h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

Why are they no good for these measurements?

If the tests measure anything objective then there should be a way to apply these tests to animals or machines.

If they do not measure anything objective then they are being used under completely false assumptions.

The tests have been standardized. Which is to say that they have been biased to provide results that correspond to the expectations of the creators of the test.

Barton

dealing with your points in no particular order, the tests have not been standardised, they hae been normalised (look up the term normalised in a book on statistics if you don't know it).

The tests are no good for machines and animals because "they have not read the clasics", ie there are parts of the tests that rely on the fact that you have a human type of intelligence, can take the same type of instruction as a human and understand it, and generally have been raised as a human in a humaniform body.

The tests do measure something, as the spotting of what they tested lead to the discovery of "g" the general intelligence, and to the specific intelligences as well. This in turn lead to trying to measure "g" directly, which is why the "evoked potential" tests were developed in the first place (as far as I can find out so far).

PS I have since found out that there is a fundamental problem central to the entire idea of intelligence tests. If you use closed questions (binary or multiple choice) then you eliminate the skew that you get with open questions, but you also bias them in favour of beaurocrats, as it has been found that creative people do badly on these closed question tests compared to other tests. now how the hell am I supposed to add that to the review?

You are correct to correct me for using standardized where I should have used normalized. I do understand the distinction and I was being sloppy. However, correcting my terminology does not answer the questions I was asking. After all, you knew what I was trying to say, you corrected it.

The tests did not test as the testors thought the test should test. So, they played with the whole thing till they did, deliberately adding bias into the test interpretations till the testors felt that the test was fair. Then the tested and, suprise, they got the results they had biased the test to give them.

I think the information about closed and open ended testing is every bit as significant an indication that the tests have in built bias AND that there are more ways to measure these qualities that we collectively call intelligence than any one test is likely to be able to handle -- assuming it can be measured and catalogued at all.

Think about the thousands of different situations in which we might consider intelligence to be required and consider the many more different kinds of successful 'solutions' to those 'problems' that depend on the expectations of those making up the tests and answers. Often times, it is plain that a questions was restructured because someone got creative and found a solution that was not the answer expected. What does that say about the test?

How are you going to include that info? There must be, at least, 50 ways (to love your liver) but one of them should not be leaving that info out.

Barton

I don't want to get into the whole subject of psychometric testing, as whole books have been writen about it, and it is at least as complex as intelligence, and fairly controversial as well, but here goes.

As far as I can find, psychometric testing started with galton in 1883 trying to do the first cognative tests. These were general, but he introduce methods that are still central to the science today. in 1904 the first intelligence tests were designed for a specific purpose, (as were all later psychometric tests). As the tests were modified to be a better fit to the original purpose, it was found that other variables were showing up, as well as general intelligence, so tests were developed to measure them as well, so we get neuroticism/stability and introvert/extrovert scales. there are also tests that will detect lying to make yourself look good, and a number of other seperate factors.

As each factor is detected, a seperate test tends to be developed, with some of the questions correlelating with more than one factor. It seems to be impossible to design a single test that can measure all of the variables without producing something that is so long that no-one will actually be able to take it. so you end up with a large number of tests, each capable to some extent of giving you rankings for that particular variable, and hopefully being designed to be orthoganal, so that there is as small a number of tests needed as possible.

now that I have hopefully answered your question, where in this large number of reviews should the information from this set of posts go, in what order, and how should it connect to the other entries?

Perhaps you should start an article called "Objections to IQ Testing" with some attention to issues that are misunderstandings and areas that justify some caution in using such tests.

Barton

To understand the mechanics of how these tests work, you need to do one first.As any IQ tester is well aware there is no definate answer to some specific questions.If you feel an answer is correct in your view & you state a reason why, you will be marked accordingly.

sometimes true.

but only for tests that have been constructed with open questions.

a lot of test varients have non-obvious open-questions which the setters hadn't seen any other answers to, so they give specific answers in their answer-sheet.

This is fine unil it encounters a tester who marks to the answer sheet, who will mark you down for not getting the answer on the sheet, even if your answer is valid with working out shown.

Well Im sorry to tell you that youve been given wrong info mate.When I did one I found more than one answer & wrote both down explaining why.I was later told that both answers were correct & will be added to the answer sheet.

then you had good testers.

that is how they all should act, but quite a few don't.