious how many of the child prodigy super IQ folk end up as mathematicians or chess grand masters. There are always a few exceptions. I'd take these numbers with a pinch of salt:
It is fairly obvious that there is a long tail on the positive IQ axis with just a handful of truly exceptional individuals on it.
Most people do fall within the range of validity though.
-- Regards, Martin Brown
Those actually aren't the only options!
That means the populations that get 95% of the test right or less or wrong. For a 100-item test, that split lies between those who only get five or six right or wrong.
For three standard deviations it's 99.73% so they get them all wrong or all right much of time.
One that basis you can't say much about what the tails look like. Your nice Gaussian distribution is actually a lumpy histogram.
If IQ means anything all that real, which I rather doubt.
If you accept that concept of a broad underlying general intelligence is valid. Spearman was convinced of it, but nobody seems to have a clue what it might be.
If you accept the premise that evolution leaves most of our capacities somewhat screwed up, as it tests random variations for survival value, Spearman's "g' might be an inverse measure of how much pointless variation you are lugging around.
-- Bill Sloman, Sydney
On Oct 30, 2017, snipped-for-privacy@ieee.org wrote (in article):
No, because the show the tails very clearly. I use this approach to analyze rare glitches in timing systems. In the case of IQ distributions, the tails would very clearly show any deviation from gaussian, because gaussians transforms into parabolas in log histograms.
I have not found any of the raw million-sample databases, or I?d do my own log histogram.
Not exactly. I?d recommend that you read The Man Who Loved China:
Winchester, Simon (2008). The Man Who Loved China: The Fantastic Story of the Eccentric Scientist Who Unlocked the Mysteries of the Middle Kingdom. New York: HarperCollins.ISBN
Also see.
How many schools use IQ tests, versus SATs and GMATs and the like. Nor do all schools use the same approaches. So it?s hard to see how such a sweeping conclusion could be true.
Yep.
Sure people test for character - such tests are usually styled as personality tests. They certainly interview for personality, looking for a ?good fit? or conversely to exclude bad fits.
Joe Gwinn
Discardtribution is actually a lumpy histogram.
And if you pick a test with only one question the distribution will be even lumpier.
but someone with extremely high IQ will pick B. 31 and wait to be
And someone with a even higher IQ will pick 32 because they are smart enough to realise tPost Discardhat there are two possible answers and based on all the previous questions the expected answer is 32.
Either that or the person who can ace the test is really smart. Discard
.Well I do know one man who was a child prodegy, got his undergraduate degree at Harvard in three years, and ended up a professor at Harvard. He did spend a few years working on Wall Street and funded a math prize at Harvard.
Dan
Only if you assume the questions are all roughly equal difficulty. The way it is usually done is that some questions are all but impossible except for geniuses to get more discrimination on the high score tail.
On a 5 way multiple choice test (and many of them are) it is possible to score 20% by random choice (eg. ticking A every time) and it is quite possible to score 30-50% on some tests by ruling out the obviously wrong answers and picking between the rest at random.
You make a few of the questions much harder or with an easy answer and a clever answer to distinguish the high tail more clearly. Talking here of exams in general although it applies to IQ too.
It measures something - a rough proxy for pattern matching ability in a Western culture. It goes horribly wrong when applied to Eskimos where they will stop at the first question they cannot answer.
It doesn't guarantee success either but then nothing ever does.
I don't see that a broad underlying general intelligence is an absolute necessity here. I have seen examples of genius level along one axis and dullard along another. Mathematicians tend to perform unusually well at visual-spatial reasoning and number sequences but be poor at word play.
-- Regards, Martin Brown
When the tests are designed to be hard to parse, they're no longer math problems. Worse, those that can be parsed multiple ways (or the author is pig-ignorant of English punctuation).
te:
te:
up
earch
drsity
sn't
.If there's no information there, the tails can be as visible as you like, b ut they don't tell you anything.
ls
Timing systems work rather differently from test subjects.
You need to think about what the test psychologists are using - tests conta ining no more than about a hundred items (usually less) which can be answer ed either correctly or incorrectly. The median test subject gets half of th e items right (if the test is well-designed). The two standard deviations s ubjects get five out of every hundred right or wrong. The three standard de viations subjects - a score of 99.73% - get one right or one wrong in about one test in three with a 100-item test.
This doesn't tell much about what is happening in the tails of the distribu tion. My point is that if you want to say anything worth hearing about the structure of the tails of the distribution you need separate tests - much e asier for the dumb tail, and much harder for the bright tail - so that the scores differ enough to tell how the tails spread out.
do my own
And it wouldn't tell you anything useful.
-- Bill Sloman, Sydney
The test items do test different kinds of intellectual skills. My wife and I find different items easy and difficult.
If there was a serious graduation in difficulty for a particular skills, th e test outcomes would tend to come out closer to 50% than they actually do.
So what.
Not if you want the results to be a nice Gaussian.
If there isn't a broad underlying general intelligence, there's not a lot o f point in testing for it.
Reification is a seductive intellectual error.
My wife and I both did very well in school, but in different subjects. We h ave compared our performances on an IQ test - not for any good reason - and we do find different items hard for one and easy for the other.
The tests - examinations - that got us both into university tested our skil ls in different subjects. We both happen to score well on IQ tests - but do well on different items, which all get lumped together. This makes the IQ test less informative than the examination system (if cheaper).
-- Bill Sloman, Sydney
IQ tests are designed to test general intelligence (which probably doesn't exist).
Test items that can be parsed multiple ways are aren't designed right and should get weeded out at the validation stage. There's a lot of incompetence around - and krw does exhibit it frequently - but it's rarely interesting.
-- Bill Sloman, Sydney
On Oct 31, 2017, snipped-for-privacy@ieee.org wrote (in article):
That?s an assertion, not a proof. As others have said, the questions are not all the same, there?s a wide range. And different people will interpret them differently, injecting some variation.
Granted, but the point is that log histograms are good for examining distribution tails.
Real IQ tests have hundreds of questions, and take about 90 minutes to take.
And a large sample handled correctly will tell you what the tails look like; and it will no longer be necessary to speculate.
In any event, this debate is a century old and we are not going to settle it here, so it?s time to stop.
Joe Gwinn
:
te:
ve a
nded
-by
dn't
es
ore
fore > > > > then).
ut.
here.
e,
An assertion of the obvious, which you do seem to be slow to take on board.
a wide range.
I've said it too. Sadly, the questions aren't neatly graded in order of inc reasing difficultly, so that everybody who gets "a" can be relied on get "b ". They test different skills,and the same score can be made up from correc t answers to different questions.
Different people will find them more or less aligned with their skills, whi ch is another - more substantial - source of variation.
Even Spearman agreed that intelligence could be broken up into specific fac tors, but he was fixed on the idea that because they all correlated to some extent there had to be an underlying general factor.
If there is information there to examine. Regular IQ test don't give you mu ch information about the way the tails of the distribution perform - the ou tliers either get all the items right. or all of them wrong.
be
half
hree
g
tests
so
?d do my own
ke.
So what?
ke;
The only way to handle a large sample correctly is to separate out the high and low scoring tails and test them separately, with harder and easier tes ts.
If anybody has actually done this, I've not heard about it.
it
The debate about IQ stopped quite a while ago. There is no general intellig ence in any sense that makes it worth looking for. IQ tests do have their u ses - they don't tell you all that much, but they are quick and cheap.
"The Bell Curve" milked the subject for a "controversial" book that sold we ll to rich Republicans, as it was intended to. The fact that it was a load of nonsense generated to enough academic outrage to get it into the newspap ers, which sold even more copies.
-- Bill Sloman, Sydney
The ones where you are expected to run out of time before finishing the test really do separate the sheep from the goats. They are also sometimes done with component axes available as independent scores. The right combination of speed and accuracy is key to a high score there.
The way I have seen it done is a long test paper with the questions getting progressively more difficult and a fixed time to do it. It probably isn't classed as an IQ test but the questions were similar.
I am not a great fan of IQ as a metric but it does measure something that is an engineering approximation to general intelligence despite its obvious limitations. You get a better idea if the various components of lateral thinking, mathematics, pattern matching, spatial reasoning and wordplay scores are shown separately (I have probably missed some).
The catch is that an IQ score does not mean success. I have known some very intelligent people crash and burn - including one national genius.
The thing we call general intelligence subjectively to me seems like a form of lazy pattern matching and with it high level abstraction.
IQ tests work fairly well to spot child prodigies. Though exceptional talent at some other skill from a very young age will also do that.
-- Regards, Martin Brown
If you have ever seen raw exam scores they are never really Gaussian - some additional adjustments have to be done to normalise the overall distribution to match previous years. Above my pay grade that step...
I have fought hard to ensure that the true right answer got top marks where the official marking scheme had a partly wrong answer though. It does happen (usually due to misprints in the exam paper) but sometimes really crass errors make it into public exams - this was a classic!
Even it is is a crude approximation it is still better than nothing.
Perhaps. I am always wary of principal component analysis as a way of fitting data but sometimes it does have its uses.
Many of the people I have known could have had other careers had things been just slightly different. Musician or mathematician is one example that springs immediately to mind some even do both professionally.
IQ is crude, but the component tests that make up the overall score do provide more detail if you choose to look at it that way. Your position appears to be that because the IQ test measure is imperfect (and it is) that we should not measure anything at all.
You could use age at which a baby starts to speak in sentences (or show an unusual aptitude for some skill) as a rough proxy for intelligence.
-- Regards, Martin Brown
Not necessarily, the real experiment would have been an increase from the historical 275 PPM to the now 375 PPM. or even the indoor 600PPM, then testing cognition. Not 1000PPM or a very high 2500 PPM. Mikek
On the contrary, I'm all for using better measures - which we do have, and have had for a long time.
The sole virtue of the IQ test is that it is quick and cheap to mark.
I don't object to it's use when a quick and dirty test is good enough, but taking it any more seriously than that is simply nuts.
Except that you haven't defined "intelligence". We all know roughly what it means, but the more carefully you try to to define it, the more elusive and multi-faceted the concept turns out to be.
The equation of IQ with "mental age" is an old and long-discarded approach.
-- Bill Sloman, Sydney
Facts don't matter to true believers.
The current value is 403.38 pmm at Maunu Loa.
Why not? The test was designed to look for an effect, and a range of CO2 levels makes the picture clearer (if it was the CO2 in the added gas that made the difference, which seems unlikely).
-- Bill Sloman, Sydney
As krw persistently reminds us. All the fact that matter to him are recorded inside his head. Pity he got so many of them wrong.
-- Bill Sloman, Sydney
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.