Examples of cultural bias are given in the paper "Cultural Bias in Standardized Intelligence Testing" by Nicolas Bommarito:
Bias in an item on intelligence tests can result from cultural familiarity with one of the objects or words that occur in the question. For example, one item that appears in a form with pictures and in another form with words, gives a set of four instruments, harp, drum, violin, and piano. The test taker is then asked to cross out the one that doesn't belong. (Eells 258). Over half of the lower status children picked the harp, rather than the correct answer of drum, probably because they were unfamiliar with the harp as an instrument from a lack of exposure to it. A similar bias occurred on the WWI alpha IQ test, which contained obviously biased terms requiring knowledge of brand names and baseball players, and also on the Weschler Intelligence Scale for Children (WISC), which calls for knowledge of English weights and measures, literature, and history. (Mensh 1991:50). Questions like these seem far from measuring any type of innate intelligence, instead it appears as if they measure, at least to some degree, knowledge acquired though experience and thus environment. The may seem to be of minor importance in the overall results, but since speed is a requirement this unfamiliarity can be a severe handicap. For example, the WISC picture-completion subset allows only fifteen seconds for picture identification, an amount of time that requires significant familiarity.
Looking at the one test available from the site:
http://www.mensa-test.com/
Intelligence Test, Part 1: You have to work out what the letters mean. See No 0 as an example. It doesn't matter if you write the answers in uppercase or lowercase, but the answers must be exactly as expected (no additional intervals or dashes and the spelling must be correct).
0 24 H in a D = 24 hours in a day
1 26 L of the A
2 7 D of the W
3 7 W of the W
4 12 S of the Z
5 66 B of the B
and it goes on, another page of endless and pretty appalling culturally conditioned testing - try translating it into French, or even transplanting it from America to Scotland. Another blogger notes of this kind of rubbish that their should be a disclaimer saying: "Test results are only accurate for Americans, between the age of 25 - 34, and who had never voted Republican, and have consumed more than 3 beers in the past hour [you must be relaxed]"
And here is another one:
Tennis is to racquet as golf is to:
a. club
b. strike
c. bat
d. swing
This question does not just measure mental skills like insight or creativity; instead, it measures a specific piece of cultural knowledge that must be learned. To the extent that a child's environment does not include such knowledge of tennis or gold, they will not be able to answer the question.
Mensa's own site (http://www.mensa.org.uk/iq-levels/) says that:
IQ or Intelligence Quotient is an attempt to measure intelligence. This means many things to many people but generally the attribute of intelligence refers to quickness of mental comprehension (or mental agility). Intelligence is often confused with knowledge, wisdom, memory, or other attributes and in general has a variety of meanings depending on the context in which it is used. The term IQ usually refers to the attempt to measure a person's mental agility.
This is an incredible idea - that the "mental agility" - whatever that is - can be somehow divorced as a discrete measurable item apart from knowledge, wisdom, memory or other attributes. In theory then, it should be simple to devise an single intelligence test for human beings, cats, ants or spiders. After all, if it does not require the other attributes - which can be screened out of the testing process - then why not? Of course, the ability to read may be helpful - I believe that is called knowledge - which will cause trouble to some spiders.
Language, is of course, encoded in a particular culture, and the choice of words - try toilet, loo, lavatory etc - is also often highly correlated with class. Paulo Freire's Pedagogy of the Oppressed demonstrates how choice of language reflects social and cultural bias. Even pictures can reflect bias, for example, those of Robert Yerke's IQ test for immigrants made assumptions about knowledge of American society.
For the pictorial tests, go to:
http://www.understandingrace.com/history/science/race intel.html
Part Six of the beta IQ test (link above), developed by psychologist Robert M. Yerkes, was given to Army recruits to determine their innate intelligence. Yerkes intelligence exams (alpha, beta and individual) were culturally biased, taken under markedly different conditions and tended to reflect years in the U.S. and familiarity with dominant culture, rather than innate intelligence. Nevertheless, the early 20th century IQ exams have been the basis for standardized tests ever since.
There is a mathematical paper by the Princeton mathematician W. Smith, which looks in detail at the mathematics and logic behind both IQ and Artificial Intelligence. Regarding the founding work of Spearman in establishing IQ, he gives this verdict (which most mathematicians would agree with):
Spearman's 93-page 1904 paper (which Jensen calls "one of the 3 or 4 most important papers in the history of mental testing") in fact was quite shoddy both mathematically, statistically, and methodologically, and to add injury to insult, R.B.Fancher redid Spearman's calculations and found about 50 erroneous results apparently due to wrong arithmetic and with the signs of the errors showing an amazing fortunate tendency to "improve" the validity of his conclusions (Spearman's uncorrected numbers were then reprinted by Jensen [86] p.24). Specifically, Spearman seemed unaware of much of linear algebra, for example never mentioning eigenvalues and eigenvectors in his paper.
It is quite an amazing thing to me as a mathematician, but the psychologists and educationalists who devise IQ tests seem to care remarkably little about whether the "correct" answers on their tests actually are correct. For example, the criteria for the "right answer" to Raven's matrices are pseudo-logical or aesthetic, but there is never any proof any answer is right or wrong. So this test - despite the plentitude of praise heaped upon it by Jensen at every opportunity - is purely a "popularity test" or "conformity test" testing whether your aesthetic preferences (or preferences when employing pseudo-logical argumentation), happen to agree with more or less of society (specifically, the normalization subsample of society) or with the test-creator, and is not a test of how good you are at finding the "right answer."
He concludes that:
Since "intelligence"is about "the ability to solve problems," one might have proposed the naive idea for an IQ test that a problem-task be described (in some language) and then the testee tries to solve it. Bad idea: it is important to be able to solve problems that do not have (or do not have obvious) descriptions and definitions at all - in real life, often a large part of solving the problem first is to find a good problem statement. This whole paper is, in fact, an excellent example of that. In real life, one determines "what the problem is" oneself and then determines "how good the solution is" oneself also, usually with the aid of some disagreeable interaction with the external world.
The advantage of that is to locate "intelligence" within an evolutionary context as something which helps survival, by providing a kind of "mental agility" to solve hitherto unseen problems. This problem solving ability would very likely involve reasoning, but also the ability to deal with real world problems rather than total abstractions, and considerable knowledge may also help. That is why this is the approach being taken by different researchers in Artificial Intelligence. We can program a computer to do a set task, but how do we program it to learn intelligently and solve problems by itself. By divorcing IQ from the evolutionary context, and making it something like a Platonic ideal, as Mensa does, the end result is a set of exercises in trivia.
References:
The Mismeasure of Man, Stephen Jay Gould,
Betrayers of the Truth, William Broad, Nicholas Wade (good on the fraud perpetrated by Cyril Burt)
http://www.nicbommarito.com/writing/nonfiction/anthroiq.pdf
http://www.pineforge.com/newman4study/resources/iqtests.htm
http://www.pipeline.com/%7Erougeforum/freirecriticaledu.htm
http://www.autcom.org/articles%5CIQFallacy2.html
http://www.math.temple.edu/~wds/homepage/iq.pdf
http://www.personality-and-aptitude-career-tests.com/iq-test-scales.html
http://www.mun.ca/biology/scarr/b4241c2.html
http://www.donkeytest.com/history.html
http://www.mun.ca/biology/scarr/2250 Living with our Genes.html
0 коммент.:
Отправить комментарий