The long and the short of it is that IQ tests are of questionable value irrespective of the particular test or the person administering them - all they really establish is your aptitude at passing IQ tests. 'Intelligence' is not really a quality that can be accurately measured with a few checkboxes.

There is a reason why there is no Guinness record for 'highest IQ' any more - and that's because IQ tests are about as as reliable and relevent as phrenology.
( Happy Phantom has been to Hastings, Brighton, and Eastbourne too, Mon 22 Nov 2010, 18:12, 1 reply)

I disagree with this thoroughly.
The main reason I disagree is that the results of a wide variety of intelligence tests correlate well, pointing to an underlying factor. In the abscence of general intelligence it is hard to see why someone who does well on raven's progressive matrices would also score highly on the stanford binet, or indeed the battery of test within the WAIS. Also, the results of intelligence test correlate well with things like academic performance and job performance, as well as (broadly speaking) subjective assessments.
There is a massive body of work on intelligence in psychology, spanning over 100 years. If it were a simple matter that intelligence test "are about as as reliable and relevent as phrenology" then there would lots of research supporting this. But there isn't.
( Sextant goes to him is emotionally erect, on this day, Tue 23 Nov 2010, 4:11, closed)

and another thing while I'm ranting here
how would you establish that a test didn't measure intelligence? you would need a way to measure intelligence, ie you'd need an intelligence test. It's a bit of a chicken and egg situation.
The way psychometrists have gone about this is to have a think about the things which we believe are markers of intelligence (learn easily, good problem solvers, good at sums, clever at business, articulate, quick witted etc), test these things, and then refine retest etc. In the process we find try and weed out as much of the stuff which is influenced by other factors as possible using statistics. So, we want to control for things such as age, education, linguistic background, culture etc. By and large intelligence tests do a good job of this, although it is up to the person administering the test to make sure that they are using an appropriate test, and score it appropriately.
The observation was made probably when academic tests were invented that those who did well at one subject tended to do well on others, and those that did poorly tended to do poorly at everything. So the idea of some general quality called intelligence is something which has come to us through casual observation.
What psychometrics does is to develop better ways of measuring these qualities, and look at their correlations. The process of developing, refining, and validating intelligence tests has be going on for over 100 years.
Ultimately, what intelligence tests measure needs to be compared and validated against reality. This is constantly happening in research, and we know for example that your score on an intelligence test is good predictor at how you will perform in practical situations where intelligence matters.
The APA's response to the Bell Curve controversy is a good starting point.
www.gifted.uconn.edu/siegle/research/Correlation/Intelligence.pdf
( Sextant goes to him is emotionally erect, on this day, Tue 23 Nov 2010, 5:08, closed)

Yes, some tests correlate well, but that's because they're essentially testing the same thing. And yes, intelligence tests have existed for a good long while - many of the older ones have been conclusively proven to be a load of rubbish, as they concluded wonderful things like 'black people are thick', on account of having massive inbuilt cultural bias.

As for what intelligence is, that's a whole different issue altogether, and the criteria seem rather arbitrary. It's all a matter of how the brain functions, certainly - but why discount people who have an outstanding mechanical ability, a talent for writing music, perfectly-accurate recall, a gift for languages, above average empathy, or one of a million other evolutionarily-useful aptitudes? At what point did we decide that 'you have a good brain because you're pretty good at one fairly esoteric aspect of puzzle solving' was the start and end of the story, and by doing so, conclude that only the people who were proficient at the aptitudes being tested could be labelled as 'intelligent'? Is this not somewhat circular and self-fulfilling? It all seems rather contrived and artificial.
( Happy Phantom has been to Hastings, Brighton, and Eastbourne too, Tue 23 Nov 2010, 8:57, closed)

^this
It's not surprising that tests correlate with each other, because they use each other for calibration - an intelligence test is deemed to be valid precisely *if* it correlates with the others! Suppose I made a new test which placed more emphasis on, say, musical ability or joke-telling. Then many people would perform better on that test than on a standard IQ test, and others would perform worse. Who is to say which one is 'right'?
General intelligence is a myth, a reification of an arbitrary measurement.

I'd also like to call SGB out on the 'that's not a 'real' IQ test' argument. That's a bit like saying 'If you really want to measure someone's shoe size, you have to do hundreds of measurements to get the precise shape of their foot'. True, that gives you a much better measurement of someone's feet, but it's not shoe size any more. IQ is what IQ tests measure. And the fact that different IQ tests give different (if correlated) results just illustrates the point!
( flatfrog Staring vacantly into space, Tue 23 Nov 2010, 10:09, closed)

On the other hand
If you take a series of varied challenges that require mental capacity to solve, you can always grade people relative to each other.

It might not be a bona fide IQ test, but if person A scores 100, person B scores 120 and person C scores 160, you can to an extent say that person A is not as good at it as the other two.

If the tests are half decent quality, and you get a good few results, then I think as a simple intelligence test they are good enough.

Online tests are pretty unreliable though. Seems to me they add on 10 - 20 points becuase they are selling certificates. I suppose there's some irony in someone getting 160 in an IQ test, and being stupid enough to send these people £25 for a bit of paper though.
( Not Pineapple anymore, Wed 24 Nov 2010, 15:58, closed)